Luminous Landscape Forum
Site & Board Matters => About This Site => Topic started by: Graeme Nattress on October 18, 2007, 06:50:49 pm
-
Optical low pass filters (OLPF's) or Anti Alias filter don't come in different "strengths". The filter works in two passes, each layer splits the light in two either horizontally or vertically, so by combing them together, you get vertical and horizontal filtering. The distance of seperation of the two rays of light is governed by the thickness of that layer, so if you want, (or need to as you don't have square photosites) you can adjust the filter accordingly. You choose the thickness of the filter in relation to the spacing of the photosites on the sensor.
Aliasing is what occurs when too high a frequency (too detailed information) enters a sampling system. A CMOS or CCD sensor has a regular array of photosites, and is a sampled system. If you put in too much detail, any detail beyond which the sensor can handle is "folded back" as an "alias" into recordable and visible frequencies, producing aliassing artifacts known as moire or "jaggies".
Once aliasing gets into a sampled system, it is very hard to remove. That is because the aliasses occur in the same levels of detail as real detail in the image. The more detailed the information that caused the alias, the lower the frequency it folds back to, corrupting more and more real image information. That means you cannot remove aliases without also removing real image data.
The thickness of the OLPF is usually chosen to split the light so that it matches the pitch of the green photosites on the bayer array. Green is by far and a way the largest component of luma, and it has the closest spacing on the bayer array, so by setting the thickness for that, you're using the least optical low pass filtering you can get away with. If you were to set the thickness for that of the red or blue, you'd be reducing the resolution of the final image too much, but you would avoid chroma aliasing. As it stands, the best compromise (and all engineering is a series of educted compromises) is to filter the green correctly, and hope that chroma moire doesn't intrude too much.
One thing that does not remove moire or jaggies, but actually makes things worse, is downsampling. Because downsampling is a filter followed by decimation process where all frequencies not allowed in the small image are removed in the large image, then pixels thrown away to create the small image, if you have aliasing you have frequencies that will not get removed by the downsampling filter as they're folded back into frequencies where detail exists that you wish to keep. Often poor image downsampling filters are used, and these actually create more aliasing on the small image as they're not strong enough to filter out the too high frequencies, and this only makes matters worse.
Now, most of what I've written also applies to three chip systems (like in video cameras) or single chip depth based systems like Foveon. With bayer pattern sensors there's an extra complexity caused by the bayer pattern itself and how it works, and that is when you get aliassing artifacts, you don't just get luma aliasing, but you get chroma aliasing too, and that appears as funky coloured edges to sharp objects. This is quite objectionable - even more so than the pure luma aliasing that you get with the other approaches. Bayer demosaic algorithms that reconstruct the RGB rely on analysis of the surrounding image content to determine the best possible guess of what the colour should be. If there are aliases in there, the algorithm cannot detect which is real detail and which is aliassed, thus causing results that are not as good as they could be.
Aliasing doesn't just add artifacts to an image that should not be there. There are other implications, that perhaps are not as critical for stills cameras, but they do effect my area of speciality, which is moving images. Compression systems work on frequencies, and aliases add extra high frequencies that don't correlate to image content, making the image harder to compress. Also, as I deal with very high definition moving images, anything you might want to do to a still, like "paint" out a problem, is not applicable or appropriate.
The major problem with OLPFs is not that they're in a camera - to me, they're a necessary part of the design to make sampling theory work without producing artifacts - but that because of their physical nature, they're not a very "steep" filter so it's hard to remove unwanted frequencies without effecting wanted frequencies. Of course, the lens' MTF acts as a low pass filter, so we don't have vast amounts of high frequency detail entering the system, but we do have enough to be an issue, especially on the size of photosites we're using to get low noise. A steeper OLPF would obviously solve this, but they don't exist. Similarly and OLPF that would filter the red and blue stronger than the green would help, but they don't exist either.
The strange thing is, in my mind, the more resolution in terms of pixel-count the more you should be using a OLPF as you're more likely to be scaling that image down. They best way around the issues of an OLPF, is to oversample, capturing more pixel resolution than you need, and using a proper downsampling filter to reduce the resolution, but increasing pixel-sharpness in the process. I personally don't think that aliasing helps algorithms to intelligently upscale images aliassed images, on edges, tend to have pixels that are less correlated with their surroundings, and hence there's no local information to properly infer edge direction, and you can get into the situation where there is no "good" direction through that pixel as it becomes like a saddle-point.
To remove the necessity of an OLPF, you either need to shrink the photosite size (probably leading to noisier images) or go to larger sensors at the same time (and increase your lens cost and size) or go to poorer MTF lenses. Or put up with aliasing artifacts. I think I'm sensitive to them - I don't like them one little bit, or what they do to algorithms for working with images.
Graeme
-
Excellent post, right on. I'd add one thing:
Optical low pass filters (OLPF's) or Anti Alias filter don't come in different "strengths". .... You choose the thickness of the filter in relation to the spacing of the photosites on the sensor.
I find no discomfort in the idea of referring to the "spread" (you call it thickness) of the OLPFs as a "strength", and it's pretty clear that that's what's meant. As you note, this can be set to different pitches, e.g., the pitch of the green pixels is a common pitch for it to be set at, you go further to talk about this being an engineering tradeoff between blurring and moire in the red/blue channels. I don't see any a priori reason to believe that different cameras don't tweak the "thickness", "spread" or "strength" of these filters based on fluctuations over and above tweaks necessary to compensate for changes in pixel size.
But... yeah, I twigged on the same line in the 1Ds3 article you did. In the end, it's a great article, and none of our tech geeking invalidates anything else MR said about the 1Ds3.
--Joe
-
Given it's a filter, if you look at it, the horizontal would look like (if you set the pixel pitch to that of the sensor photosite) [1/4 1/2 1/4], and similarly vertical. That would make the 3x3 that it generates:
[1/16 1/8 1/16]
[1/8 1/4 1/8]
[1/16 1/8 1/16]
or something like that. Now it's possible by changing the thickness of the layers to alter the "width" of that filter, but the "strength" of the filter is fixed, if you get my thinking. Yes, it's arguing semantics to an extent...
Now, if you don't make the pitch so that it aligns with the photosites, I'm not sure how it would work if make it wider. I can sortof see it working if you made it thinner though.
Indeed, great article, but I hate aliases more than I hate the resolution loss. The lack of sharpness can be quickly and easily fixed with some edge contrast or other sharpening tools, or downsampling and alliasing can not. To me, it's as simple as that.
Graeme
-
And I forgot to add that fill-factor and microlenses also contribute to the overall effect of the OLPF. They cannot provide an anti-aliasing effect, but they can minimise to an extent the appearance of what aliasses do arise.
Graeme
-
But, at least as I read Michaels report - MFDB's "typically" don't have an AA filter? I don't really understand how the absolute size of the sensor is relevant to the issue of moire.
... and I don't know how Graeme's first post relates to this quote from the article:
"Here's the Achilles heal of almost all DSLRs. They have an AA filter primarily to save battery power and processing time when computing a JPG. Medium format backs, for example, typically don't have AA filters because they don't need to produce in-camera JPGs, and therefore have no need to demosaic the Bayer pattern in-camera."
I do confess that this is the first time I've heard the reason for having an AA filter expressed this way.
I also wonder at "The reason that anti-aliasing is even mentioned in this review, though the whole AA filter issue applies to almost every DSLR on the market, is that the IIIs is the __first DSLR to offer resolution in what has, until now, been medium format territory.___" I would have thought the 1ds2 had, at least at one time (ie the Kodak DCS), medium format resolution.
The kind of landscape shooting I do just doesn't raise the circumstances eg: fabric or as per the generic example here: http://en.wikipedia.org/wiki/Anti-aliasing (http://en.wikipedia.org/wiki/Anti-aliasing) that I would think would cause problems.
As for down sampling, why wouldn't someting like up rez 200% gaussian blur of even 1 pixel, or .5 and downsample from there work? In any event if I'm downsizing it's most likely for the web and I really don't care as much as I would for a printed image.
-
Probably because MF backs are large and a large OLPF without defects is probably quite expensive indeed.
I don't see how not having to demosaic in camera to produce a JPEG has any effect on the engineering decision to use an OLPF or not. Because moire and aliasing cannot be removed without removing real image detail, no amount of funky demosaic algorithm will get you out of jail free unless you slap a blur on the image of significant size.
Because aliases mirror around the maximum recorded frequency, their effects can be at quite a low frequency if the input frequency is high enough. At some point the fold back is at a low enough resolution that you'd have to blur the image to the point of destruction to get rid of the aliases. They're insidious once they get inside the system, which is why, normally, people using sampled systems take great care to avoid letting them in the first place.
Graeme
-
Well, the article certainly provides an insight into why it is thought AA filters are needed, but doesn't really explain why Bayer type cameras without either AA filters or microlenses, those ridiculously expensive MFDBs that knock your socks off, get by without them. (I think I'm right in saying that MFDBs do not have microlenses, but if I'm wrong I'm sure you'll correct me ).
It's interesting to note also that Sigma's Foveon cameras are typically low resolution (the SD14 just 4.3mp) yet the clarity and 3-dimensionality is much admired by some. The cameras are also criticised for producing lots of aliasing and artificial detail, due to lack of an AA filter, which cause the images to appear sharper than they actually are, and no doubt some of that extra clarity is due to the fact that all the pixels are real, ie. non-interpolated.
Nevertheless, if aliasing is not such a big disadvantage with such a low-pixel-density sensor as the SD14, one wonders if it is really necessary with higher pixel density cameras such as the 1Ds3 and particularly the 400D and 40D.
-
Because moire and aliasing cannot be removed without removing real image detail, no amount of funky demosaic algorithm will get you out of jail free unless you slap a blur on the image of significant size.
[a href=\"index.php?act=findpost&pid=147085\"][{POST_SNAPBACK}][/a]
I can't quite see the logic of this. Because the Foveon type sensors can actually resolve beyond the Nyquist limit, the false detail is of a higher frequency than the highest real detail, in which case it should be possible to apply a low pass filter in post processing which is just sufficient to remove those highest frequencies which are false, but not the next highest frequencies which are real, if you get my point.
By way of clarification, although I understand that the real high frequency detail that is mixed up with the aliased artifacts, and which is of the same frequency as the false detail, cannot be separated by software, we are, after all, comparing images with and without AA filters.
The image without the AA filter would presumably not contain such real, high frequency detail. At best it would be attenuated.
As already explained, these low pass AA filters are not 'brick wall' filters that are able to cut off all frequencies immediately above the Nyquist limit whilst allowing all frequencies immediately below the Nyquist limit to pass. Frequencies just below the Nyquist limit will be significantly attenuated. If not, there would be no point and no interest in removing the AA filter.
Removal of the AA filter should allow those highest frequencies near the Nyquist limit, including real detail, to be recorded. Software which attempts to remove aliasing artifacts cannot distinguish between the high frequency artifacts and real detail of the same frequency which is however only present due to the lack of an AA filter.
Removal of such real detail during the process of removing the artifacts simply brings us back to square one. We have an image no more detailed than what we would have got if we hadn't removed the AA filter.
However, not all scenes have fine, evenly structured detail which is the subject of aliasing. Are you able to follow my convoluted argument?
-
ALthough the Sigma doesn't interpolate the luma, the chroma is pretty poor because silicon being a rather poor colour filter, so heavy noise reduction, which by definition in the case of a single image, must use surrounding pixels - ie interpolation - is used. I clearly see aliases on practically every in-focus Sigma shot I've seen. But when you've got such a low pixel count, you need all the help you can get, even if it does lead to nasty artifacts.
If you'd like that level of pixel sharpness, but without the artifacts or need for heavy NR, just properly downsample your bayer DLSR image. Works a treat.
I'm not sure about micro-lens status on MFDBs - Perhaps it's because of the large sensor that people have to stop down a lot shooting with them, and then the aperture diffraction blurs the image enough that an extra OLPF is not really necessary? Perhaps a filter-wheel then that plops in the right OLPF for the aperture you're shooting at then??
As for knocking socks off, yes the mirror slapback on the Hassy did indeed knock my socks off :-)
Graeme
Well, the article certainly provides an insight into why it is thought AA filters are needed, but doesn't really explain why Bayer type cameras without either AA filters or microlenses, those ridiculously expensive MFDBs that knock your socks off, get by without them. (I think I'm right in saying that MFDBs do not have microlenses, but if I'm wrong I'm sure you'll correct me ).
It's interesting to note also that Sigma's Foveon cameras are typically low resolution (the SD14 just 4.3mp) yet the clarity and 3-dimensionality is much admired by some. The cameras are also criticised for producing lots of aliasing and artificial detail, due to lack of an AA filter, which cause the images to appear sharper than they actually are, and no doubt some of that extra clarity is due to the fact that all the pixels are real, ie. non-interpolated.
Nevertheless, if aliasing is not such a big disadvantage with such a low-pixel-density sensor as the SD14, one wonders if it is really necessary with higher pixel density cameras such as the 1Ds3 and particularly the 400D and 40D.
[a href=\"index.php?act=findpost&pid=147086\"][{POST_SNAPBACK}][/a]
-
Hi!
I think I'm right in saying that MFDBs do not have microlenses, but if I'm wrong I'm sure you'll correct me smile.gif
At least P30 and P21 have micro lenses.
Cheers,
J
-
The way I read Michael's post is "don't expect a significant resolution difference compared to the 1ds2", and he is not even talking about wide lenses.
Considering that the d2x was IMHO in the same league as the 1ds2 already, it might that the actual gap in resolution between the 1ds3 and the d3 is much smaller than the pixel count might lead one to think.
TBC with testing of course.
Cheers,
Bernard
-
Considering that the d2x was IMHO in the same league as the 1ds2 already, it might that the actual gap in resolution between the 1ds3 and the d3 is much smaller than the pixel count might lead one to think.
[a href=\"index.php?act=findpost&pid=147132\"][{POST_SNAPBACK}][/a]
Most of the world still seems to be convinced that more megapixels are better though.
-
Most of the world still seems to be convinced that more megapixels are better though.
[a href=\"index.php?act=findpost&pid=147133\"][{POST_SNAPBACK}][/a]
Yep, but too bad for them.
I tend to be fact based. I'll probably rent a 1ds3 with a 16-35 f2.8 and will compare it to the D3 and Mamiya ZD to see to what extend "most of the world" is right to believe that more megapixels actually help.
Cheers,
Bernard
-
Yep, but too bad for them.
...
I would say too bad for those who believe that more pixels is not always a good thing! I'd trade a couple of MP for better dynamic range.
-
I would say too bad for those who believe that more pixels is not always a good thing!
[a href=\"index.php?act=findpost&pid=147138\"][{POST_SNAPBACK}][/a]
More pixels are always a good thing all other things being equal. I wouldn't be interested in having to use a 30+MP AA filtered sensor to get less detail that I am already getting with my 22MP ZD.
I'd trade a couple of MP for better dynamic range.
[a href=\"index.php?act=findpost&pid=147138\"][{POST_SNAPBACK}][/a]
Same thing here.
cheers,
Bernard
-
I tend to be fact based. I'll probably rent a 1ds3 with a 16-35 f2.8 and will compare it to the D3 and Mamiya ZD to see to what extend "most of the world" is right to believe that more megapixels actually help.
Cheers,
Bernard
[a href=\"index.php?act=findpost&pid=147135\"][{POST_SNAPBACK}][/a]
Bernard
Unless you want to handicap the 1Ds lll (and I'm sure you don't) I wouldn't think the 16-35 would be the best lens to try it with. My thinking is that only decent primes or the best zooms - e.g. 70-200/2.8 - would be up to the task. The 24-105/4 is pretty sharp too. Pity about the distortion and vignetting, but at least they're easily fixable in software. Hopefully Canon will release some decent ultrawide glass before too long.....
I may get the chance to compare 'apples to oranges' and shoot a 1Ds lll alongside my ZD shortly. I'll be interested to see how they compare.
Regards
Frank
-
Bernard
Unless you want to handicap the 1Ds lll (and I'm sure you don't) I wouldn't think the 16-35 would be the best lens to try it with. My thinking is that only decent primes or the best zooms - e.g. 70-200/2.8 - would be up to the task. The 24-105/4 is pretty sharp too. Pity about the distortion and vignetting, but at least they're easily fixable in software. Hopefully Canon will release some decent ultrawide glass before too long.....
[a href=\"index.php?act=findpost&pid=147146\"][{POST_SNAPBACK}][/a]
As a landscape guy, I use wides very often, the 16-35 f2.8 II is the lens I would be using and it is the reason why I would test the 1ds3 with this lens.
I am not really interested in doing a generic comparison, I would focus on my applications.
Cheers,
Bernard
-
As a landscape guy, I use wides very often, the 16-35 f2.8 II is the lens I would be using and it is the reason why I would test the 1ds3 with this lens.
I am not really interested in doing a generic comparison, I would focus on my applications.
Cheers,
Bernard
[a href=\"index.php?act=findpost&pid=147147\"][{POST_SNAPBACK}][/a]
Are you happy with the 16II?
-
I tested two copies of the 16-35 f2.8 II and gave them all back. Sorry but the lens is still crap, perhaps they should have dones something like 16-28 or so. I also use wides, but I only use Leica below 20mm than zeis and canon starts at 35 ;-)
-
Removal of such real detail during the process of removing the artifacts simply brings us back to square one. We have an image no more detailed than what we would have got if we hadn't removed the AA filter.
No, in fact it might bring us back to square -3 since aliasing will be apparent (and mixed with the signal) at frequencies lower than the cutoff frequency.
-
Well, the article certainly provides an insight into why it is thought AA filters are needed, but doesn't really explain why Bayer type cameras without either AA filters or microlenses, those ridiculously expensive MFDBs that knock your socks off, get by without them.
I think it is a matter of expectations and workflow, rather than a technical one.
With a DSLR, it is expected that you will shoot hundreds, or even thousands of frames in a day (if covering an event). At the end of the shoot, you want to have images that are immediately usable. A slight loss of resolution that still makes it possible for you to produce a great 20x30 print (according to MR's eyes) doesn't make the image as unusable as some strange artefacts.
With a MFDB, you are supposed to work much more carefully and slowly, producing less frames, examining them one at a time. Since you spent the same amount of money as for as for two cars just for the sake of superior resolution over a product that was already pretty good, you also expect to get the ultimate resolution, so anything that impacts it should be avoided.
-
I think it is a matter of expectations and workflow, rather than a technical one.
With a DSLR, it is expected that you will shoot hundreds, or even thousands of frames in a day (if covering an event). At the end of the shoot, you want to have images that are immediately usable. A slight loss of resolution that still makes it possible for you to produce a great 20x30 print (according to MR's eyes) doesn't make the image as unusable as some strange artefacts.
With a MFDB, you are supposed to work much more carefully and slowly, producing less frames, examining them one at a time. Since you spent the same amount of money as for as for two cars just for the sake of superior resolution over a product that was already pretty good, you also expect to get the ultimate resolution, so anything that impacts it should be avoided.
[a href=\"index.php?act=findpost&pid=147237\"][{POST_SNAPBACK}][/a]
I had similar thoughts myself. I think this is probably the correct explanation.
I get the impression that most people who use DSLRs shoot jpeg most of the time. They don't expect or want to have deal with moire issues and I believe that once the moire effects have been converted and compressed to jpeg, it's more difficult to remove them, is it not?
Perhaps the time has come for Canon to offer two versions of its DSLRs, starting with the 40D and 1Ds3; one with an AA filter for the happy snapshooter and one without an AA filter for the agonised, creative perfectionist.
-
Are you happy with the 16II?
[a href=\"index.php?act=findpost&pid=147176\"][{POST_SNAPBACK}][/a]
I have tried 4 copys. 3 were really bad. The 4th was much better then the old 16-35mm. I really don't understand the quality control of Canon. I bought the 50mm f1.2. A lot of CA. I went to the sevice of Canon. They could make it a little better but the lens still gives at f5.6 more CA then all other lenses including a chaep Sigma 50 lens. Hopefully I can get a better copy.
-
No, in fact it might bring us back to square -3 since aliasing will be apparent (and mixed with the signal) at frequencies lower than the cutoff frequency.
[a href=\"index.php?act=findpost&pid=147227\"][{POST_SNAPBACK}][/a]
Can you amplify on that please. Are you saying you can have aliasing artifacts, specifically moire, which is of a lower frequency than natural detail in the image which is unaffected by aliasing?
-
Are you happy with the 16II?
[a href=\"index.php?act=findpost&pid=147176\"][{POST_SNAPBACK}][/a]
I feel it is a significant improvement over the v1 that I had and as such am very pleased with it on my 1Ds2.
-
Can you amplify on that please. Are you saying you can have aliasing artifacts, specifically moire, which is of a lower frequency than natural detail in the image which is unaffected by aliasing?
[a href=\"index.php?act=findpost&pid=147344\"][{POST_SNAPBACK}][/a]
BTW, JPEGs don't make it harder to remove moire - it's just as hard with RAW - once it's "in" it's "in".
Yes, aliasing is caused by there being, like a "mirror" at the highest frequency that a sampling system can sample, and this mirror "folds back" any too high detail into lower levels of detail. Ever seen a cart wheel appear to run backwards on film - that's a great example of aliasing (temporal aliasing in that example. The more excessive the high frequencies you try to put in, the lower the frequency of the alias. That's how they can corrupt an image so badly you can't remove them.
Graeme
-
Sure, moire can corrupt an image, but it's rare. I've been playing with some RAW images from a 5D that had the AA removed by Marc McCalmont (he generously lent them to me to examine). I see a lot more false color fringing at high contrast borders than moire. The high contrast edge color fringing is eliminated pretty much 100% by the Lightroom, remove color fringing from "all edges" command. . . .
And, yes, the files are sharper. But I'm not sure it's a very significant difference, after I go through my normal sharpening routine. I'm trying to figure out if I think I can make a good enough 30" print from a 5D without AA filter to make it worthwhile. At this point, my tentative conclusion is that whether the 5D has an AA filter or not, the maximum print size that I feel is of exhibition quality does not change -- usually 24" in the long dimension, though for some images I can go to 28" in the long dimension (again with or without the AA filter). What's missing from the 5D prints when you enlarge them past 24", to my eyes, is not sharpness but detail. The detail is in the files with the 5D with AA; it's just not as sharp. But the files are sharp enough, when properly post-processed. For files from either with or without an AA filter, I find it's important to add digital grain before printing above 18". It adds a false impression of detail that's convincing.
-
Are you happy with the 16II?
[a href=\"index.php?act=findpost&pid=147176\"][{POST_SNAPBACK}][/a]
I am happy with the copy i purchased. It is sharper than my 17-40mm f4 and sharper than my 24-105mm f4 (where they overlap). It is probably as good as it gets for Canon at present. The main difference I have noticed with this lens, is the lack of CA compared to the 17-40mm.
Dave
down in NZ
-
the main reason for me not to buy the 16-35 II is the 82mm filter
-
the main reason for me not to buy the 16-35 II is the 82mm filter
That's a rather foolish criterion for buying/not buying a lens. You might look into considering other things like image quality, sharpness, maximum aperture, etc. when making a lens purchase.
-
the new 16-35 II is not better than the 17-40 , or not enough to change (for me of course)
and the cost of my 77mm filters is higher than the price of the lens, only the Singh Ray Vari ND filters cost 400$
if this lens had the same quality than the Canon EF 85 L II I'll buy it at once
a main reason is also that for a landscape photographer f/2.8 is stupid when you allways use f/11 to f/16
your reasons and mine are different
I am still waiting for a great Canon 24 mm
-
the new 16-35 II is not better than the 17-40 , or not enough to change (for me of course)
and the cost of my 77mm filters is higher than the price of the lens, only the Singh Ray Vari ND filters cost 400$
WTF??? The 16-35 is over $1200; the only filters you need are a circular polarizer (a good B+W brand one is less than $300) and UV/protection filter (<$100). Definitely not "more than the cost of the lens".
-
WTF??? The 16-35 is over $1200; the only filters you need are a circular polarizer (a good B+W brand one is less than $300) and UV/protection filter (<$100). Definitely not "more than the cost of the lens".
[a href=\"index.php?act=findpost&pid=147552\"][{POST_SNAPBACK}][/a]
Do serious photographers still bother with UV filters? I thought it had been established that modern lens coatings are so hard there's really no need for an additional piece of protective glass.
-
WTF??? The 16-35 is over $1200; the only filters you need are a circular polarizer (a good B+W brand one is less than $300) and UV/protection filter (<$100). Definitely not "more than the cost of the lens".
[a href=\"index.php?act=findpost&pid=147552\"][{POST_SNAPBACK}][/a]
as I said a vari-ND filter alone cost 400 $, a good slim POl filter >= 150 $, GND filters each one > 100 $
I am not using UV filter ... or time to time a special UV-IR filter for high mountain
-
Do serious photographers still bother with UV filters? I thought it had been established that modern lens coatings are so hard there's really no need for an additional piece of protective glass.
In most cases, no; a lens hood does a better job of protecting the front element from accidental impacts. But with wide angle lenses like the 17-40 and 16-35, the lens hood is too stubby to offer much protection, so a UV filter still makes sense in situations where encounters with rocks and such are possible.
But GND filters are unnecessary with digital; HDR blending works better in a lot more situations and is much cheaper. There's no reason to spend that much on filters any more.
-
Jonathan I am afraid you dont understand much about landscape photography
HDR is one thing , shooting in the field another one
-
Jonathan I am afraid you dont understand much about landscape photography
HDR is one thing , shooting in the field another one
Let's see, I've shot over 20,000 frames in Yellowstone, Yosemite, Grand Canyon, Bryce Canyon, Zion National Park, as well as a bunch of less well-known places all over the US. Except in certain areas of Nebraska, the Dakotas, and other plains areas, or out in the middle of a large body of water, how many times have you encountered a photogenic landscape scene where the division between bright and dark areas followed a straight line? In my experience, there's generally things like mountains, trees, buildings, and other things that start looking very unnatural when they cross the line of a grad filter, and fixing them is a huge hassle of burning and dodging in Photoshop.
If you expose properly, you can get a good image with a single capture in most instances, and if you can't, HDR blending works far better than ND grads with most subjects.
-
ok Jonathan :-) ... do as you like
-
Jonathan I am afraid you dont understand much about landscape photography
HDR is one thing , shooting in the field another one
[a href=\"index.php?act=findpost&pid=147579\"][{POST_SNAPBACK}][/a]
Jonathan isn't the only one not to understand much about landscape photography then... I agree 100% with his conclusions on this topic.
Cheers,
Bernard
-
do as you like
I keep using Singh Ray GND filters, and their very good ND filters
for exposure I am using a Sekonic LD 758 DR
-
I'm fine! It's the Universe that's f***ed up!
-
Do serious photographers still bother with UV filters? I thought it had been established that modern lens coatings are so hard there's really no need for an additional piece of protective glass.
Call me old fashioned - B+W UVs on all my Zeiss & Leica glass ... just in case
-
The way I read Michael's post is "don't expect a significant resolution difference compared to the 1ds2", and he is not even talking about wide lenses.
[a href=\"index.php?act=findpost&pid=147132\"][{POST_SNAPBACK}][/a]
It seems that this was just a bad dream, the updated version of Michael's review doesn't seem to contain these comments anymore. I guess that we'll know more after his return from Africa.
Cheers,
Bernard
-
Yes, it will be interesting to find out what Michael found out. Aliasing and optical low pass filtering is a very interesting topic of discussion, and certainly rouses debate. But that's what I like about photography - it's a nice mix of the technical and artistic!
Graeme
-
But that's what I like about photography - it's a nice mix of the technical and artistic!
Right on, Brother!
-
I read the review within a day of posting and I never saw anything about a comparison in resolution to the 1DsmkII.
It seems that this was just a bad dream, the updated version of Michael's review doesn't seem to contain these comments anymore. I guess that we'll know more after his return from Africa.
Cheers,
Bernard
[a href=\"index.php?act=findpost&pid=147617\"][{POST_SNAPBACK}][/a]
-
I read the review within a day of posting and I never saw anything about a comparison in resolution to the 1DsmkII.
[a href=\"index.php?act=findpost&pid=147710\"][{POST_SNAPBACK}][/a]
You are correct, it wasn't there in the first review. My point was not about the comparison with the 1ds2, but on the effects of AA filters.
In the initial review, there was a long section on AA filters, their values and downsides, and clear comments that the 1ds3 appears to have a stronger AA filter than the 1ds2, which requested more sharpening to have pleasing results in print. There were also some comments on the result of the comparison with the P45+ focusing on the difference in per pixel sharpness.
Besides, the introduction was starting with Michael saying that he would once more write his impressions as is, and that some people would not like them. This introduction being followed by a detailed discussion on AA filters clearly gave the impression that Michael was not too impressed with the resolution of the 1ds3. Most of his positive comments have not been touched.
Unless I am mistaken, these sections appear to have been removed from the current version of the article.
The only comment left is that the prints are "sharp".
Cheers,
Bernard
-
At the end of the review this comment is found ...
In a version of this review which was online for a few hours on Oct 18-19, there was a discussion of antialiasing filters and how I felt that there would be advantages to the 1Ds MKIII not having one, for a variety of reasons. Due to a mix-up, an early version which was not intended for publication because of mistakes in my initial analysis, found its way online in error.
I regret any confusion that this may have caused.
-
Well the problem is that when all these comparisons are made, many feel that after RAW conversion no alterations should be made to the image when comparing different image parameters. The problem of course is that of course the CCD MF back images captured with no AA filter are going to look greatly different. In my experience the CCD MF images do not take as much sharpening (and they do not need it) as the canon images. So there is going to be a subjective component when applying sharpening. In the end, I am very interested in seeing a RAW comparison between the canon 1DsmkIII and the MF backs after the "optimized" workflow.
The 40D images are soft but take agressive USM -- but sometimes depending on the image, I feel like there are some observed artifacts to get the image to where I would like it to be.
At the end of the review this comment is found ...
[a href=\"index.php?act=findpost&pid=147720\"][{POST_SNAPBACK}][/a]
-
Well the problem is that when all these comparisons are made, many feel that after RAW conversion no alterations should be made to the image when comparing different image parameters. The problem of course is that of course the CCD MF back images captured with no AA filter are going to look greatly different. In my experience the CCD MF images do not take as much sharpening (and they do not need it) as the canon images. So there is going to be a subjective component when applying sharpening. In the end, I am very interested in seeing a RAW comparison between the canon 1DsmkIII and the MF backs after the "optimized" workflow.
The 40D images are soft but take agressive USM -- but sometimes depending on the image, I feel like there are some observed artifacts to get the image to where I would like it to be.
[a href=\"index.php?act=findpost&pid=147739\"][{POST_SNAPBACK}][/a]
Having worked with the same camera with and without the AA filter I prefer the images (in general) without the AA filter and much less software sharpening to the stock camera with the AA filter and twice as much (or more) software sharpening. The difference (subjective) is subtle but nevertheless there. The final prints both optimized need to be compared.
Marc
-
Having worked with the same camera with and without the AA filter I prefer the images (in general) without the AA filter and much less software sharpening to the stock camera with the AA filter and twice as much (or more) software sharpening. The difference (subjective) is subtle but nevertheless there. The final prints both optimized need to be compared.
[a href=\"index.php?act=findpost&pid=147752\"][{POST_SNAPBACK}][/a]
To me, the biggest problem with AA filters is that they are there even when you don't need them. They are needed when the optics are very sharp, and they serve no purpose (and cause no significant losses, either) when they are very soft, but in that range where the optics are just a little soft (which is true of most Canon wide-angle lenses), then they add an extra level of softness, perhaps unnecessarily.
What is needed is something like a vibrating sensor, which replaces the functionality of the AA filter with mechanical motion, sweeping the sensor in small circles at least fast enough to do a full circle at the fastest shutter speed or practical flash duration, and the firmware in the camera could have information about what diameter of circle (if any) is needed for each lens at each f-stop. A user preference could scale the diameter even further.
-
Indeed, the OLPF is there if you need it or not. However, I think it wouldn't be easy to make a vibrating sensor produce the nice gaussian distribution that the OLPF proveds and could provide for artifacts of it's own if it's distribution was more regular and less smooth.
Graeme
-
Whilst searching the internet for software solutions to aliasing artifacts, I came across this interesting patent by Kodak which addresses (I think) the problem of the lower frequency artifacts below the Nyquist limit.
Here's an extract from the full text.
...... the invention comprises a system for reducing sub-Nyquist aliasing artifacts in an image below a visually perceptible level, where the system includes: a source of image data having sub-Nyquist aliasing artifacts; a visual perception stage utilizing a visual perception algorithm to identify the location and characteristics of the sub-Nyquist aliasing artifacts, thereby generating artifact coordinates and parameters; and an artifact removal stage for processing the sub-Nyquist aliasing artifacts by reference to the artifact coordinates and parameters to reduce their visibility, thereby providing an artifact corrected image.
While the aliasing is not always totally removed, the advantage of the method according to the present invention is that aliasing is rendered less visible in situations where it might otherwise be visible, all the while using a minimal amount of processing. This provides a feed forward method of minimizing certain aliasing artifacts while not imposing a heavy computational load on the image processing system.
The process seems to involve scaling the source image to a reduced resolution image, then back up again to full resolution, then comparing the upscaled image to both the downscaled image and the original image using a lot of fancy algorithms.
..another extract from http://www.freepatentsonline.com/EP1471728.html (http://www.freepatentsonline.com/EP1471728.html)
(e) scaling the high resolution source image to reduce the pixel resolution of the high resolution source image to a display resolution, thereby generating a reduced resolution image; and
(f) predicting the visual perception difference between the high resolution source image and the reduced resolution image, thereby identifying the artifacts in the reduced resolution image.
(g) scaling the reduced resolution image back up to the resolution of the high resolution source image, thereby producing an upscaled image;
(h) differencing the high resolution source image and the upscaled image and providing difference components; and
(i) using the difference components and the high resolution source image, the scaled image and the upscaled image to determine parameters of the artifacts, said parameters including artifact type, artifact location, and other parameters such as artifact frequency and phase.
-
What is needed is something like a vibrating sensor, which replaces the functionality of the AA filter with mechanical motion, sweeping the sensor in small circles at least fast enough to do a full circle at the fastest shutter speed or practical flash duration, and the firmware in the camera could have information about what diameter of circle (if any) is needed for each lens at each f-stop. A user preference could scale the diameter even further.
[a href=\"index.php?act=findpost&pid=147832\"][{POST_SNAPBACK}][/a]
Move a lens element. Which is already being done for IS. Then you can move it much faster than you could move the relatively heavy sensor. (I foresee hacked IS FW to do this on cameras with the AA filter removed.)
In practice I think you would need more than one full circle to deal with Nyquist. Hmm... I'm not used to thinking of signal fade in cameras but it could be an issue with a flash setup. How constant is the output vs. time? Should I consider the optical input a Rayleigh faded signal?
Ack -- I'm mixing work with hobby. This is bad. Very bad. Need to go take pictures and not think about this anymore.
- Dan
-
Move a lens element. Which is already being done for IS. Then you can move it much faster than you could move the relatively heavy sensor. (I foresee hacked IS FW to do this on cameras with the AA filter removed.)
In practice I think you would need more than one full circle to deal with Nyquist. Hmm... I'm not used to thinking of signal fade in cameras but it could be an issue with a flash setup. How constant is the output vs. time? Should I consider the optical input a Rayleigh faded signal?
[a href=\"index.php?act=findpost&pid=147993\"][{POST_SNAPBACK}][/a]
Increase the speed, then (optical element or sensor).
Well, the best solution is to oversample the optics and omit the AA filter, but that requires reading out many more pixels and introduces storage/tranmission issues.
-
Move a lens element. Which is already being done for IS. Then you can move it much faster than you could move the relatively heavy sensor.
A full-frame sensor is smaller and lighter than most lens elements, and moving the sensor would work with all lenses.
-
A full-frame sensor is smaller and lighter than most lens elements, and moving the sensor would work with all lenses.
[a href=\"index.php?act=findpost&pid=148031\"][{POST_SNAPBACK}][/a]
We're already moving the sensor or lens for image stabilisation purposes, depending on design. Are we suggesting here a sacrificing of IS so we can dispense with AA filters?
-
We're already moving the sensor or lens for image stabilisation purposes, depending on design. Are we suggesting here a sacrificing of IS so we can dispense with AA filters?
[a href=\"index.php?act=findpost&pid=148036\"][{POST_SNAPBACK}][/a]
Why would you need to sacrifice sensor-based IS? You just superimpose the two needed motions. For lens-based IS, there is no conflict.
-
A full-frame sensor is smaller and lighter than most lens elements, and moving the sensor would work with all lenses.
[a href=\"index.php?act=findpost&pid=148031\"][{POST_SNAPBACK}][/a]
The IS lens element is small and relatively light. Without taking apart my lenses and 5D I can't say whether the IS element is lighter than the sensor and sensor packaging. Though on principal I will accept that the sensor is lighter, so it's really a question about the packaging around the sensor (how much mass do we need to wiggle).
Summing the IS and AA motions into a single lens is simple, at least in theory. Of doing it this way means that you only get the AA advantage with some lenses. Moving the sensor works with all lenses on that body. Of course as mentioned in another reply you do need faster reads from the sensor which causes other problems.
- Dan
-
Why would you need to sacrifice sensor-based IS? You just superimpose the two needed motions. For lens-based IS, there is no conflict.
[a href=\"index.php?act=findpost&pid=148225\"][{POST_SNAPBACK}][/a]
Okay! I'll defer to your superior knowledge on such matters. In my simple-minded way I just assumed if I reach out to catch a cricket ball and then attempt to catch something else at the same time, I might miss the cricket ball. Generally I believe I can only do one thing at precisiely one time. However, I can walk and chew gum at the same time because I don't use my legs for chewing.
-
Minor correction Ray:
The Sigma SD14 produces a 4.67 megapixel output file from 14 million sampling sites. The optical resolution for "all" colors and B&W is very close to 1800 lines per image height horizontally and vertically which puts it around the 9-10 megapixel equivalence in b&w and better than 12 megapixel CFA equivalency for reds.
Best regards,
Lin
Well, the article certainly provides an insight into why it is thought AA filters are needed, but doesn't really explain why Bayer type cameras without either AA filters or microlenses, those ridiculously expensive MFDBs that knock your socks off, get by without them. (I think I'm right in saying that MFDBs do not have microlenses, but if I'm wrong I'm sure you'll correct me ).
It's interesting to note also that Sigma's Foveon cameras are typically low resolution (the SD14 just 4.3mp) yet the clarity and 3-dimensionality is much admired by some. The cameras are also criticised for producing lots of aliasing and artificial detail, due to lack of an AA filter, which cause the images to appear sharper than they actually are, and no doubt some of that extra clarity is due to the fact that all the pixels are real, ie. non-interpolated.
Nevertheless, if aliasing is not such a big disadvantage with such a low-pixel-density sensor as the SD14, one wonders if it is really necessary with higher pixel density cameras such as the 1Ds3 and particularly the 400D and 40D.
[a href=\"index.php?act=findpost&pid=147086\"][{POST_SNAPBACK}][/a]
-
Minor correction Ray:
The Sigma SD14 produces a 4.67 megapixel output file from 14 million sampling sites. The optical resolution for "all" colors and B&W is very close to 1800 lines per image height horizontally and vertically which puts it around the 9-10 megapixel equivalence in b&w and better than 12 megapixel CFA equivalency for reds.
Best regards,
Lin
[a href=\"index.php?act=findpost&pid=148420\"][{POST_SNAPBACK}][/a]
Okay! 4.3 or 4.7, lets not nit pick. We're all waiting for 10mp.
-
LOL - what would you use for a lens? The 3.4 produces over 1500 lines, the 4.67 about 1760 so the 10 in that progression would produce around 3700 lines horiz and vertical - probably out resolve any current 35mm lenses??
Best regards,
Lin
Okay! 4.3 or 4.7, lets not nit pick. We're all waiting for 10mp.
[a href=\"index.php?act=findpost&pid=148434\"][{POST_SNAPBACK}][/a]
-
LOL - what would you use for a lens? The 3.4 produces over 1500 lines, the 4.67 about 1760 so the 10 in that progression would produce around 3700 lines horiz and vertical - probably out resolve any current 35mm lenses??
Best regards,
Lin
[a href=\"index.php?act=findpost&pid=148436\"][{POST_SNAPBACK}][/a]
I don't know if it would outresolve all curent 35mm lenses, but I think that the ability to outresolve all current lenses would be a good thing. Basically that would lead to oversampling, which whilst not efficient, may give the best results. I think the sigma sensor has a crop factor of 1.5 so 4.7 Mp* 1.5^2 is around 10.6Mp. If the noise is under control I would be very happy with that I suspect (though perhaps I'd really want 25MP foveon...)
Mike
-
Minor correction Ray:
The Sigma SD14 produces a 4.67 megapixel output file from 14 million sampling sites. The optical resolution for "all" colors and B&W is very close to 1800 lines per image height horizontally and vertically which puts it around the 9-10 megapixel equivalence in b&w and better than 12 megapixel CFA equivalency for reds.[a href=\"index.php?act=findpost&pid=148420\"][{POST_SNAPBACK}][/a]
No, the Sigma doesn't do that many lines. It can do that many "lucky lines" with luck of alignment. The same lines can go totally unseen, with no contrast whatsoever, just by shifting 1/2 pixel relative to the sensor. That is not "resolution". A slight mismatch in frequency results in a low-frequency modulation between no lines and high-contrast lines.
-
Hey John,
We've been down this road before. Indeed it does do 1760 lines vertical and horizontally measured as all others have been on b&w standardized optical resolution charts. "Luck" has nothing to do with it.
I have no idea where you come up with these silly arguments about Sigma resolution. I've been measuring it and using the SD9, SD10 and SD14 since they were in beta and the optical resolution is precisely as measured for all colors. Really you need to stick to dealing with things you actually have experience with rather than introducing FUD about a camera you have no hands-on experience with.
Best regards,
Lin
No, the Sigma doesn't do that many lines. It can do that many "lucky lines" with luck of alignment. The same lines can go totally unseen, with no contrast whatsoever, just by shifting 1/2 pixel relative to the sensor. That is not "resolution". A slight mismatch in frequency result in a low-frequency modulation between no lines and high-contrast lines.
[a href=\"index.php?act=findpost&pid=148461\"][{POST_SNAPBACK}][/a]
-
Hi Mike,
Actually the Sigma's all have a crop factor of 1.7x.
Best regards,
Lin
I don't know if it would outresolve all curent 35mm lenses, but I think that the ability to outresolve all current lenses would be a good thing. Basically that would lead to oversampling, which whilst not efficient, may give the best results. I think the sigma sensor has a crop factor of 1.5 so 4.7 Mp* 1.5^2 is around 10.6Mp. If the noise is under control I would be very happy with that I suspect (though perhaps I'd really want 25MP foveon...)
Mike
[a href=\"index.php?act=findpost&pid=148457\"][{POST_SNAPBACK}][/a]
-
Lin, they're not silly arguments, they're based upon sampling theory and the lack of optical low pass filtering on the Sigma cameras. So if you've got 2640 x 1760, you should be able to measure that, however, you'll also get nasty aliasing. You don't get one without the other. Some people like aliasing, others don't. I personally like to keep it at a minimum. That's why I use high resolution zone plates for testing resolution and aliasing so that I can easily see what's going on. However, I do note from the Foveon pdfs that they do say that the requirement for optical low pass filtering is not removed by their chip design.
Graeme
Hey John,
We've been down this road before. Indeed it does do 1760 lines vertical and horizontally measured as all others have been on b&w standardized optical resolution charts. "Luck" has nothing to do with it.
I have no idea where you come up with these silly arguments about Sigma resolution. I've been measuring it and using the SD9, SD10 and SD14 since they were in beta and the optical resolution is precisely as measured for all colors. Really you need to stick to dealing with things you actually have experience with rather than introducing FUD about a camera you have no hands-on experience with.
Best regards,
Lin
[a href=\"index.php?act=findpost&pid=148476\"][{POST_SNAPBACK}][/a]
-
Hey John,
We've been down this road before. Indeed it does do 1760 lines vertical and horizontally measured as all others have been on b&w standardized optical resolution charts. "Luck" has nothing to do with it.
Luck has everything to do with it. Lines that are converging will naturally have zones of high contrast, but also zones that are almost completely grey. The grey "crossover" zones in the converging lines are just as significant as the areas with contrast; you just want to see the good parts and ignore the trash, and count only the good parts as a quality of the imaging. The people who decided how a resolution test should be interpreted did not bother to stop to think about aliasing, so you relish this loophole to pretend that artifical resolution is something good.
I have no idea where you come up with these silly arguments about Sigma resolution. I've been measuring it and using the SD9, SD10 and SD14 since they were in beta and the optical resolution is precisely as measured for all colors. Really you need to stick to dealing with things you actually have experience with rather than introducing FUD about a camera you have no hands-on experience with.
[a href=\"index.php?act=findpost&pid=148476\"][{POST_SNAPBACK}][/a]
I don't need my own Sigma and test chart to see that it violates the basic rules of consistent imaging. Chance of pixel/subject alignment is crucial for an appearance of high resolution.
You should know very well that the camera can't resolve *ANY* lines at all at the nyquist if you just move the sensor just 1/2 pixel. You don't see this with the converging lines, because some part of them is always guaranteed to have some contrast. In the experiments that other people did with printed sheets of checkerboard patterns at the nyqist, the checkerboard only appeared in a fraction of the checkerboard areas. Large areas of grey appeared as well. This is not what was envisioned in the original concept of "resolution", I would think. Play with your "resolution" loophole all you want, but aliased imaging falls flat on its face when you qualify with "consistent resolution with accurately placed points and edges", which is what matters most.
More pixels is what cameras need; not aliasing, not anti-aliasing. In the interim, anti-aliasing is the reasonable solution.
-
Hi Graeme,
It's a silly argument to tell me that the resolution of my camera is not what I measure it to be. I'm not a beginner either with sampling theory or Physics (Masters Degree) or with digital cameras.
Whether one believes that a low pass filter is necessary is a subjective issue and real-world results tell me that I prefer the IQ from my SD14 to any of my other (many) CFA dSLR's and fixed lens digital cameras. I'm comparing it with my Canon D30, 10D, 1D, 1DS, 1D Mark II and 40D. I'm comparing it with my Nikon D2Xs and with my Kodak 16 back and my Kodak DCS-760 which I "prefer" with the AA filter removed.
The term "nasty aliasing" is another "subjective issue". Is there some stairstep aliasing especially on some man made objects? Yes. Can I deal with it in post, very easily on a selective basis? Yes. Is it an issue for "most" of my photography? Absolutely not. In wildlife and nature photos I get amazing enlargements from my SD14 and even from my SD10. I have images posted at PMA at A0 size and larger which have no visible aliasing and have received rave reviews. So the bottom line is it's very easy for me to compare the output with that of my numerous CFA counterparts and I like what I get. The optical resolution is what it measures. There are "fewer" artifacts in the capture than from the CFA counterparts.
Sampling theory is fine to debate, but real world results are what work for the photographer.
Best regards,
Lin
Lin, they're not silly arguments, they're based upon sampling theory and the lack of optical low pass filtering on the Sigma cameras. So if you've got 2640 x 1760, you should be able to measure that, however, you'll also get nasty aliasing. You don't get one without the other. Some people like aliasing, others don't. I personally like to keep it at a minimum. That's why I use high resolution zone plates for testing resolution and aliasing so that I can easily see what's going on. However, I do note from the Foveon pdfs that they do say that the requirement for optical low pass filtering is not removed by their chip design.
Graeme
[a href=\"index.php?act=findpost&pid=148485\"][{POST_SNAPBACK}][/a]
-
John,
The "bottom line" is "results". I know what results I get from my Sigma and from my CFA cameras and which I and my customer's prefer. You can argue sampling "theory" till the cows come home and that doesn't change real world "results."
Mike Chaney (author of Qimage) has about put his Canon 5D to rest since he began using the SD14 - he has good reasons and is as good at "pixel peeping" as anyone around. The real world performance in produced IQ of the SD14 is outstanding and that's an incontrovertible fact. The detail and sharpness in enlargements are superior to what I get from my CFA counterparts. After 40+ years of making a pretty good living doing this I really do know what makes a good image!
As I said, we've been down this road before. You see aliasing everywhere you look - I'm amazed that you ever really "enjoy" an image, LOL. Thousands of professionals in the photography business passing by the Sigma booth at Photokina and PMA concur that the images produced by the SD14 are outstanding. That's what's important to me. I know what works and what sells and what people like. I'm not selling images to sampling theory pixel peepers, I'm selling to patrons of fine art and their opinions are what count.
So keep your head in those figures and sampling theory and keep wishing for something better - maybe it will happen for you. For me, I'll do just fine with my "nasty aliasing" and SD14 thank you...
Lin
Luck has everything to do with it. Lines that are converging will naturally have zones of high contrast, but also zones that are almost completely grey. The grey "crossover" zones in the converging lines are just as significant as the areas with contrast; you just want to see the good parts and ignore the trash, and count only the good parts as a quality of the imaging. The people who decided how a resolution test should be interpreted did not bother to stop to think about aliasing, so you relish this loophole to pretend that artifical resolution is something good.
I don't need my own Sigma and test chart to see that it violates the basic rules of consistent imaging. Chance of pixel/subject alignment is crucial for an appearance of high resolution.
You should know very well that the camera can't resolve *ANY* lines at all at the nyquist if you just move the sensor just 1/2 pixel. You don't see this with the converging lines, because some part of them is always guaranteed to have some contrast. In the experiments that other people did with printed sheets of checkerboard patterns at the nyqist, the checkerboard only appeared in a fraction of the checkerboard areas. Large areas of grey appeared as well. This is not what was envisioned in the original concept of "resolution", I would think. Play with your "resolution" loophole all you want, but aliased imaging falls flat on its face when you qualify with "consistent resolution with accurately placed points and edges", which is what matters most.
More pixels is what cameras need; not aliasing, not anti-aliasing. In the interim, anti-aliasing is the reasonable solution.
[a href=\"index.php?act=findpost&pid=148487\"][{POST_SNAPBACK}][/a]
-
If you like the look of aliasing artifacts or not, that's subjective. That they exist, and can be measured is not subjective but objective. Quite frankly, if you had a detailed enough resolution chart, you'd still be measuring resolution way beyond the pixels on your sensor if you don't stop when aliasing starts to occur. It's a joke to suggest that heavily aliased data is real resolution when it's an undesirable artifact. If you could keep your resolution as-is, but not have any aliasing, would that not be beneficial to your images?
Graeme
-
If you could keep your resolution as-is, but not have any aliasing, would that not be beneficial to your images?
Graeme
[a href=\"index.php?act=findpost&pid=148502\"][{POST_SNAPBACK}][/a]
But that's the problem, you cannot keep resolution as-is without any aliasing. For some types of images, a bit of aliasing is preferable to the detail loss from an AA filter.
Maybe I don't want it completely gone, but defintitely very weak. For instance the D70 was known for having a fairly weak AA filter, and I never had any problems with moire or obvious aliasing. What's disappointing to some is that on the newer DSLR's there seems to be a trend towards stronger filters, not weaker.
-
Hi Graeme,
This is where sampling theory and photographic reality collide. When I look at a b&w resolution chart at Nyquist a CFA sensor produces mush. When I shoot a landscape with grass, pine needles, leaves, etc., and enlarge a CFA print I get to a point where "mush" is very apparent. When nine converging lines on a resolution chart can no longer be resolved the AA filter produces a blob of indistinguishable grey plus color moire. In real-world photography this point is evident throughout when I reach the visible limits of resolution. These artifacts are far more objectionable to me than what you describe.
When my SD10 or SD14 stop resolving nine converging lines, I see five lines. Is this "real"? No - it's false detail but I still see "lines" not mush. Whether I see five blades of grass in my enlargement where there should be nine, or five pine needles where there should be nine on distant trees is largely unimportant as long as I still see something which resembles grass or pine needles or fur on wildlife (whose counting?). In my enlargements this false detail looks much more realistic than having an indistinguishable blob at the point of resolution exhaustion.
Would I like more "resolution?" Sure. But no matter how much "resolution" I have, an antialiasing filter simply at some point in the equation produces moire and blur which I don't want in my enlargements.
In the real world this lets me make larger prints and tighter crops from my Foveon based Sigma than with even my 12 megapixel CFA sensor and "still" hold the interest of my audience. The reality is that with AA filtering not having visible stairstep aliasing eventually causes color moire and resolution extinction blur and having more resolution only pushes the point at which it appears further along the enlargement scale, it doesn't prevent it. All that it does is soften the edges and smooth out stairsteps. For some that may be a wonderful thing, but my walls are full of huge prints which show no visible aliasing and that's what I want an what my customers want.
Why do you think AA filters are rarely used on MF backs??? Why did Kodak not use one on their 14 series and why did they have a removable AA filter on earlier DCS pro-line cameras?? It's because for many jobs it produces a superior product - not "nasty aliasing". Yes indeed whether or not one "likes" the look of the image (not likes or dislikes aliasing) is the true issue. One doesn't look at the printed image from an inch away and never see a single pixel - that's what interpolation is all about. The issue of "aliasing" with the Sigma/Foveon sensor is a very minor one and the improvements in sharpness, enlargeability and realistic appearance of prints more than compensates the small amount of controllable stairstep aliasing. Unless you are pixel peeping with zoom it's rare to even see this type aliasing in prints.
In short, complaining about a Sigma SD14's "aliasing" is a solution in search of a problem.
Best regards,
Lin
If you like the look of aliasing artifacts or not, that's subjective. That they exist, and can be measured is not subjective but objective. Quite frankly, if you had a detailed enough resolution chart, you'd still be measuring resolution way beyond the pixels on your sensor if you don't stop when aliasing starts to occur. It's a joke to suggest that heavily aliased data is real resolution when it's an undesirable artifact. If you could keep your resolution as-is, but not have any aliasing, would that not be beneficial to your images?
Graeme
[a href=\"index.php?act=findpost&pid=148502\"][{POST_SNAPBACK}][/a]
-
Hi Graeme,
This is where sampling theory and photographic reality collide. When I look at a b&w resolution chart at Nyquist a CFA sensor produces mush. When I shoot a landscape with grass, pine needles, leaves, etc., and enlarge a CFA print I get to a point where "mush" is very apparent. When nine converging lines on a resolution chart can no longer be resolved the AA filter produces a blob of indistinguishable grey plus color moire. In real-world photography this point is evident throughout when I reach the visible limits of resolution. These artifacts are far more objectionable to me than what you describe.
When my SD10 or SD14 stop resolving nine converging lines, I see five lines. Is this "real"? No - it's false detail but I still see "lines" not mush. Whether I see five blades of grass in my enlargement where there should be nine, or five pine needles where there should be nine on distant trees is largely unimportant as long as I still see something which resembles grass or pine needles or fur on wildlife (whose counting?). In my enlargements this false detail looks much more realistic than having an indistinguishable blob at the point of resolution exhaustion.
That's all well and good, but confusing aliasing artifacts with actual resolution, especially when publishing camera performance specifications, is deceptive and misleading. If you personally like the look of aliasing artifacts in your work, great. But not everyone shares your view.
Would I like more "resolution?" Sure. But no matter how much "resolution" I have, an antialiasing filter simply at some point in the equation produces moire and blur which I don't want in my enlargements.
I find it rather humorous that someone claiming to have a master's degree in physics thinks that an anti-aliasing filter causes moire. Its primary purpose is to prevent moire, a task which it does quite effectively. Does it introduce blur? Absolutely. But moire? Spare us.
-
LOL - what would you use for a lens? The 3.4 produces over 1500 lines, the 4.67 about 1760 so the 10 in that progression would produce around 3700 lines horiz and vertical - probably out resolve any current 35mm lenses??
Best regards,
Lin
[a href=\"index.php?act=findpost&pid=148436\"][{POST_SNAPBACK}][/a]
I can't follow the logic here, Lin. You can't have greater resolution than the number of pixels and the SD14 has only 4.67 million of them. If every one of them has a different value, that's a maximum of 4.67m different data points that comprise the image. The 40D could have 10m different data points.
The fact that the 40D cannot resolve up to the Nyquist limit is not the fault of the lens but is due to a combination of AA filter and pixel interpolation which is a design feature of the Bayer CFA. Interestingly, Phil Askey of dpreview has just completed a review of the 40D and comments that the images are a little softer than he'd like and suggests it's probably due to an AA filter which is too strong. Nevertheless, the 40D still manages an absolute resolution of 2200 lines horizontally. The 1Ds2 manages 2800 lines and I'm sure a 10mp foveon sensor could beat that, if noise is not too great.
-
Jonathan,
No one said you had to "share my view" - and this is not about any confusion at all between "actual resolution and artifacts" - those are your misguided assumptions. Also you apparently don't understand the difference between color moire and luminance moire.
Who published performance specifications which are deceptive and misleading??? Are you pretending to know more about the performance of the Foveon processor than Carver Mead and Dick Lyons? If so you are sadly mistaken.
Would you like for me to point out the quite obvious differences in the presence of color moire between same frames with the Foveon processor and a CFA processor. Would you then like to explain why there is no color moire with the Foveon? I don't "pretend" about anything, including my credentials, and I understand CFA technology very well. I clearly stated that the moire I'm referring to is beyond Nyquist but perhaps you missed that part. Also you apparently do not do enlarging and printing or you would already understand the effects of the combination of CFA sensor and AA filter designed to prevent moire below Nyquist limits combine to produce bad moire moire in that portion of unresolved data. Not only blurred but blurred with color moire. In copious enlargements this shows up as blotches of color smear making far worse artifacts than produced by the limited luminance moire and stairstep aliasing of the Foveon processor.
As I said earlier - no one is trying to force anyone to "like" aliasing or like Foveon. But what I am saying is that the resolution claimed for the Foveon processor is as accurate as any claimed for CFA sensors - anyone claiming otherwise is either ignorant of the facts or deliberately lying for some unknown reason.
Lin
That's all well and good, but confusing aliasing artifacts with actual resolution, especially when publishing camera performance specifications, is deceptive and misleading. If you personally like the look of aliasing artifacts in your work, great. But not everyone shares your view.
I find it rather humorous that someone claiming to have a master's degree in physics thinks that an anti-aliasing filter causes moire. Its primary purpose is to prevent moire, a task which it does quite effectively. Does it introduce blur? Absolutely. But moire? Spare us.
[a href=\"index.php?act=findpost&pid=148539\"][{POST_SNAPBACK}][/a]
-
When my SD10 or SD14 stop resolving nine converging lines, I see five lines. Is this "real"? No - it's false detail but I still see "lines" not mush. Whether I see five blades of grass in my enlargement where there should be nine, or five pine needles where there should be nine on distant trees is largely unimportant as long as I still see something which resembles grass or pine needles or fur on wildlife (whose counting?).[a href=\"index.php?act=findpost&pid=148533\"][{POST_SNAPBACK}][/a]
But they *don't* look like blades of grass or pine needles. They look like Tetris pieces *colored* like blades of grass and pine needles.
-
Hi Ray,
Don't confuse optical resolution with display pixel count. The SD14 has 14 million sampling sites (photosites). The true correlation between optical resolution and pixels is with the number of sampling sites not with the number of display pixels.
The "measured" b&w resolution of the "3.4 megapixel" SD9/SD10 Foveon processor is identical or better than a six megapixel CFA sensor. The number of display pixels is irrelevant. Because CFA sensors have approximately an equal number of display pixels to the number of sampling sites this display pixel count is commonly, but improperly used in conjunction with the word "resolution". It's only because of this misconception that it's been widely accepted. Does an 8 megapixel cell phone have the same optical resolution as a Canon 30D? Why? They each produce a display with eight million pixels. The measured optical resolution if far different. So it follows that a CFA sensor can have less resolution than the theoretical limits imposed by the number of sampling sites but not more. This is also true of the Foveon processor. The Foveon processor is actually less efficient than the CFA processor of equal photosite count. The 10.3 million Foveon processor used in the SD9/SD10 only produces a measured count of 1550 lines horizontal/vertical (in reality, it can't be more than the vertical display pixel count of 1512 lines) but human visual acuity and standardized resolution chart print error account for the variance as observed by testing (Phil Askey on dPReview's figures). The same margin for error applies to all other cameras tested so if I accept their numbers then I must also accept the "measured" Foveon numbers of 1550 lines vertical and horizontal. The "measured" color resolution is identical to the b&w resolution for this processor.
So there is no question that the Foveon processors produce the resolution measured as accurately as the CFA sensors. To assume otherwise is illogical.
A 10 megapixel (using a display pixel count) Foveon processor, if such a device could be built, would produce optical resolution of about 3700 lines (the 4.46 megapixel sensor produces 1760 lines). Remember, a Foveon processor with a 10 megapixel display would use slightly over 30m sampling sites.
Yes, the 40D has 10m sampling sites, the SD14 has 14m sampling sites. I agree with Phil - my 40D produces excellent color accuracy but noticeably softer images than my 1D Mark II or my 1DS. Also note that the resolution extinction figures are not great for the 40D. Compare them to the Sony Alpha 100 for instance. This is very likely an issue with AA filtering as Phil suggests.
For landscapes I very much prefer my D2XS or my Sigma SD14 to any of my Canon cameras including my 1DS which has a relatively "weak" AA filter (to use a term some dislike here - LOL).
Best regards,
Lin
I can't follow the logic here, Lin. You can't have greater resolution than the number of pixels and the SD14 has only 4.67 million of them. If every one of them has a different value, that's a maximum of 4.67m different data points that comprise the image. The 40D could have 10m different data points.
The fact that the 40D cannot resolve up to the Nyquist limit is not the fault of the lens but is due to a combination of AA filter and pixel interpolation which is a design feature of the Bayer CFA. Interestingly, Phil Askey of dpreview has just completed a review of the 40D and comments that the images are a little softer than he'd like and suggests it's probably due to an AA filter which is too strong. Nevertheless, the 40D still manages an absolute resolution of 2200 lines horizontally. The 1Ds2 manages 2800 lines and I'm sure a 10mp foveon sensor could beat that, if noise is not too great.
[a href=\"index.php?act=findpost&pid=148565\"][{POST_SNAPBACK}][/a]
-
The issue with detail as it approaches the Nyquist limit is do you wish either of these two alternatives:
1) the detail blends out to nothing - a blur - the camera is saying "I don't know what's here"
or
2) the detail continues, but, it's not necessarily accurate - you eventually end up with a fizz of uncorrelated pixels, the camera is saying - "there's something there, but what I'm telling you is not correct - or it might be correct, but I'm not telling you which"
From my POV: 1) above is how I see the world.
-
Hi Graeme,
Yes, for some images and some purposes I agree; for others I disagree which makes it fortunate that we have choices.
Best regards,
Lin
The issue with detail as it approaches the Nyquist limit is do you wish either of these two alternatives:
1) the detail blends out to nothing - a blur - the camera is saying "I don't know what's here"
or
2) the detail continues, but, it's not necessarily accurate - you eventually end up with a fizz of uncorrelated pixels, the camera is saying - "there's something there, but what I'm telling you is not correct - or it might be correct, but I'm not telling you which"
From my POV: 1) above is how I see the world.
[a href=\"index.php?act=findpost&pid=148576\"][{POST_SNAPBACK}][/a]
-
Do nine lines changing to five at Nyquist look like lines? Do they look like black and white "Tetris pieces?" Does the fact that a blade of grass is green suddenly make it different? Does the fact that a pine needle is green make it different?
Grass beyond Nyquist still resembles grass - pine needles beyond Nyquist still resemble pine needles and ferret fur beyond Nyquist still resembles ferret fur - LOL. You may see it quite differently and perhaps that's what happens when "you pixel peep" at levels never seen by one who views a print. My customers and users of the SD14 see it differently and isn't it a great thing that you have a choice?
Best regards,
Lin
But they *don't* look like blades of grass or pine needles. They look like Tetris pieces *colored* like blades of grass and pine needles.
[a href=\"index.php?act=findpost&pid=148570\"][{POST_SNAPBACK}][/a]
-
Don't confuse optical resolution with display pixel count. The SD14 has 14 million sampling sites (photosites). The true correlation between optical resolution and pixels is with the number of sampling sites not with the number of display pixels.
[a href=\"index.php?act=findpost&pid=148573\"][{POST_SNAPBACK}][/a]
Lin,
Surely the SD14 has 4.67m sites with regard to spatial resolution. The other 9.3m are stacked underneath. Remove the CFA and AA filter from a 40D and you'd get a spectacularly high resolution B&W image, far sharper than anything the SD14 could produce. Is this not so?
-
Hi Mike,
Actually the Sigma's all have a crop factor of 1.7x.
Best regards,
Lin
[a href=\"index.php?act=findpost&pid=148477\"][{POST_SNAPBACK}][/a]
Excellent, that'll be 13.5Mp (40.5M sample sites) then. That ought to outresolve my 5D every which way.
Please can I have one with a canon lens mount and is compatibility (so I don't have to buy new glass), sensible shooting speed (3fps or better), decent autofocus and a sensible number of focus points, a focus screen I can use manually (a split image?) and low noise up to iso 400 at least (I can always use the 5D for higher iso work).
After that you can start work on the high resolution version.
Oh, please keep it nice guys - they are only tools.
Mike
-
The same margin for error applies to all other cameras tested so if I accept their numbers then I must also accept the "measured" Foveon numbers of 1550 lines vertical and horizontal.
[a href=\"index.php?act=findpost&pid=148573\"][{POST_SNAPBACK}][/a]
Only a zealot would accept 1550 lines in a camera that only has 1512.
You'll believe anything that supports your faith, won't you.
The smart thing would have been to ask, "what's wrong with this test?".
-
If pigs had wings they "might" be able to fly.
Why don't you "try" removing the AA filter and CFA from a 40D and test it for yourself. Actually I'll tell you where you can have the AA filter removed for about $500.00. The point is you can't remove the CFA but you can remove the AA filter. And with it removed the answer is no, it doesn't produce spectacular "resolution" and it isn't a "sharper" image than the SD14 produces. It's been done with a number of Canon and Nikon CFA cameras.
We're not talking about theoretical constructs which don't exist, we're talking about real world photography with cameras which are available. Kodak made a nice six megapixel b&w pro-body camera with optional AA filtering. It had issues.
Best regards,
Lin
Lin,
Surely the SD14 has 4.67m sites with regard to spatial resolution. The other 9.3m are stacked underneath. Remove the CFA and AA filter from a 40D and you'd get a spectacularly high resolution B&W image, far sharper than anything the SD14 could produce. Is this not so?
[a href=\"index.php?act=findpost&pid=148620\"][{POST_SNAPBACK}][/a]
-
Only a zealot would fail to read plain English which is ostensibly their "first language" and attempt such a pathetic distortion in their reply.
Only an armchair "expert" would continue to argue ceaselessly about the performance of a camera they have never held in their hand. It never ceases to amaze me how you have all the answers about something you have so little experience with.
Lin
Only a zealot would accept 1550 lines in a camera that only has 1512.
You'll believe anything that supports your faith, won't you.
The smart thing would have been to ask, "what's wrong with this test?".
[a href=\"index.php?act=findpost&pid=148644\"][{POST_SNAPBACK}][/a]
-
Only a zealot would fail to read plain English which is ostensibly their "first language" and attempt such a pathetic distortion in their reply.
Only an armchair "expert" would continue to argue ceaselessly about the performance of a camera they have never held in their hand. It never ceases to amaze me how you have all the answers about something you have so little experience with.
[a href=\"index.php?act=findpost&pid=148647\"][{POST_SNAPBACK}][/a]
You don't have to hold a camera in your hands to see that it can't resolve more lines than it has, regardless of the definition of resolution.
You don't have to hold a camera in your hands to see nasty aliasing artifacts in the images. The aliasing issues of a camera are not going to change because I am holding it.
You are coming across as a very irrational person, arguing against cold hard facts, like the role of alignment in the "resolution" of an aliasing sampler. You have some kind of psychological or economic stake in the Sigma camera, or the images you've taken with it, and it has you writing some ridiculous things.
-
Jonathan,
No one said you had to "share my view" - and this is not about any confusion at all between "actual resolution and artifacts" - those are your misguided assumptions. Also you apparently don't understand the difference between color moire and luminance moire.
I'm familiar with both, and an anti-aliasing filter prevents the occurrence of both. What's your point?
Would you like for me to point out the quite obvious differences in the presence of color moire between same frames with the Foveon processor and a CFA processor. Would you then like to explain why there is no color moire with the Foveon?
I'm well aware of the reasons why not, and why a Bayer-pattern sensor exhibits color moire if it has no AA filter. I've shot over 120,000 frames with various Canon DSLRs (10D, 1D-MkII, and 1Ds), and I have not yet encountered color moire in any of them.
The 10.3 million Foveon processor used in the SD9/SD10 only produces a measured count of 1550 lines horizontal/vertical (in reality, it can't be more than the vertical display pixel count of 1512 lines) but human visual acuity and standardized resolution chart print error account for the variance as observed by testing (Phil Askey on dPReview's figures).
This is the deceptive and misleading stuff I'm talking about. It is impossible to capture actual image detail above the Nyquist limit. As has been pointed out by others already, at Nyquist you can either capture perfectly alternating black and white lines or 50% gray, depending on the alignment of the subject with the sensor. Above Nyquist, things get even worse, and counting the aliasing artifacts you get under such conditions as actual image detail in your "measurements" (as you and Phil are doing) is simply bullshit.
-
O.K. Jonathan,
You're the "expert" here. I'm truly surprised that you are not making your living testing cameras because you obviously know more about them than the people who build them and the one's who do make their livings testing them. I'll have to recommend you to dPReview so you can help "tutor" them with their reviews. LOL...
Wow, I'm impressed that you have taken over 120,000 frames and never encountered color moire in any of them. I took well over 100,000 frames before you were born so you're not impressing me with your "experience". I also have taken many more than 1,000,000 frames with digital cameras including six Canon dSLR's, numerous Kodak Pro bodies, Sigma SD9, SD10 and SD14 and several Nikon Pro bodies. I have encountered color moire on numerous occasions so what you are demonstrating is the "lack" of your experience.
Thanks for the "lesson"....
Lin
I'm familiar with both, and an anti-aliasing filter prevents the occurrence of both. What's your point?
I'm well aware of the reasons why not, and why a Bayer-pattern sensor exhibits color moire if it has no AA filter. I've shot over 120,000 frames with various Canon DSLRs (10D, 1D-MkII, and 1Ds), and I have not yet encountered color moire in any of them.
This is the deceptive and misleading stuff I'm talking about. It is impossible to capture actual image detail above the Nyquist limit. As has been pointed out by others already, at Nyquist you can either capture perfectly alternating black and white lines or 50% gray, depending on the alignment of the subject with the sensor. Above Nyquist, things get even worse, and counting the aliasing artifacts you get under such conditions as actual image detail in your "measurements" (as you and Phil are doing) is simply bullshit.
[a href=\"index.php?act=findpost&pid=148650\"][{POST_SNAPBACK}][/a]
-
Yep, I've got over $40,000 invested in Canon bodies and lenses, over $35,000 invested in Nikon and Kodak lenses and bodies and about $5,000 invested in Sigma so I have a huge economic investment in Sigma.
I use what works and I know what my images look like. You see "aliasing" everywhere you look. Fully half your posts on dPReview involve some discussion of aliasing and about 100% of your post concerning Sigma/Foveon concern your obsession with trying to convince people of the poor performance of the Sigma cameras. I'm not the one writing "ridiculous things" John, look in the mirror.
Nasty "aliasing" yep, that's your subjective opinion. It's not shared by the majority. In fact the "majority" of people listening to your constant obsessive/compulsive rhetoric about aliasing whom I've had conversations with think you are trolling the majority of the time.
You're all talk and no show, John. You talk the talk but you don't walk the walk. Were are your superior samples of CFA images with no aliasing. Where are your comparison shots of Foveon and CFA images? Were are "any" images you've taken?????
I'll give you this, you really do like to talk - LOL
Lin
You don't have to hold a camera in your hands to see that it can't resolve more lines than it has, regardless of the definition of resolution.
You don't have to hold a camera in your hands to see nasty aliasing artifacts in the images. The aliasing issues of a camera are not going to change because I am holding it.
You are coming across as a very irrational person, arguing against cold hard facts, like the role of alignment in the "resolution" of an aliasing sampler. You have some kind of psychological or economic stake in the Sigma camera, or the images you've taken with it, and it has you writing some ridiculous things.
[a href=\"index.php?act=findpost&pid=148648\"][{POST_SNAPBACK}][/a]
-
As far as I'm concerned - this discussion is over. I will not discuss more subjective issues with those who attempt to argue about images they have no experience with and who continually twist my words to make straw man arguments.
Use whichever camera pleases you and I'll do likewise.
Lin
-
Wow, I'm impressed that you have taken over 120,000 frames and never encountered color moire in any of them. I took well over 100,000 frames before you were born so you're not impressing me with your "experience". I also have taken many more than 1,000,000 frames with digital cameras including six Canon dSLR's, numerous Kodak Pro bodies, Sigma SD9, SD10 and SD14 and several Nikon Pro bodies. I have encountered color moire on numerous occasions so what you are demonstrating is the "lack" of your experience.
And how many of the cameras exhibiting moire had AA filters?
And with that resume, arguing that aliasing artifacts captured above Nyquist count as validly measured resolution is even more absurd. You should know better.
BTW, I was born in 1970. You exposed >100,000 frames before then?
As to comparisons, I've done them, both RAW-to-RAW and print to print. Foveon is noticeably better per-pixel than Bayer, but not to the extent that the Foveon kool-aid drinkers claim. When both are optimally sharpened, 3MP Foveon is about as good as 4-5MP Bayer. A well-executed 10D shot will beat a similarly well-done SD9 frame, but not by much. There's only so far Foveon's improved color resolution over Bayer can compensate for the spatial resolution disadvantage.
-
Hi!
I have downloaded the PIMA/ISO 12233 testcharts for the Nikon D100 and the Sigma SD10 as JPEGS from the DPReview site and run them trough Imatest.
MTF 50 Figures (corrected):
Nikon D100: 1272 LW/PH
Sigma SD10: 1556 LW/PH
MTF at Nyquist limit:
Nikon: 0.147
Sigma: 0.534
I have no opinion on the figures. I have taken the images from the same test.
I have also run the same test on Canon 10D and got 1328 LW/PH at MTF 50. I never shoot JPEG myself, but JPEG is what I can find at DPReview.
Best regards
Erik
And how many of the cameras exhibiting moire had AA filters?
And with that resume, arguing that aliasing artifacts captured above Nyquist count as validly measured resolution is even more absurd. You should know better.
BTW, I was born in 1970. You exposed >100,000 frames before then?
As to comparisons, I've done them, both RAW-to-RAW and print to print. Foveon is noticeably better per-pixel than Bayer, but not to the extent that the Foveon kool-aid drinkers claim. When both are optimally sharpened, 3MP Foveon is about as good as 4-5MP Bayer. A well-executed 10D shot will beat a similarly well-done SD9 frame, but not by much. There's only so far Foveon's improved color resolution over Bayer can compensate for the spatial resolution disadvantage.
[a href=\"index.php?act=findpost&pid=148692\"][{POST_SNAPBACK}][/a]
-
Hi!
I have downloaded the PIMA/ISO 12233 testcharts for the Nikon D100 and the Sigma SD10 as JPEGS from the DPReview site and run them trough Imatest.
MTF 50 Figures (corrected):
Nikon D100: 1272 LW/PH
Sigma SD10: 1556 LW/PH
MTF at Nyquist limit:
Nikon: 0.147
Sigma: 0.534
I have no opinion on the figures. I have taken the images from the same test.
I have also run the same test on Canon 10D and got 1328 LW/PH at MTF 50. I never shoot JPEG myself, but JPEG is what I can find at DPReview.
Best regards
Erik
[a href=\"index.php?act=findpost&pid=148699\"][{POST_SNAPBACK}][/a]
A MTF 50 of 1556 LP/PH is not attainable by a camera with a PH of 1512 pixels.
MTF at Nyquist of 0.534 indicates severe aliasing and possibly sharpening artifact if resharpening were applied. This is a know limitation of Imatest.
-
Why don't you "try" removing the AA filter and CFA from a 40D and test it for yourself.
[a href=\"index.php?act=findpost&pid=148645\"][{POST_SNAPBACK}][/a]
Lin,
With all respect, when I expressed that in the colloquial fashion, "If you remove the CFA and AA filter', I meant of course, if the CFA and AA filter is removed in the design stage. In other words, if Canon were to design a B&W only camera.
I'm addressing your statement that little purpose would be served in increasing the pixel count of the current 4.67mp of the SD14 because lenses are not good enough. This does not make sense to me because it's clear that real spatial resolution is limited to the 4.67m spatial photosites.
The fact that the Foveon sensor can 'make more' of those 4.67m pixels because they are real pixels which don't have to be interpolated is a separate issue from the resolving limits of lenses.
It seems clear to me that a B&W version of the 40D, without CFA and AA would deliver a much sharper and more detailed image that the SD14 could produce, and therefore a 10mp Foveon sensor ( with 30m collection points) would be required to match in color the resolution of the B&W designed 40D.
Are we now clear on that point?
-
Hi!
Thanks for pointing out this. I'm quite aware of both issues, but didn't feel competent enough to discuss. Wanted just to present the figures. Probably not a very good idea to present the data out of context.
Best regards
Erik
A MTF 50 of 1556 LP/PH is not attainable by a camera with a PH of 1512 pixels.
MTF at Nyquist of 0.534 indicates severe aliasing and possibly sharpening artifact if resharpening were applied. This is a know limitation of Imatest.
[a href=\"index.php?act=findpost&pid=148712\"][{POST_SNAPBACK}][/a]
-
I also have taken many more than 1,000,000 frames with digital cameras including six Canon dSLR's, numerous Kodak Pro bodies, Sigma SD9, SD10 and SD14 and several Nikon Pro bodies. I have encountered color moire on numerous occasions so what you are demonstrating is the "lack" of your experience.
Yep, I've got over $40,000 invested in Canon bodies and lenses, over $35,000 invested in Nikon and Kodak lenses and bodies and about $5,000 invested in Sigma so I have a huge economic investment in Sigma.
Rather curious statements, since your Canon image sample page (http://www.lin-evans.net/canon/canon.htm) only has D30 images, and your Nikon image sample page (http://www.lin-evans.net/nikon/nikon.htm) has only Coolpix 990 samples. And your Recent Images page (http://www.lin-evans.net/recent2/recent2.html) only has 10D and 1D samples. If your web site is intended to impress your clients, you might want to let them know you have something better than the Canon 10D, 1D, and D30 (only one of which breaks the 6MP barrier) and a 3.3MP Nikon point-and-shoot, all of which are several generations old technologically. And your site makes no mention and has no samples of anything shot with any Kodak or other brand pro bodies.
If these cameras are the straw man you're arguing against WRT Bayer sensors, I'm NOT impressed.
-
Jonathan,
Do you really have a point or are you simply showing more ignorance?
If I posted images from every digital camera I own and use or have owned and used over the years would that make you happy? LOL..
The site you apparently visited is a sample site with images from a few cameras which I find interesting. The CP990 and CP950 because of their use in digiscoping, the Sony DKC-ID1 and DSC-D700, a few Olympus models because of their place in history, the Canon D30 because it was Canon's first dSLR, the Canon 1D because it was Canon's first Pro-body. The Sigma SD9 and SD10 and Sony R1 because of their place in digital camera history.
I test cameras and peripherals for manufacturers and have done so for many, many years. I currently have over 30 digital cameras ranging from professional bodies to consumer level digicams. Some of the links are very old with images dating back to the 90's. I have used and discarded or sold another 30 or more over the years.
My customers don't even visit my samples site and could care less which equipment I use. They are only concerned with results and they pay me very well for giving them that.
In answer to your earlier question - yes, well over 100K images and well before 1970. In 1960 alone over 30K frames.
You appear to be a half way intelligent boy, if you would spend a bit of your time and intelligence actually "reading" what I said rather than trying to demonstrate the extent of your "knowledge" you would immediately see that you have misquoted me, constructed a straw man argument from your misquotes, and displayed your ignorance based on your misconceptions.
You are wrong about the comparative resolution of the SD9/SD10 viz CFA counterparts. Period... Numerous testers who do this for a living disagree with your conclusions and have concrete "proof" of their findings.
Impress you? Don't flatter yourself. I could care less whether you are impressed. Color moire with cameras having AA filters? Yes, in every case. I won't waste my time producing a sample using my 1DS (since you appear to have this camera) photographing a herringbone suit coat to demonstrate just how "easy" it is to produce color moire with the 1DS. Making statements like you made (120,000 frames - no moire) simply tells anyone with any true experience with this issue that you either have no clue or you have never shot subjects prone to moire. Some software companies have made a fortune creating software to ameliorate color moire for cameras with AA filters as well as for cameras without AA filters.
Don't bother replying unless you simply want to attempt to impress others with your vast knowledge of optical physics because this is my last comment on this issue.
Concerning the Foveon processor you simply have no clue.
Lin
Rather curious statements, since your Canon image sample page (http://www.lin-evans.net/canon/canon.htm) only has D30 images, and your Nikon image sample page (http://www.lin-evans.net/nikon/nikon.htm) has only Coolpix 990 samples. And your Recent Images page (http://www.lin-evans.net/recent2/recent2.html) only has 10D and 1D samples. If your web site is intended to impress your clients, you might want to let them know you have something better than the Canon 10D, 1D, and D30 (only one of which breaks the 6MP barrier) and a 3.3MP Nikon point-and-shoot, all of which are several generations old technologically. And your site makes no mention and has no samples of anything shot with any Kodak or other brand pro bodies.
If these cameras are the straw man you're arguing against WRT Bayer sensors, I'm NOT impressed.
[a href=\"index.php?act=findpost&pid=148746\"][{POST_SNAPBACK}][/a]
-
Jonathan,
Do you really have a point or are you simply showing more ignorance?
Just this:
Spatial aliasing artifacts are not the same as actual resolution above Nyquist. If you want to claim that Foveon caputes detail up to Nyquist, fine. I agree that it does. But claiming that any sampled system can capture valid data beyond Nyquist is disingenuous. If you have a mathematically demonstrable method for reliably and predictably capturing >Nyquist frequencies with a sampled system, you should patent it and make a huge amount of money not only in digital imaging, but digital audio as well. Otherwise, let's quit defending flawed testing methodologies that fail to distinguish between actual captured image detail and aliasing artifacts and claiming that said test methods yield accurate performance measurements. Not everyone agrees that aliasing artifacts are desirable, especially in the world of digital audio.
I did a comparison a few years ago between the SD9 and the 10D that compared RAW images and prints, and concluded that the 10D had a slight edge on the SD9 in overall detail and print sharpness. The 10D was softer straight out of the RAW converter, but could be sharpened to about 70% of the SD9's per-pixel image quality with deconvolution-based capture sharpening (I used Focus Magic as opposed to USM). I don't have access to the files and prints; they're in storage in the states right now. Can you point me to a comparison that shows a Foveon image that is superior to a Bayer image with more than twice the pixel count? I'd be happy to change my opinion if you can show me some evidence.
-
Yep, I've got over $40,000 invested in Canon bodies and lenses, over $35,000 invested in Nikon and Kodak lenses and bodies and about $5,000 invested in Sigma so I have a huge economic investment in Sigma.
I wrote, "You have some kind of psychological or economic stake in the Sigma camera ...". Do you know what the word "or" means, and how it is different from the word "and"?
The psychological part was the part I was leaning towards. A psychological investment does not necessarily parallel an economic one. A psychological one could simply be a need to identify with something perceived as the under-dog, non-mainstream, or something like an ego thing relating to the questioning of your perceptual faculty.
I use what works and I know what my images look like. You see "aliasing" everywhere you look. Fully half your posts on dPReview involve some discussion of aliasing and about 100% of your post concerning Sigma/Foveon concern your obsession with trying to convince people of the poor performance of the Sigma cameras. I'm not the one writing "ridiculous things" John, look in the mirror.
I am just trying to bring some honesty into the the threads where aliased photography is praised. I'd hate to see someone rush out and buy an aliasing camera and a cache of proprietary lenses because of fanatical claims, and then find out they don't like it. A person has a chance of listening to claims like yours, and my objections, and look closer to see if what I am saying is true or objectionable to them. Not everyone likes mortar that disappears and reappears in different parts of walls, or the edges of roof tiles that come and go due to luck of alignment, or natural textures that should have equal angles of opportunity being emphasized horizontally and vertically. They may not start noticing these things until after they have made an investment. And there are lots of folks who had Sigmas and could deal with the aliasing; if you don't remember those posts, then your memory is very selective.
IMO, people who see aliasing as image detail are living in a state of ignorant bliss. Are they lucky that they're not bothered? I don't know ... but I suspect that if they can't notice that points and edges are in the wrong places or inconsistently present at all, they can't be fully appreciating what I know of as Reality; they're just satisfying a need to feel like they have succeeded in focusing on the print or screen.
Nasty "aliasing" yep, that's your subjective opinion. It's not shared by the majority. In fact the "majority" of people listening to your constant obsessive/compulsive rhetoric about aliasing whom I've had conversations with think you are trolling the majority of the time.
And who do you converse with? Other fanatics.
I doubt that any more than a small minority of people who read my posts think I'm a troll, and I'm not a troll, because I say what I really think; not what I think will stir people up as an end in itself.
You're all talk and no show, John.
Show what? Did I ever say, "Lin, my pictures are better than yours"? It's just like you to shift contexts to create illusions. I've said times before, that a camera is capable of taking great photographs despite artifacts. AOTBE, however, I don't want aliased capture.
You talk the talk but you don't walk the walk.
What walk did I claim to be doing but am not? Again, you are trying to create illusions. You are coming across as a megalomaniac who thinks that he can bend reality through his mighty implications.
Were are your superior samples of CFA images with no aliasing. Where are your comparison shots of Foveon and CFA images? Were are "any" images you've taken?????
None of your business, and that is irrelevant. Again, you are trying to create an illusion; the illusion that we are having a photo competition, when in fact, I was only talking about aliasing and resolution.
I'll give you this, you really do like to talk - LOL
[a href=\"index.php?act=findpost&pid=148661\"][{POST_SNAPBACK}][/a]
I do like to write - about things that I have taken a special interest in.
-
It's such a pity that Lin seems to have taken offense at what is merely a bit of robust criticism.
I actually quite admire the Foveon system. In my view, that's what a pixel should be, a composite of a red, green and blue element. I was quite dismayed when I first learned that most digital cameras have to interpolate the other 2 colors from surrounding values. I thought that was sort of cheating .
Unfortunately, all systems have their weaknesses and I guess Foveon's is the absorption of the blue and red light that takes place as it passes through the layer(s) of silicon above. Without some technological breakthroughs I would guess that a 10mp sensor would be too noisy in the red and blue channels, unless the sensor size were increased. A 12mp full frame 35mm Foveon sensor might be viable though.
-
The advantages of the Foveon (and clever it is) are the co-sited nature of the components, so no interpolation is needed to figure what's going on where. The disadvantage is that silicon is used as a colour filter, silicon is not a terribly good colour filter. That leads to issues with noise, which is a shame.
Foveon does not have the advantage that it does not require an OLPF, which is how Sigma configure their Foveon cameras. Foveon themselves in their white papers state quite clearly that their Foveon sensor still requires an OLPF.
Graeme
-
Foveon does not have the advantage that it does not require an OLPF, which is how Sigma configure their Foveon cameras. Foveon themselves in their white papers state quite clearly that their Foveon sensor still requires an OLPF.
[a href=\"index.php?act=findpost&pid=148992\"][{POST_SNAPBACK}][/a]
Aliasing can be nasty, especially when it produces colored moiré patterns with Bayer pattern sensors. Fortunately for Sigma, the moiré with their sensor is monochrome.
Whether or not aliasing is readily visible depends on the nature of the subject. It is quite visible with regularly repeating patterns near the frequency of the sensor grid, such as fabrics or the tiles on a roof. Most naturally occurring objects such as landscapes do not have such patterns. An exception is feathers photographed close up.
The Nikon D70 has a weak OLPF and in my personal experience, I have never noted moiré in landscapes, but it has ruined some family pictures where strong moiré appeared in fabrics. The D200 has a weaker OLPF and moiré is less of a problem, but more sharpening is required.
The best sharpening algorithm for images degraded by OLPF is of interest. Bruce Fraser used unsharp masking with the radius selected according to the resolution of the sensor and the amount according to the strength of the filter. Jonathan Wienke has been experimenting with a deconvolution algorithm rather than unsharp masking and has reported good results. Comments on this matter are welcome.
Bill
-
Aliasing can be nasty, especially when it produces colored moiré patterns with Bayer pattern sensors. Fortunately for Sigma, the moiré with their sensor is monochrome.
It is possible to get moire that is colored, however. the difference is, true color moire can occur with B&W subjects. An LCD color display, shot at pixel-pitch-to-dot-pitch ratios near unity will result in color artifacts not directly representative of the subject as well, although they will not look as bizarre as when they are demosaiced from a CFA:
(http://www.pbase.com/jps_photo/image/37173876/original.jpg)
Whether or not aliasing is readily visible depends on the nature of the subject. It is quite visible with regularly repeating patterns near the frequency of the sensor grid, such as fabrics or the tiles on a roof. Most naturally occurring objects such as landscapes do not have such patterns. An exception is feathers photographed close up.
[a href=\"index.php?act=findpost&pid=149000\"][{POST_SNAPBACK}][/a]
I think it varies greatly with the viewer, too, and this, perhaps, is the reason why people feel so differently about aliasing. I can see aliasing in totally randomly-arranged subject lines and textures. I can see it quite clearly, even in sand. I see it as a propensity for things to line up and snap to a grid, and it looks very unnatural to me. I don't even have great eyesight; it's a brain-thing, IMO. My brain sees the weightedness of points and edges of light and dark, and sees them as artifically distributed. The "sharpness" of aliased imaging is about as useful to my brain as looking at an analog image through a sheet of glass with little magnifying glasses in rows and colums that show you only the center of the dots, but fill them out. Sharper transitions, but more erroneous representations of the areas.
Another problem with aliased capture, especially with aliased capture at low pixel resolutions, is that the captured image is very fragile in terms of freedom of resampling; you can not properly resample the low-res, aliased image at ratios like 60 to 80% or 120% to 170% and keep any of that content (real or fake) near the nyquist. It's re-distributed at lower frequencies, whereas, with a pixel-softer but higher-MP image, you have much greater flexibility as you are not smearing and redistributing anything at the original nyquist, because there is almost nothing there. One option is to use nearest neighbor, as that always maintains or increases pixel contrast in the result in randomly distributed subjects, but that results in even more aliasing, and this time, snap to left, snap to right, snap up, and snap-down to the new grid, depending on column and row in the original, as opposed to just snapping symmetrically in the original capture.
-
Another problem with aliased capture, especially with aliased capture at low pixel resolutions, is that the captured image is very fragile in terms of freedom of resampling; you can not properly resample the low-res, aliased image at ratios like 60 to 80% or 120% to 170% and keep any of that content (real or fake) near the nyquist. It's re-distributed at lower frequencies, whereas, with a pixel-softer but higher-MP image, you have much greater flexibility as you are not smearing and redistributing anything at the original nyquist, because there is almost nothing there. One option is to use nearest neighbor, as that always maintains or increases pixel contrast in the result in randomly distributed subjects, but that results in even more aliasing, and this time, snap to left, snap to right, snap up, and snap-down to the new grid, depending on column and row in the original, as opposed to just snapping symmetrically in the original capture.
[a href=\"index.php?act=findpost&pid=149103\"][{POST_SNAPBACK}][/a]
John,
That all sounds very nasty. One wonders how those very demanding MFDB users cope with all these problems. I think the pixel density of a 22mp digital back is no greater than that of the SD14 and I'm getting the impression that aliasing from Bayer type sensors that don't have an AA filter could be worse than aliasing from Foveon type sensors, at least with regard to color.
-
Another problem with aliased capture, especially with aliased capture at low pixel resolutions, is that the captured image is very fragile in terms of freedom of resampling; you can not properly resample the low-res, aliased image at ratios like 60 to 80% or 120% to 170% and keep any of that content (real or fake) near the nyquist. It's re-distributed at lower frequencies, whereas, with a pixel-softer but higher-MP image, you have much greater flexibility as you are not smearing and redistributing anything at the original nyquist, because there is almost nothing there.
Capturing at a high enough sampling rate that there is little near-Nyquist signal is a good way to increase the fidelity of a sampled waveform. Witness the growing popularity of 48-96KHz sample rates in digital audio as an example. Even though 44.1KHz is technically enough to capture up to 20KHz without aliasing, the fidelity of the recording suffers as you approach Nyquist. I use this principle when stitching panoramas; I upsize the source images to the largest ACR output dimensions before feeding them to the stitcher program to be bent, folded, spindled, and mutilated. The stretching and warping inherent to stitching is less destructive to the images as a result, and it is possible to make alignment adjustments <1 pixel (relative to the original resolution) when specifying anchor points and making any manual alignment adjustments after stitching (I always output to a layered PSD file so I can manually align and blend the layers together).
-
Capturing at a high enough sampling rate that there is little near-Nyquist signal is a good way to increase the fidelity of a sampled waveform. Witness the growing popularity of 48-96KHz sample rates in digital audio as an example. Even though 44.1KHz is technically enough to capture up to 20KHz without aliasing, the fidelity of the recording suffers as you approach Nyquist.
Frequencies near the nyquist, and near fractions like 3/5, 2/3, 4/5, and 3/4 of the nyquist suffer amplitude problems, and a sampled frequency sweep near these values results in periodic amplitude modulation. The things the don't tell you in basic sampling theory.
I use this principle when stitching panoramas; I upsize the source images to the largest ACR output dimensions before feeding them to the stitcher program to be bent, folded, spindled, and mutilated. The stretching and warping inherent to stitching is less destructive to the images as a result, and it is possible to make alignment adjustments <1 pixel (relative to the original resolution) when specifying anchor points and making any manual alignment adjustments after stitching (I always output to a layered PSD file so I can manually align and blend the layers together).
[a href=\"index.php?act=findpost&pid=149195\"][{POST_SNAPBACK}][/a]
Same goes for rotation, perspective correction, CA correction, lens correction, etc, etc. While not quite the same thing as originally sampling at a higher rate, I like to upsample images before doing these things (if not already oversampled), to reduce the damage.
-
I can see aliasing in totally randomly-arranged subject lines and textures. I can see it quite clearly, even in sand. I see it as a propensity for things to line up and snap to a grid, and it looks very unnatural to me. I don't even have great eyesight; it's a brain-thing, IMO. My brain sees the weightedness of points and edges of light and dark, and sees them as artifically distributed. The "sharpness" of aliased imaging is about as useful to my brain as looking at an analog image through a sheet of glass with little magnifying glasses in rows and colums that show you only the center of the dots, but fill them out. Sharper transitions, but more erroneous representations of the areas.[a href=\"index.php?act=findpost&pid=149103\"][{POST_SNAPBACK}][/a]
This sounds more like a description of pixelation than aliasing. If I photograph a single thin hair against a dark background, with a lens which out-resolves the sensor, the image might well be a single pixel wide, with all sorts of artificial looking consequences (jaggies, snapping to grid). Some people probably find these artifacts as objectionable as aliasing. The cure in both cases is the same -- blur the optical image so that it no longer out-resolves the sensor. However, in this case (unlike aliasing), the blurring can also be performed after the fact, on the digital image.
-
What you describe as pixilation is actually aliasing.
Graeme
-
No-one has really explained how photographers who spend spend huge sums of money on MFDBs and associated equipment in order to get the ultimate image quality, manage to deal with this very prevalent and nasty aliasing.
The only reasonable explanation I've heard so far is that professional photographers know what they are doing and deal with it by either reducing its effect through software or by recognising and avoiding situations where aliasing is going to be a problem.
Since time is money and quality is paramount for the professional, an extra couple of thousand for an AA filter on top of the $30,000 or so for the average MFDB would be money well spent, I would have thought, in order to avoid the hassles of aliasing, if it's as bad as Graeme, John and Joanathan claim.
-
No-one has really explained how photographers who spend spend huge sums of money on MFDBs and associated equipment in order to get the ultimate image quality, manage to deal with this very prevalent and nasty aliasing.
[a href=\"index.php?act=findpost&pid=149412\"][{POST_SNAPBACK}][/a]
I haven't really noticed any correlation between "professionals with MF digital" and the interest in details such as RAW data, and there aren't a lot RAW files available to examine, so there is a bit of a dearth of useful infomation.
The internet is full of all kinds of stuff, but very little of it is RAW files, especially RAW files shot for specific tests. Conversions have too much opportunity to hide artifacts.
-
No-one has really explained how photographers who spend spend huge sums of money on MFDBs and associated equipment in order to get the ultimate image quality, manage to deal with this very prevalent and nasty aliasing.
An AA filter is only needed when the lens outresolves the sensor. When a lens is designed to cover a larger image circle, the LP/mm where MTF is 50% decreases. Medium format lenses can resolve more total detail than a DSLR lens, but do so with a lower MTF at a given LP/mm than the equivalent DSLR lens. Given that DSLRs and MFDBs have pixel densities in roughly similar ranges, the MFDB is less likely to encounter aliasing because the lens is less likely to outresolve the sensor, meaning an AA filter isn't necessary. And if aliasing does occur, it's less noticeable when part of a 39MP image than part of a 10MP image, even if it's the same on a per-pixel level.
The other factor to consider is that software is getting better at dealing with color aliasing (moire) during Bayer interpolation and later on in the workflow.
-
An AA filter is only needed when the lens outresolves the sensor. [a href=\"index.php?act=findpost&pid=149564\"][{POST_SNAPBACK}][/a]
So at the end of the day, ___IF__ it's determined that, for example, the 1ds3 is lens limited, ie the sensor outresolves the lens(es) then it would be appropriate to wonder if an AA filter is necessary? And not to give Canon credit where they probably don't deserve it, ___IF___ we assume Canon executed a well engineered design of the sensor/AA filter "system" in the 1ds3 then the fact they included an AA filter means that the sensor does not out resolve the lens(es)?
(Ignoring for the moment whether __some__ lenses do or some don't outresolve...)
-
Given that the 1Ds MKII has the same pixel pitch (or thereabouts) of the 20D, I don't think the sensor is out-resolving the glass. I don't think it's lens limited (certainly not in the centre of the image).
Graeme
-
So at the end of the day, ___IF__ it's determined that, for example, the 1ds3 is lens limited, ie the sensor outresolves the lens(es) then it would be appropriate to wonder if an AA filter is necessary?
Yes it would. An AA filter is probably still appropriate, but it doesn't need to be as strong as on previous models, given that the likelihood of resolution being lens-limited is much higher. The more pixels you put behind a given lens, the weaker the AA filter needs to be to avoid aliasing.
-
Yes it would. An AA filter is probably still appropriate, but it doesn't need to be as strong as on previous models, given that the likelihood of resolution being lens-limited is much higher. The more pixels you put behind a given lens, the weaker the AA filter needs to be to avoid aliasing.
[a href=\"index.php?act=findpost&pid=149593\"][{POST_SNAPBACK}][/a]
This is what's puzzling about the impression Michael initially posted, because he felt that the AA filter on the 1Ds3 was stronger than the 1Ds2, not weaker. I don't recall hearing compaints about the 1Ds2 filter being too weak, so if Michael's impression is true you have to wonder what Canon was thinking.
-
This is what's puzzling about the impression Michael initially posted, because he felt that the AA filter on the 1Ds3 was stronger than the 1Ds2, not weaker.
And note that Michael removed those comments because he decided they were not accurate...
Also keep in mind that if the AA filter in the 1Ds-III is the same strength per-pixel as the 1Ds-II, it is still weaker overall on a per-image basis.
-
And note that Michael removed those comments because he decided they were not accurate...
Has he said that his observations about the relative AA filter strengths were not accurate? I thought he removed all the AA discussion from the review because his comments about in-camera JPEG being the reason for AA filters were inaccurate.
Also keep in mind that if the AA filter in the 1Ds-III is the same strength per-pixel as the 1Ds-II, it is still weaker overall on a per-image basis.
[a href=\"index.php?act=findpost&pid=149644\"][{POST_SNAPBACK}][/a]
I guess we'll just have to wait for some real, technical reviews to see.
-
An AA filter is only needed when the lens outresolves the sensor. When a lens is designed to cover a larger image circle, the LP/mm where MTF is 50% decreases. Medium format lenses can resolve more total detail than a DSLR lens, but do so with a lower MTF at a given LP/mm than the equivalent DSLR lens. Given that DSLRs and MFDBs have pixel densities in roughly similar ranges, the MFDB is less likely to encounter aliasing because the lens is less likely to outresolve the sensor, meaning an AA filter isn't necessary. And if aliasing does occur, it's less noticeable when part of a 39MP image than part of a 10MP image, even if it's the same on a per-pixel level.
[a href=\"index.php?act=findpost&pid=149564\"][{POST_SNAPBACK}][/a]
Those photographers who have invested in the new Digitar and Rodenstock lenses which have a reduced image circle more appropriate for the 'cropped' MF format of the P22, P25 and P45 must be hopping mad at the amount of aliasing they are getting. Some of those new MF lenses have an MTF response higher that that of any 35mm lens that I've seen. Their performance looks more like that of the Zuiko lenses for the 4/3rds format, you know, something like 70% MTF at 60 lp/mm (can't remember off-hand the precise figures).
-
Those photographers who have invested in the new Digitar and Rodenstock lenses which have a reduced image circle more appropriate for the 'cropped' MF format of the P22, P25 and P45 must be hopping mad at the amount of aliasing they are getting. Some of those new MF lenses have an MTF response higher that that of any 35mm lens that I've seen. Their performance looks more like that of the Zuiko lenses for the 4/3rds format, you know, something like 70% MTF at 60 lp/mm (can't remember off-hand the precise figures).
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=149734\")
Hi!
In some occasions there are moire problems with digital backs and not only with Schneider and Rodenstock (Linos) digital-optimized lenses. Software is getting better in dealing with moire in PP.
Two different moire examples can be found in [a href=\"http://luminous-landscape.com/forum/index.php?showtopic=13563]http://luminous-landscape.com/forum/index....showtopic=13563[/url]
(have a look at the links in the first post, 33mpix with different lenses) and http://luminous-landscape.com/forum/index....showtopic=12976 (http://luminous-landscape.com/forum/index.php?showtopic=12976) (22mpix).
I want to emphasize that I don't wish anyone to draw any conclusions about these particular backs or the photographers work based on these couple of pictures, these are only to demonstrate the different moire effects encountered in different situations.
Cheers,
J
-
Those photographers who have invested in the new Digitar and Rodenstock lenses which have a reduced image circle more appropriate for the 'cropped' MF format of the P22, P25 and P45 must be hopping mad at the amount of aliasing they are getting.
I don't think that's necessarily the case, at least most of the time. Improved RAW interpolation algorithms and software tools are reducing the likelihood of color aliasing being a problem even with troublesome subjects. And luminance moire is less unnatural looking when part of a ~40MP image than in a ~4MP image, at least at viewing distances where one is looking at the entire image.
-
Time to put up or shut up.
Let's see some Foveon images with all this "nasty aliasing" you've been peddling. I've been listening to you and John and Graeme wax eloquent about all the problems with Foveon images and not a hint of anything except hot air from all of you.
My Foveon images and numerous others hang at PMA in plain sight at greater than A0 size and not a single comment from the thousands of visitors about "nasty aliasing". In fact the vast preponderance of comments have been exceedingly complementary. I can show you a 2.5 gigapixel stitched image made with an SD14 which prints beautifully at "any" size. So how about the three of you get together and show all your results with a Foveon processor which substantiate your continuing BS??
Lin
I don't think that's necessarily the case, at least most of the time. Improved RAW interpolation algorithms and software tools are reducing the likelihood of color aliasing being a problem even with troublesome subjects. And luminance moire is less unnatural looking when part of a ~40MP image than in a ~4MP image, at least at viewing distances where one is looking at the entire image.
[a href=\"index.php?act=findpost&pid=149757\"][{POST_SNAPBACK}][/a]
-
Hi Folks
I can see this discussion about the Foveon sensor performance has ruffled a few feathers. I'm a veteran of this, being reknowned on the sigma forum of "another place" as being somewhat more cautious than is typical for that forum.
I'd like to add one or two thoughts of my own...
I recently purchased a used Sd9 in order to do my own testing of the Foveon concept (and perhaps to improve my credibilty in the endless debates).
My favourite camera is the 350D (mainly for the compact size and decent results) so I was very interested to see how the Sd9 compared to that camera. In a nutshell, the answer is pretty well. I found no discernable detail difference between the two cameras even pixel peeping at 100% and certainly not in prints.
There is a difference in the way aliasing is handled though: the Canon shows the usual "maze" type artifacts and moire while the SD9 tends to reproduce shrinking regular patterns as a smaller number of larger blocks at the resolution limit. Neither is ideal, but is really a matter of taste as to which is least bad.
Here's a link to a 100% crop comparison that shows the similarity in detail and the difference in aliasing: http://forums.dpreview.com/forums/read.asp...essage=23492002 (http://forums.dpreview.com/forums/read.asp?forum=1027&message=23492002)
In addition to the Sd9 v 350D shootout, I've been working with a Kodak 14nx owner who recently acquired a Sd14 to test how these cameras compare for resolution. I've been delayed in writing up the exercise because of other committements but as a quick summary: it is impossible to tell which camera produced the prints in prints up to 24x16 inches.
I'm not sure where this fits with the general debate but I've no doubt the SD14 is capable of producing prints every bit as good as a camera like the 5D - up to this print size at least.
I think at the end of the day it is important to reference all debates to physical evidence. Hopefully, I can get my act together and actually publish my evidence soon, but in the meantime, can I suggest that the debate here would be moved forward more efficiently if arguments were supported by some comparison images from real world photography?
Regards
Dave Millier
-
You need to stop bringing sensibility into the debate. We have analytical preconceived notions to argue about.
I think the only reason this thread still persists is because the shopkeeper is away for a few weeks ;-)
-
Hey, let's not forget the subjective entertainment value to help us wile away those boring work hours ...
-
So how about the three of you get together and show all your results with a Foveon processor which substantiate your continuing BS??
Are you referring to the BS where I agree that Foveon images are better per-pixel than Bayer? Or the BS where I point out that it is mathematically impossible for any sampled system to record an accurate reproduction of the input above Nyquist, which you seem to be claiming with your defense of the test results showing Foveon resolving more lines of resolution than it actually has?
Contrary to your claims, I am NOT anti-Foveon. Capturing all 3 primaries at each pixel site is better than capturing only one. But every comparison I have made and seen indicates that the advantages of Foveon, while real, are significantly exaggerated by the more enthusiastic proponents of the technology such as yourself. I have never seen any evidence to support the notion that Foveon pixels are twice as good as Bayer--4MP Foveon = 8MP Bayer in overall image quality and detail--my analysis indicates that 3MP Foveon = ~5MP Bayer when both are optimally sharpened. That is a solid performance advantage for the Foveon, at least on a per-pixel basis.
Foveon's crippling weakness right now is its low pixel count. No matter how good your ~4MP are, there's only so much detail you can cram into those pixels. Here's an experiment anyone can try at home to demonstrate this. Start out with a 1Ds RAW (http://www.visual-vacations.com/images/2004-07-14-0167.tif), and open it in Photoshop in 16-bit mode at its original resolution (4064x2704). Make a copy of it. Now downsize the copy using Bicubic Sharper, and print both files the same size (say 12x18 inches) and compare the prints. Undo the changes, and repeat the resizing, reducing by a greater percentage each time. How small can you resize the file before significant loss of image quality and detail starts showing up in the print?
When you size the image down to 60% of the original pixel count (77.46% of original linear dimensions) the damage to the image is very slight; one has fewer pixels, but they are better quality, and there is little difference to be seen in prints. Reducing to 50% of the pixel count (70.71% of original linear dimensions), one starts seeing a loss of fine detail. The individual pixels in the resized image are better than in the original image, but there aren't enough of them to keep the high-frequency detail and the image as a whole suffers. It follows then that given a perfect sensor that records all color channels for each pixel, one can achieve similar image quality with somewhere between 50-60% of the pixel count of a comparable Bayer sensor. This notion tracks fairly well with the comparisons I've done between the 10D and SD9; Foveon approaches, but does not quite match the IQ and detail of a Bayer sensor with double the pixel count.
If the most exuberant claims about Foveon were true (X Foveon photosites = X Bayer photosites), then you could reduce any Bayer-sensor image to 57.74% of its original pixel dimensions (which will leave you with 1/3 of the original pixel count) with little or no loss of image quality or detail when printed. But this is most certainly not the case; reducing 4064x2704 pixels down to 2347x1561 tosses out an awful lot of fine detail. The remaining pixels are very nice on a per-pixel basis, but the image as a whole has suffered irreparable damage; much of the fine detail present in the original capture is gone. There are simply not enough pixels to hold all of it. So the most extreme claims about Foveon are simply not credible.
I would LOVE to see a Foveon sensor with a pixel count that is more competitive with Bayer-sensor cameras. But given the current pixel count limitations of available Foveon sensors, I can do better with my 1Ds.
-
the Canon shows the usual "maze" type artifacts and moire while the SD9 tends to reproduce shrinking regular patterns as a smaller number of larger blocks at the resolution limit. Neither is ideal, but is really a matter of taste as to which is least bad. [a href=\"index.php?act=findpost&pid=149831\"][{POST_SNAPBACK}][/a]
Maze artifacts come from poor demosaic artifacts. The demosaicing I'm using the the RED camera will never produce such nasties! I don't know why people code demosaics that bad....
Graeme
-
I have no intention of disputing your numbers but it is conceivable that some bayer sensors are worse than you might expect, therefore making the Foveon seem a better performer.
As always, platonic theories must be tempered by boring facts. In my previous post, I provided some explicit evidence that the SD9 can compete with the 8MP 350D - something that surprised me. There appear to be only two explanations: either the Foveon is better than you (and me) expected or the Canon sensor is worse. Do you have any comment on this?
Are you referring to the BS where I agree that Foveon images are better per-pixel than Bayer? Or the BS where I point out that it is mathematically impossible for any sampled system to record an accurate reproduction of the input above Nyquist, which you seem to be claiming with your defense of the test results showing Foveon resolving more lines of resolution than it actually has?
[snip]
I have never seen any evidence to support the notion that Foveon pixels are twice as good as Bayer--4MP Foveon = 8MP Bayer in overall image quality and detail--my analysis indicates a bit less than 50% advantage, so that 4MP Foveon = ~5.5MP Bayer when both are optimally sharpened. That is a solid performance advantage for the Foveon, at least on a per-pixel basis.
[a href=\"index.php?act=findpost&pid=149856\"][{POST_SNAPBACK}][/a]
-
Are you referring to the BS where I agree that Foveon images are better per-pixel than Bayer? Or the BS where I point out that it is mathematically impossible for any sampled system to record an accurate reproduction of the input above Nyquist, which you seem to be claiming with your defense of the test results showing Foveon resolving more lines of resolution than it actually has?
Let me quote myself from a previous ansewer to Ray, and you can explain which part of this rather plain English you fail to understand............
-------------------------------------------
.........The 10.3 million Foveon processor used in the SD9/SD10 only produces a measured count of 1550 lines horizontal/vertical (in reality, it can't be more than the vertical display pixel count of 1512 lines) but human visual acuity and standardized resolution chart print error account for the variance as observed by testing (Phil Askey on dPReview's figures). The same margin for error applies to all other cameras tested so if I accept their numbers then I must also accept the "measured" Foveon numbers of 1550 lines vertical and horizontal. The "measured" color resolution is identical to the b&w resolution for this processor. ..............
----------------------------------------------
Contrary to your claims, I am NOT anti-Foveon. Capturing all 3 primaries at each pixel site is better than capturing only one. But every comparison I have made and seen indicates that the advantages of Foveon, while real, are significantly exaggerated by the more enthusiastic proponents of the technology such as yourself. I have never seen any evidence to support the notion that Foveon pixels are twice as good as Bayer--4MP Foveon = 8MP Bayer in overall image quality and detail--my analysis indicates a bit less than 50% advantage, so that 4MP Foveon = ~5.5MP Bayer when both are optimally sharpened. That is a solid performance advantage for the Foveon, at least on a per-pixel basis.
Did I say you were anti-Foveon? When and where did I say this?? If you would spend more time reading and understanding and less time with knee jerk reactions based on what you apparently "want" to believe you would know that what I said was that the "nasty aliasing" issue is BS.
Foveon's crippling weakness right now is its low pixel count. No matter how good your ~4MP are, there's only so much detail you can cram into those pixels. Here's an experiment anyone can try at home to demonstrate this. Start out with a 1Ds RAW (http://www.visual-vacations.com/images/2004-07-14-0167.tif), and open it in Photoshop in 16-bit mode at its original resolution (4064x2704). Make a copy of it. Now downsize the copy using Bicubic Sharper, and print both files the same size (say 12x18 inches) and compare the prints. Undo the changes, and repeat the resizing, reducing by a greater percentage each time. How small can you resize the file before significant loss of image quality and detail starts showing up in the print?
And what I'm telling you ois that Foveon doesn't have a "crippling weakness". Forget your "experiments" and look at the images and the huge prints at PMA and Photokina. Your entire argument is predicated on your experience with CFA cameras and your incorrect assumptions about Foveon "weakness". I'm telling you that people who use the cameras on a daily basis including CFA cameras and including the 1DS and 5D and D2X don't see a "crippling weakness". People just as skilled or more skilled than you with pixel peeping such as the author of Qimage (Mike Chaney) have put aside their full frame Canons and favor the 1DS for precisely IMAGE QUALITY.
When you size the image down to 60% of the original pixel count (77.46% of original linear dimensions) the damage to the image is very slight; one has fewer pixels, but they are better quality, and there is little difference to be seen in prints. Reducing to 50% of the pixel count (70.71% of original linear dimensions), one starts seeing a loss of fine detail. The individual pixels in the resized image are better than in the original image, but there aren't enough of them to keep the high-frequency detail and the image as a whole suffers. It follows then that given a perfect sensor that records all color channels for each pixel, one can achieve similar image quality with somewhere between 50-60% of the pixel count of a comparable Bayer sensor. This notion tracks fairly well with the comparisons I've done between the 10D and SD9; Foveon approaches, but does not quite match the IQ and detail of a Bayer sensor with double the pixel count.
Working with Foveon images and working with CFA images are totally different issues. Foveon not only matches the IQ and detail of a Bayer sensor with double the file size in pixels, the 3.4 megapixel Foveon exceeds the measured detail of the 6.3 mp Bayer and gives better IQ per the subjective evaluation of many, many photographers.
If the most exuberant claims about Foveon were true (X Foveon photosites = X Bayer photosites), then you could reduce any Bayer-sensor image to 57.74% of its original pixel dimensions (which will leave you with 1/3 of the original pixel count) with little or no loss of image quality or detail when printed. But this is most certainly not the case; reducing 4064x2704 pixels down to 2347x1561 tosses out an awful lot of fine detail. The remaining pixels are very nice on a per-pixel basis, but the image as a whole has suffered irreparable damage; much of the fine detail present in the original capture is gone. There are simply not enough pixels to hold all of it. So the most extreme claims about Foveon are simply not credible.
I don't know what the "most extreme claims" about Foveon are and really could care less. What I do know very well is real-world "results" based on personally inspecting many, many thousand Sigma prints and many hundred thousand CFA prints.
I would LOVE to see a Foveon sensor with a pixel count that is more competitive with Bayer-sensor cameras. But given the current pixel count limitations of available Foveon sensors, I can do better with my 1Ds.
That's "your" opinion. My opinion is that I can't and it's an opinion shared by many who have real-world experience with both.
Lin
[a href=\"index.php?act=findpost&pid=149856\"][{POST_SNAPBACK}][/a]
-
Foveon not only matches the IQ and detail of a Bayer sensor with double the file size in pixels, the 3.4 megapixel Foveon exceeds the measured detail of the 6.3 mp Bayer
Yes, when you buy into invalid measurement methodology that gives a higher resolution rating than the sensor has line pairs. That claim is no different than saying that a CD with a 44.1KHz sampling rate can meaningfully record a 25KHz audio signal.
As to the subjective opinion of professional photographers, I think its safe to say that a lot more pros are shooting with Canon DSLRs than Sigma, most because they believe they are getting better image quality with Canon. Sigmas aren't the only DSLR capable of capturing a large file that can print large with high quality.
-
You still can't read??? Still attributing things to me which I never said. Don't you get tired of being wrong? So what's your "valid measurement methodology?" Maybe you can inform the rest of the world and camera reviewers will get results "you" like - LOL. Perhaps if they keep testing they will find a way to substantiate measurements which you agree with. Strange thing is that when using the identical approach for CFA and Foveon they all seem to come up with similar results, non of which seem to agree with your assessment. Imatest analysis (oops - know issues causing the significantly better Foveon results viz 6mp Bayer CFA, hmmm), plain old b&w standardized resolution chart photos, color resolution chart photos, all seem to validate better Foveon results. Eye witnesses with prints in hand saying no difference between enlargements with 3.4 megapixel Foveon and 8 megapixel CFA - hmmm.
Yes, when you buy into invalid measurement methodology that gives a higher resolution rating than the sensor has line pairs. That claim is no different than saying that a CD with a 44.1KHz sampling rate can meaningfully record a 25KHz audio signal.
As to the subjective opinion of professional photographers, I think its safe to say that a lot more pros are shooting with Canon DSLRs than Sigma, most because they believe they are getting better image quality with Canon. Sigmas aren't the only DSLR capable of capturing a large file that can print large with high quality.
[a href=\"index.php?act=findpost&pid=149875\"][{POST_SNAPBACK}][/a]
I think it's safe to say that there are a lot more Canon, Nikon, Olympus, Pentax, Sony, Fuji and Kodak dSLR's than Sigma. I think it's also safe to say that the majority of professionals have a rather large investment in lenses and are unlikely to jump from one brand to another unless they are not getting results which are satisfactory. I get great results from a number of different dSLR's, fixed lens digicams and platforms and that includes Sigma. No "nasty artifacts" - no problems with "crippled" resolution or poor image quality.
I've yet to have a customer ask me which type of "camera" I used to produce an image. They either like it or they don't. The majority of the time they do or I wouldn't have had a successful career and enough income to buy all the equipment I need or want.
It's obvious we are never going to agree about this so not much sense continuing. You love numbers and tests - I love "results". I can't sell numbers and tests but I sure can sell results.
Yes, it's possible to get great large prints from CFA cameras as well. My D2XS produces incredible quality as does my SD14 and SD10 and my 1DS. They each have strengths and weaknesses. To my knowledge I've never said othewise. What I have said is that my SD14 produces, in my opinion, the best IQ of all for many of my purposes. I much prefer it for landscapes. It doesn't produce "nasty artifacts" any more than my CFA cameras do. It gives up nothing in IQ to any of my CFA cameras for subjects and under conditions it's amenable to. It's not a good low light, high ISO camera. It's not very useful for shooting birds in flight or for shooting sports. Horses for courses.....
Lin
-
Whatever happened to the Sigma compact? That was a camera I would have been and might still be interested in, a good way to test the Foveon sensor out.
And why are there no higher pixel count sensors from foveon? Seems a shame we don't have a 10mp (x3 if you like) foveon based dslr to choose from.
Quentin
-
Jonathan
I believe it was the host of this very site who published an article called "Cognitive Dissonance"...
I have crossed swords with Lin many times over Foveon but once again it all comes down to evidence. I've posted a comparison between 3.4MP X3 and 8MP bayer. It's not a lie - it shows the SD9 really does match the Canon for detail in every practical sense and this holds true for every base ISO comparison I've made.
I'd value your opinion of the evidence rather than the theory. Do you see anything in those crops that surprises you; do you have an alternative explanation of how the Foveon does so well?
I don't personally believe that the Foveon sensor is better than 2x more efficient than Bayer sensors in an absolute sense. My "theory" is that the 8MP canon sensor isn't really very much more detailed than its 6MP predecessors and that is the likely explanation of why the Foveon holds up. Whatever the reason, there is a piece of evidence that requires a better analysis than blindly quoting numbers...
Yes, when you buy into invalid measurement methodology that gives a higher resolution rating than the sensor has line pairs. That claim is no different than saying that a CD with a 44.1KHz sampling rate can meaningfully record a 25KHz audio signal.
As to the subjective opinion of professional photographers, I think its safe to say that a lot more pros are shooting with Canon DSLRs than Sigma, most because they believe they are getting better image quality with Canon. Sigmas aren't the only DSLR capable of capturing a large file that can print large with high quality.
[a href=\"index.php?act=findpost&pid=149875\"][{POST_SNAPBACK}][/a]
-
I have crossed swords with Lin many times over Foveon but once again it all comes down to evidence. I've posted a comparison between 3.4MP X3 and 8MP bayer. It's not a lie - it shows the SD9 really does match the Canon for detail in every practical sense and this holds true for every base ISO comparison I've made.
[a href=\"index.php?act=findpost&pid=149892\"][{POST_SNAPBACK}][/a]
I agree. One always needs the evidence otherwise it's not clear what people are referring to.
I've looked at your 100% crops comparing the 350D with the SD9 and I see relative advantages and disadvantages to both images.
To my eyes the SD9 image looks marginally sharper all over. Not in the sense of providing more detail but in the sense of greater accutance, as though one were comparing a prime lens with a zoom lens. I'm not sure if this is due to greater sharpening of the SD9 image, however.
The balustrade and the red tiled roof tend to highlight the SD9 aliasing problems. The vertical struts in each section of the balustrade are at least identifiable in the 350D image, although the number of struts visible in each section steadily diminishes as one moves from far right to left. In the SD9 image, the struts are clearly visible and countable only in the far right segment. Strong aliasing is very noticeable as one moves from right to left and before long everything is complete mush and color artifacts with no countable struts visible at all.
Nevertheless, I can see why some people would prefer the SD9 image and others the 350D image.
Thanks for providing the comparison images, Dave.
-
Hi Ray,
You have them backward. The top image is Sigma, the bottom Canon. The mush and color aliasing are on the Canon image. The Sigma image still shows the struts as identifiable only fewer than are actually there up until the sixth group segment.
The Sigma won't produce color moire. Jonathan might learn something by looking at this example of color moire he has never seen in 120,000 frames. Also it may be a learning experience for John and his "Tetris" pieces theory. The number of struts in the Sigma image diminish and are not correct, but they are still identifiable as "struts" very far past Nyquist. Most human observers looking only at the aliased struts when asked what they "thought" they were seeing would properly identify these as "struts" where seen in isolation in the Canon image it would be difficult to say unless the image was desaturated. Likewise with grass, pine needles, leaves, etc., which is one reason many like the Sigma landscapes. They don't see blobs of indiscriminate color and shapelessness in their landscape enlargements. They may see aliasing but the aliasing looks a lot better and much more closely resembles "reality" to the human brain than what is there with an AA filtered Bayer image. Of course one could always argue that an image shouldn't be enlarged to the degree that this color moire and shapeless blobs become visible - that's an issue for a different discussion.
Best regards,
Lin
I agree. One always needs the evidence otherwise it's not clear what people are referring to.
I've looked at your 100% crops comparing the 350D with the SD9 and I see relative advantages and disadvantages to both images.
To my eyes the SD9 image looks marginally sharper all over. Not in the sense of providing more detail but in the sense of greater accutance, as though one were comparing a prime lens with a zoom lens. I'm not sure if this is due to greater sharpening of the SD9 image, however.
The balustrade and the red tiled roof tend to highlight the SD9 aliasing problems. The vertical struts in each section of the balustrade are at least identifiable in the 350D image, although the number of struts visible in each section steadily diminishes as one moves from far right to left. In the SD9 image, the struts are clearly visible and countable only in the far right segment. Strong aliasing is very noticeable as one moves from right to left and before long everything is complete mush and color artifacts with no countable struts visible at all.
Nevertheless, I can see why some people would prefer the SD9 image and others the 350D image.
Thanks for providing the comparison images, Dave.
[a href=\"index.php?act=findpost&pid=150034\"][{POST_SNAPBACK}][/a]
-
Hi Ray,
You have them backward. The top image is Sigma, the bottom Canon. The mush and color aliasing are on the Canon image. The Sigma image still shows the struts as identifiable only fewer than are actually there up until the sixth group segment.
[a href=\"index.php?act=findpost&pid=150085\"][{POST_SNAPBACK}][/a]
LOL! Well, there you are then. I could have sworn the bottom image, which shows obvious aliasing, was from the SD9. Considering the SD9 is 3.4mp, it would not be difficult to convince me, had the SD14 been used in the comparison, that the Foveon image would have been preferred to the 8mp Bayer type in all respects... at least in these examples. One would need to compare many different types of images to be certain about this though.
The fact that the 350D image appears to have greater accutance (like comparing a prime with a zoom) is probably attributable to its additional 2mp. The SD9 was always considered to be on a par with a 6mp Bayer type.
-
Hi!
The lower image from 350D shows a very typical Canon color aliasing. With a sharp lens (and even the WA zooms are sharp enough) there is this problem when photographing subjects with regular high frequency patterns such as fabrics, buildings etc. I have encountered this numerous times with several Canons. Digital backs aren't immune either. Software is getting better in dealing with this (Hasselblad's Phocus is advertised as particularly effective in this) but in most cases it means less than perfect results and more PP.
Cheers,
J
-
The fact that the 350D image appears to have greater accutance (like comparing a prime with a zoom) is probably attributable to its additional 2mp. The SD9 was always considered to be on a par with a 6mp Bayer type.
[a href=\"index.php?act=findpost&pid=150192\"][{POST_SNAPBACK}][/a]
He said these are both 100% crops, which is a disadvantage to the 350D, as a smaller fraction of the image is used, which always results in lower quality.
The only fair way to compare on a monitor is to use the same FOV lens with both cameras, and upsample both so that they have the subject size, and step back and view from a distance, if necessary. Even when printing small, images are resampled, as the brain-dead drivers resample before printing, so you must print large for a fair comparison.
-
Hi Ray
I don't think there is any difference in detail. Acutance is down to sharpening differences. Straight out of the raw converter, the SD9 is much, much sharper - the Canon looks typically soft. After, uprezzing to match the Canon file size both were sharpened. I don't see much of a difference but this can be fine tuned to taste and application.
The main difference is the aliasing. Both cameras alias but in different ways. That balustrade is a beautiful test chart. Essentially what you are looking at is colour aliasing vs luminance aliasing. The canon shot is contaminated by false colour (despite the presence of the AA filter), the SD9 shows no colour artifacts. But you do get false detail.
In many ways this false detail lies at the heart of the debate. Is is good, bad or neutral - it clearly isn't "correct". You see the good and the bad in this shot.
The balustrade shows what happens really neatly. Essentially as the detail shrinks below Nyquist, it stops being reproduced correctly. Instead, you get false detail but the false detail is related to the real detail in a way that is useful at creating the impression of real detail. In a print of this shot what the eye sees is a convincing impression of higher resolution than the 350D (even though technically the detail is artifacts).
The downside is the tiled roof of the boathouse. This is a standard horizontal tiling patterm yet the SD9 reproduces it as a strange diagonal striped moire effect. The 350D does badly here too: if you blow this shot up it become evident that it has produced a patchy effect. Some areas show tiles, some a blurred smudge and some a kind of maze/hatched effect. Fortunately for the 350D, the whole thing is smudged enough to mix these effects together into something that looks passable.
And there I think lies the rub: which is best (or least worse) depends on subject, angle, magnification, personal taste. And we know exactly how this kind of ambiguity leads to squabbling.
Personally, I don't think either system copes well at the extremes - the ideal solution is higher pixel count and and AA filter for both systems in my view.
But the one thing that the comparison does show is that the Foveon sensor is very capable of producing detailed images; more so than some people (who haven't used the camera) give it credit for.
The lesson? Experiment and evidence trump theory and posturing all of the time.
LOL! Well, there you are then. I could have sworn the bottom image, which shows obvious aliasing, was from the SD9. Considering the SD9 is 3.4mp, it would not be difficult to convince me, had the SD14 been used in the comparison, that the Foveon image would have been preferred to the 8mp Bayer type in all respects... at least in these examples. One would need to compare many different types of images to be certain about this though.
The fact that the 350D image appears to have greater accutance (like comparing a prime with a zoom) is probably attributable to its additional 2mp. The SD9 was always considered to be on a par with a 6mp Bayer type.
[a href=\"index.php?act=findpost&pid=150192\"][{POST_SNAPBACK}][/a]
-
John
I'm a Canon user, I got the Sd9 to test rather than to use. My reputation on the DPR Sigma forum is as a sceptic - I have few friends there and have jousted with Lin on a number of occasions, so I have no agenda to promote Foveon or Sigma.
I prepared this comparison as best I could to show the differences. The crops are 100% crops of resized files. I compared every method I could think of to equalise the comparisons including uprezzing the Sigma, down rezzing the Canon and re-sizing both to some common size. Nothing really changes anything.
I'm in the middle of another comparison (Kodak 14nx vs SD14) and am using a different approach. Prints have been made from the two cameras and I'm making make high resolution flatbed scans of the prints to show.
It's work in progress at the moment but I'm happy to share the conclusion: in prints up to 24 x 16 inches (injket and lightjet) of distant landscapes shot with finest primes at optimum apertures, it is impossible to tell which camera took the shot.
The SD14 competes 100% with 14MP bayer in landscape type shots.
As I said in another post, evidence trumps theory. I'm happy to be challenged on this but only with hard evidence; you won't change my opinion by quoting sampling theory. It's science not philosophy...
He said these are both 100% crops, which is a disadvantage to the 350D, as a smaller fraction of the image is used, which always results in lower quality.
The only fair way to compare on a monitor is to use the same FOV lens with both cameras, and upsample both so that they have the subject size, and step back and view from a distance, if necessary. Even when printing small, images are resampled, as the brain-dead drivers resample before printing, so you must print large for a fair comparison.
[a href=\"index.php?act=findpost&pid=150221\"][{POST_SNAPBACK}][/a]
-
Dave, I like the photographs you took to examine the SD9 and 350D. They show up a number of issues in both cameras quite clearly, and are wonderful diagnostic images. Thanks for sharing them with us.
I looked in the metadata and see the Canon was converted with Bibble - is this correct? The demosaic is awful. Now, as I'm saying that, it's only fair to say that such a comment comes from my experience with the the RED camera project, and we're currently producing images like this: http://www.reduser.net/forum/uploaded/9_2kpomona_01033.jpg (http://www.reduser.net/forum/uploaded/9_2kpomona_01033.jpg) at 72fps. I'm not going to say the images are perfect - there's always room for improvement, and I'd also point out that we're recording to CF card through an advanced RAW compression scheme that's lossy (uncompressed or losslessly compressed is not really possible to record to a CF card at the fps and frame sizes we're doing) with a remarkably low data rate. If you're interested, here's the forum thread about those images: http://www.reduser.net/forum/showthread.php?t=5624 (http://www.reduser.net/forum/showthread.php?t=5624) if you're interested.
I started this thread to talk about aliasing. It's all the more important as moving images show up the issues of aliasing much more clearly than stills, and you can't in all honesty, go in and paint and fix up a movie the way you can on a still.
And then we took a track that started to talk about Foveons. That's an interesting topic in it's own right, but it's also a tad "religious". They get brought into any discussion on aliasing because the don't have an optical low pass filter. Most Bayer CFA cameras do, because without, the bayer pattern reconstruction algorithms don't always work as well as you'd like and chroma fringing artifacts are a problem. Interesting that Phantom, maker of high speed Bayer CFA movie cameras have announced a "upgrade" to put an OLPF in front of their sensor. And it's removable, so if you don't want / need it, you can remove it.
But, when we talk about images, and what we see in them, we are always talking about a "reconstructed" image. Look at a RAW Bayer CFA image, as I often have to do, and it looks awful - black and white, a mosaic of little dots, and very dark as it's linear light. It needs to go through complex algorithms to allow you to see the RGB image. Those algorithms are often proprietary, and include all kinds of processing from noise reduction, demosaicing, sharpening etc. etc. Similarly a RAW Foveon image would also look very dark, with funny colours, but it would not have the mosaic pattern as it uses co-sited samples. It also needs a number of steps to turn it into a viewable RGB image that looks like the scene you've photographed. The Foveon white papers hint at the processing involved, and pixel-peeping can tell you more about what they do, but for the most part, it's proprietary. I don't know precisely what they do, but I can make educated guesses as to the nature of the processing needed to generate the RGB image, and I suspect it's of the same order of complexity as we use for demosaicing Bayer CFA images, just different.
So, when you compare an image, there's a whole host of factors you're adding into the mix, not least all the algorithms that conver the RAW to RGB. So when you start a discussion about aliasing, and try to keep it about aliasing, how it works, how OLPFs work, what positive and negative effects they have, you can keep it all quite reasonable. When you look at the real world on how cameras and photographers work with their images, the number of variables go up through the roof, and it's very hard to get to the bottom of it all.
I'd dearly like to see some Sigma images from a camera modded to include an appropriate OLPF. Lin, do you know if anyone has done this? There are enough examples of Bayer CFA cameras modded to remove the OLPF, but to see the other way around, I think, would be enlightening.
Graeme
-
I looked in the metadata and see the Canon was converted with Bibble - is this correct? The demosaic is awful. [a href=\"index.php?act=findpost&pid=150248\"][{POST_SNAPBACK}][/a]
Let's not beat about the bush here, Graeme. Are you implying, if the 350D image had been converted using ACR, the comparison would be quite different and the result would favour the 350D image?
If so, then this is a serious criticism of Bibble. Perhaps users of Bibble would like to defend their choice of converter .
-
I'm not saying if ACR did it, it would be better. I am saying that if I did it, then it may not be overall better (just becasue I have no way of telling), but we'd not see those particular artifacts. Those artifacts are most probably not in the data, and most probably in the demosaic algorithm, probably too aggressive on some gradient adaptive code to try and extract too much resolution in an inappropriate way from the image.
Graeme
-
Here's an alternative version from RSE with colour noise reduction to suppress the colour component of the aliasing (sorry, I forgot to sharpen the Canon crop).
NB: It's not visible in this crop but elsewhere in the image, the NR causes some local desaturation of points of colour so it can't be used without penalty.
http://forums.dpreview.com/forums/read.asp...essage=23493083 (http://forums.dpreview.com/forums/read.asp?forum=1027&message=23493083)
What's interesting to me about this balustrade is that although RSE has cleaned up the colour fringing and superficially it looks good, there is still a problem. As you look to the left neither camera seem to be able to quite render the right number of vertical bars but for the first 3 panes the SD9 is more accurate and after that manages a reasonable approximation while the 350D rendering is inconsistent and gradually turns into mush.
Obviously, at the limit both cameras are aliasing and producing false detail in somewhat different ways. It's a matter of taste which is least bad but for my money the Sigma's attempt at least tries to approximate the underlying detail.
More resolution is the real answer, but I think at the very least this demonstrates that the Foveon approach holds up pretty well and some of the criticisms of its aliasing performance (often from people who have never used a Sigma) are a bit over-blown.
Hopefully, people will accept that blanket statements along the lines of "this approach doesn't work because of xyz" need to be tested against the real world.
ps
I won't be abandoning my bayer cameras, however. At the current stage of Foveon development, it is competitive but I don't believe it is enough to compensate for the rather budget Sigma bodies and the fact I'm invested in other systems. But I think for someone starting out, the Sigma is worth a good look.
I'm not saying if ACR did it, it would be better. I am saying that if I did it, then it may not be overall better (just becasue I have no way of telling), but we'd not see those particular artifacts. Those artifacts are most probably not in the data, and most probably in the demosaic algorithm, probably too aggressive on some gradient adaptive code to try and extract too much resolution in an inappropriate way from the image.
Graeme
[a href=\"index.php?act=findpost&pid=150254\"][{POST_SNAPBACK}][/a]
-
Dave, thanks for that update. The new RSE image certainly looks a lot better than Bibble. I still see issues with, what looks like, their local gradient based approach, but the artifacts are much more subtle than Bibble, and indeed, it does handle the chroma oh-so-much better.
Personally speaking, I prefer the Bayer CFA image, unless I pixel peep and look at a small area, when I begin to see some Bayer artifacts that I don't like. I think that can be solved with better algorithms, but who knows? The Sigma image, when viewed close up, doesn't have those issues. However, the roof tiles aliasing are as much an artifact as the confused gradient interpolation I see in the Bayer image.
Everyone has a different tolerance for different kinds of artifacts. Some people will look at the Sigma image and see nothing wrong. Others will see everything wrong with it. Similarly for the Canon image. The criticism of aliasing is very much a two pronged thing: a) the technical aspect, where I think we can all agree we'd rather not have aliasing, all other things being equal and the mind / brain side of things, where some of us can't stand the way aliasing looks, and don't mind so much the effects of the OLPF, and others who find the sin of commision - ie aliasing effects, is worse than the sin of ommission, ie ommission of detail through using an OLPF.
The thing with a) above, is that all things are not equal. Add and OLPF and your measured resolution goes down, so you need more pixels or smaller pixels, and that effects FOV, cost or noise. All things are never equal.
When you design your own camera, you get to pick your compromises. When you don't you've got to pick someone elses compromises. That's why Nikon and Canon and Sigma and MF cameras exist - everone has their own internal set of compromises they can make, and they're all different.
Graeme
-
Perhaps the impressive point here is just how much detail appears to be captured by this Foveon sensor which has only 3.4mp in the one spatial plane as opposed to the 8mp of the 350D.
Neither image is perfect with regard to aliasing, but it seems to be nit picking to argue which is preferrable.
The Foveon design, without AA filter, would be the way to go in my opinion. However, perhaps a technological breakthrough is required before sensors with great pixel density can be constructed, considering the loss of energy which the red and green light suffers as it passes through the layers of silicon.
Bayer type sensor also block a lot of light by virtue of the CFA, causing the sensors to be less sensitive than they otherwise could be (with B&W images, for example). But the inefficiency of the Foveon system, with regard to red and green light, seems to be greater.
-
I'm in the middle of writing up the results of a Kodak 14nx vs Sigma SD14 shootout which will be published on my website shortly.
To cut a long story short, the image quality of the SD14 is more than just a pixel count advance over the SD9. Dynamic range, detail, noise and colour reproduction are all improved. The conclusion: 24 x 16 inch prints from the two cameras are indistinguishable.
I actually think the 14nx has a couple of percent more resolution in certain very specific circumstances but you would be printing 10 foot wide prints and scrutisinising with a loupe to see it.
The Sd14 is a bit of a bargain from a resolution point if view; it's just a pity the overall package is a little bit "budgety".
Perhaps the impressive point here is just how much detail appears to be captured by this Foveon sensor which has only 3.4mp in the one spatial plane as opposed to the 8mp of the 350D.
Neither image is perfect with regard to aliasing, but it seems to be nit picking to argue which is preferrable.
The Foveon design, without AA filter, would be the way to go in my opinion. However, perhaps a technological breakthrough is required before sensors with great pixel density can be constructed, considering the loss of energy which the red and green light suffers as it passes through the layers of silicon.
Bayer type sensor also block a lot of light by virtue of the CFA, causing the sensors to be less sensitive than they otherwise could be (with B&W images, for example). But the inefficiency of the Foveon system, with regard to red and green light, seems to be greater.
[a href=\"index.php?act=findpost&pid=150490\"][{POST_SNAPBACK}][/a]
-
These are great posts about the Sigma SD14. It's the first time I've been tempted to buy one, and buy one I would if I could get it with a Canon or Nikon lens mount. Bummer it doesn't exist.
-
I've been attracted by the Foveon concept, much as many other have, but one wonders why it hasn't captured a larger share of the market than appears. Could lenses be the key issue? How does the quality of Sigma lenses normally used with a camera such as the Sd14 stack-up compared with the better of the Canon L series? Would this not also be an important factor affecting comparisons? I must say, I don't find the comparisons of the images on DP Review determinative because they are small and hard to interpret on a display. Perhaps one of these days I'll seem some (at least A3) prints of real-world subjects captured and processed under good comparative test conditions. Meanwhile, I'm more persuaded by evidence of what users have found to pass muster in critical market segments.
Mark
-
Sigma make great and affordable lenses, so I don't see quality of glass being the issue. I've got the feeling that ergonomics and speed of the bodies might be the major factor? Not that Canon get the ergonomics totally right either....
I've not seen a proper dynamic range test for the Sigma - anyone got a Stouffer chart shot they could share?
Graeme
-
Could links the original Canon and Sigma RAWs be posted? I'd like to do a comparison converting both with ACR, and using my deconvolution-based capture sharpening approach.
-
I've been attracted by the Foveon concept,
[a href=\"index.php?act=findpost&pid=150556\"][{POST_SNAPBACK}][/a]
Which concept? The disregard for aliasing issues, or the concentric RGB capture? They are unrelated to each other, except for the fact that concentric RGB requires weaker AA filtering than a CFA, relative to pixel pitch.
Concentric RGB is not a concept that is unattractive to many people. In fact, it's the way most people assume digital cameras work until they find out otherwise. The Sigma implementation, however, has luminance aliasing (it doesn't have to; it's not part of the concentric RGB concept) and also has poor color discrimination in certain wavelengths, and a zero-sum blotchy noise between the blue and green channels (blotches are negative in one channel while positive in the other).
If some company came up with a concentric RGB sensor without the aliasing and color noise issues, as well as higher MP counts, it would be much more readily accepted than what Sigma is doing now.
-
Which concept? The disregard for aliasing issues, or the concentric RGB capture? They are unrelated to each other, except for the fact that concentric RGB requires weaker AA filtering than a CFA, relative to pixel pitch.
Concentric RGB is not a concept that is unattractive to many people. In fact, it's the way most people assume digital cameras work until they find out otherwise. The Sigma implementation, however, has luminance aliasing (it doesn't have to; it's not part of the concentric RGB concept) and also has poor color discrimination in certain wavelengths, and a zero-sum blotchy noise between the blue and green channels (blotches are negative in one channel while positive in the other).
If some company came up with a concentric RGB sensor without the aliasing and color noise issues, as well as higher MP counts, it would be much more readily accepted than what Sigma is doing now.
[a href=\"index.php?act=findpost&pid=150717\"][{POST_SNAPBACK}][/a]
John, what I had in mind is the concentric RGB capture. Thanks for the insight into the issues with the present implementation.
-
The co-sited nature of the Foveon sensor is indeed a most wonderful advance. However, it doesn't come without a price. The price is that silicon is not the best colour filter and the processing needed to extract colour (not luma) from the sensor is a lot more extreme than for a Bayer CFA, and that exaggerates any chroma noise.
Bayer sensors rely on us being less sensitive to chroma resolution than luma. But so does Foveon in the noise reduction used on the chroma.
With a good Bayer sensor you can do the image processing without any noise reduction of any kind, and still get a very nice looking image at base ISO. I, obviously, have access to a lot of Bayer pattern RAW data. I don't have such data for Foveon, but the Foveon white papers outline the processing needed and show how the luma and chroma get treated seperately, and the noise reduction in the basic image processing path.
Graeme
-
The co-sited nature of the Foveon sensor is indeed a most wonderful advance. However, it doesn't come without a price. The price is that silicon is not the best colour filter and the processing needed to extract colour (not luma) from the sensor is a lot more extreme than for a Bayer CFA, and that exaggerates any chroma noise.
Bayer sensors rely on us being less sensitive to chroma resolution than luma. But so does Foveon in the noise reduction used on the chroma.
With a good Bayer sensor you can do the image processing without any noise reduction of any kind, and still get a very nice looking image at base ISO. I, obviously, have access to a lot of Bayer pattern RAW data. I don't have such data for Foveon, but the Foveon white papers outline the processing needed and show how the luma and chroma get treated seperately, and the noise reduction in the basic image processing path.
Graeme
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=150747\")
The 1Ds, 1DsII, 1DsIII and 5D, from all I've seen and experienced, allow us to expose far above "base ISO" without a need for noise reduction. From what you and John are saying it looks like a present-day situation where we win some and lose some in different ways with Bayer versus Foveon, the net result being the kind of "real world" comparisons discussed above.
What would really intrigue me as a 1DsIII without the AA filter. You can get a taste for this issue here: [a href=\"http://MaxMax.com]http://www.maxmax.com/hot_rod_visible.htm[/url] They don't seem to offer this service for the 1 series - probably too risky for them, but fascinating nonetheless. The Mamiya ZD is available in a version without the AA filter, but noise becomes a problem above 100 ISO - those nasty trade-offs again!
-
The other great thing about the Foveon sensor is that it demands less resolution from the lenses because you're not wasting any of it. So if you have, for example, a full frame 35mm Foveon 10 megapixel sensor (i.e. 30 mllion photosites), the sensors would be less dense than with a Canon 5D (thus stressing the lenses resolving power less than the 5D), but should produce an equivalent of somewhere between 20-30 megapixels of Bayer data. This is the big problem with the Bayer sensors at this point -- they demand too much from imperfect lenses, which is why, that I surmise Canon and Nikon will all eventually go the Foveon, or similar, route, but probably not until the 1DsIV comes out in 4 years.
-
The other great thing about the Foveon sensor is that it demands less resolution from the lenses because you're not wasting any of it. So if you have, for example, a full frame 35mm Foveon 10 megapixel sensor (i.e. 30 mllion photosites), the sensors would be less dense than with a Canon 5D (thus stressing the lenses resolving power less than the 5D), but should produce an equivalent of somewhere between 20-30 megapixels of Bayer data. This is the big problem with the Bayer sensors at this point -- they demand too much from imperfect lenses, which is why, that I surmise Canon and Nikon will all eventually go the Foveon, or similar, route, but probably not until the 1DsIV comes out in 4 years.
[a href=\"index.php?act=findpost&pid=151245\"][{POST_SNAPBACK}][/a]
I wonder whether this line of reasoning is correct. Is there not a mixing-up of resolution and interpolation here? I'm under the impression that resolution is a luminosity-based concept which depends on the number of megapixels (strictly defined, and devoid of perceptual factors); hence a 10MP sensor has less resolution than a 16 MP sensor. The interpretation of colour, on the other hand, is concentric with Foveon and Bayer matrix with other sensors, meaning that the ultimate hue of each pixel is read and calculated differently, but I question whether that has an impact on the sensor's basic resolution.
-
I wonder whether this line of reasoning is correct. Is there not a mixing-up of resolution and interpolation here? I'm under the impression that resolution is a luminosity-based concept which depends on the number of megapixels (strictly defined, and devoid of perceptual factors); hence a 10MP sensor has less resolution than a 16 MP sensor. The interpretation of colour, on the other hand, is concentric with Foveon and Bayer matrix with other sensors, meaning that the ultimate hue of each pixel is read and calculated differently, but I question whether that has an impact on the sensor's basic resolution.
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=151278\")
Well, Mark, Phil Askey of dpreview certainly appears to be of the opinion that the 3.4mp of the SD9 and SD10 equal in resolution and detail the 6mp of the Canon 10D. Check out his review at [a href=\"http://www.dpreview.com/reviews/SigmaSD10/page18.asp]http://www.dpreview.com/reviews/SigmaSD10/page18.asp[/url]
Whilst the Foveon sensor appears to resolve slightly beyond the Nyquist limit, such detail is represented at a lower frequency, as shown in the example of the vertical struts in the balustrade from Dave. Even if you set aside the issue of just how relevant that detail is beyond the Nyquist limit, the fact is the Bayer type sensors do not resolve up to the Nyquist limit. I think the figure is something like 2.8 pixels per line pair.
It would seem therefore that a 10mp Foveon sensor should be at least equal to an 18mp Bayer type. Since there's a law of diminishing returns at work with regard to system resolution as one increases the resolving power of just the sensor without simultaneously increasing the resolving power of the lens, then it would be reasonable to expect a 10mp full frame Foveon to have greater resolution than a full frame 18mp Bayer type.
-
Ray, thanks for recalling Phil's review. On re-reading the conclusions, it seems to be that the most distinguishing feature in terms of resolution is perhaps less the method of determining hue, but more the absence of an AA filter on the Foveon sensor. I can relate to that. Did you see what a difference the AA filter makes (on www.maxmax.com)?
Mark
-
A Bayer CFA with AA will measure somewhere between 70% to 80% of the rated linear resolution, depending on algorithm and AA. A combination I know well comes out at 75%, so I'll use that figure. Therefore our 10mp Bayer CFA would give about the same measured resolution as a 10 * 0.75 *0.75 = 5.6mp Foveon without AA. That 5.6mp Foveon needs 5.6 * 3 photosites to get it's resolution, which would be: 16.875mp. Our Bayer CFA needs 10mp photosites to get it's resolution. The difference is that shrinking the Bayer CFA down to 75% reduces noise a bit, and sharpens up the image a bit (assuming a good downsampling filter is used). The downsampling doesn't add aliasing (noticibly) as we're well anti-aliased from the OLPF. Now, theoretically, the Foveon should have higher chroma resolution. However, the noise reduction used in the Foveon processing might not let the sensor keep that advantage to it's full. All this leads to, is a justification of the x2 factor that many people use to compare Foveon MP with Bayer OLPF MP. It's a rule of thumb, but a reasonable guide.
Graeme
-
Graeme,
A few things I don't understand:
(1) Why are you multiplying 10MP by 0.75^2 rather than simply 0.75?
(2) Also, why does a Foveon need 3*5.6mm photosites to get its resolution? Is there a separate photosite for each primary for each pixel, or is it three filters on top of the same photosite?
(3) When you knock 25% off the Bayer, you aren't really shrinking it are you, you are just discounting its resolution because of the "rounding the edges" impact of the AA? If so, where does this sharpening effect come from? I thought the AA blurs, and that is why you discount 25% of its native resolution.
Hence I don't really understand your conclusion.
Mark
-
Ray, thanks for recalling Phil's review. On re-reading the conclusions, it seems to be that the most distinguishing feature in terms of resolution is perhaps less the method of determining hue, but more the absence of an AA filter on the Foveon sensor. I can relate to that. Did you see what a difference the AA filter makes (on www.maxmax.com)?
Mark
[a href=\"index.php?act=findpost&pid=151295\"][{POST_SNAPBACK}][/a]
Mark,
I did have a look at those hot-rod images with AA filter removed and I didn't think there was a significant increase in resolution. The differences were rather subtle. I don't believe the lack of an AA filter alone can explain the extra resolution of the Foveon sensor. Rather I think it's a combination of the absence of an AA filter plus the fact the color values are not interpolated as in the Bayer system.
-
Ray, maybe I'm a bit thick today, but frankly I just don't see how interpolating colour information between neighbouring pixels to calculate the hue of any one pixel should affect resolution. Each photosite corresponds to a pixel and captures a certain amount of greyscale information depending on the light that strikes it from the lens - I would have thought that defines the resolution, not the calculation of the hue.
-
At the risk of sounding like a broken record, trust the evidence, not the theory (there are endless numbers of factors overlooked in simple theories).
I have a draft article comparing Sd14 (4.7MP X3) and Kodak 14nx (14MP CFA).
In prints up to 24 x 16 inches there is no practical difference in resolution. I would love to show you the article but it is with my collaborator for comment at the moment but here's one example:
http://www.whisperingcat.co.uk/scans/big1.jpg (http://www.whisperingcat.co.uk/scans/big1.jpg)
These are flatbed scans from prints. I don't know which is which...
Ray, maybe I'm a bit thick today, but frankly I just don't see how interpolating colour information between neighbouring pixels to calculate the hue of any one pixel should affect resolution. Each photosite corresponds to a pixel and captures a certain amount of greyscale information depending on the light that strikes it from the lens - I would have thought that defines the resolution, not the calculation of the hue.
[a href=\"index.php?act=findpost&pid=151339\"][{POST_SNAPBACK}][/a]
-
Dave,
You are not sounding like a broken record, and what you are saying reflects very much what's going through my mind. It's not self-evident how stacking pixels on top of eachother improves RESOLUTION compared with arraying them side by side and interpolatinfg colour information. I've seen the technical papers on Foveon's website, can't say I really understand it all, but yes, clearly there are numerous variables at play and in the final analysis what counts are (i) the measurable resolution in the file, and (ii) the quality of what emerges from the best printers and papers we have for reproducing the resolution of high frequency images.
-
In prints up to 24 x 16 inches there is no practical difference in resolution. I would love to show you the article but it is with my collaborator for comment at the moment but here's one example:
http://www.whisperingcat.co.uk/scans/big1.jpg (http://www.whisperingcat.co.uk/scans/big1.jpg)
These are flatbed scans from prints. I don't know which is which...
[a href=\"index.php?act=findpost&pid=151343\"][{POST_SNAPBACK}][/a]
Dave,
In resolution comparisons, shouldn't you be comparing subject matter with fine detail such as cat's whiskers?
-
Ray
The sample comparison is of around 1 x 1 inch section from a 24 x 16 inch print scanned at 240ppi.
On a typical monitor it is equivalent to examining the print with a low power loupe.
We shot 12 pairs of prints in total, most being landscapes, forests, trees, foliage. This was the lone sample of architecture. There were a couple of macros, too.
All indistinguishable. Three of them (including this one) particularly so for me as the photographer didn't label them!
Of course there are lots of variables in testing so I don't claim any kind of scientific veracity but for the purpose of making your mind up whether the Sd14 is for you, the tests are good enough.
The main lesson for me from all the tests is not that the Foveon sensor competes well with high end cameras but that the end results don't look particularly different. All this stuff about aliasing problems is clearly irrelevant unless you are aiming for ridiculous enlargements (several feet wide).
I would like to add, though, that there is a clear difference in quality from both cameras in 19 x 13 and 24 x 16 prints. IMO, 24 x 16 is a size too far for 12MP class cameras if optimum quality is required.
Dave,
In resolution comparisons, shouldn't you be comparing subject matter with fine detail such as cat's whiskers?
[a href=\"index.php?act=findpost&pid=151349\"][{POST_SNAPBACK}][/a]
-
I would like to add, though, that there is a clear difference in quality from both cameras in 19 x 13 and 24 x 16 prints. IMO, 24 x 16 is a size too far for 12MP class cameras if optimum quality is required.
[a href=\"index.php?act=findpost&pid=151375\"][{POST_SNAPBACK}][/a]
Of course, because once the long dimension gets to be 24 inches wide, the PPI count sinks below 180. Bound to be sub-optimal I hazard to suggest, no matter how many pixels you pile on top of eachother - the linear resolution just gets too low for optimal quality.
-
..., I can walk and chew gum at the same time because I don't use my legs for chewing.
[a href=\"index.php?act=findpost&pid=148323\"][{POST_SNAPBACK}][/a]
Aren't your legs chewing away at the remaining distance to your destination?
Great thread from the tech-gods, ... thanks to all from someone down at the mortal level who is trying to learn!
-
Graeme,
A few things I don't understand:
(1) Why are you multiplying 10MP by 0.75^2 rather than simply 0.75?
(2) Also, why does a Foveon need 3*5.6mm photosites to get its resolution? Is there a separate photosite for each primary for each pixel, or is it three filters on top of the same photosite?
(3) When you knock 25% off the Bayer, you aren't really shrinking it are you, you are just discounting its resolution because of the "rounding the edges" impact of the AA? If so, where does this sharpening effect come from? I thought the AA blurs, and that is why you discount 25% of its native resolution.
Hence I don't really understand your conclusion.
Mark
[a href=\"index.php?act=findpost&pid=151320\"][{POST_SNAPBACK}][/a]
1) 0.75 is the linear factor. Because we're dealing with mp, which is a measure of area, we need to multiply by 0.75^2.
2) Each "pixel" on the Foveon has three sensors, one each for the three depths it senses, which is later converted into RGB colour.
3) You can shrink the image, if a good downsampling filter is used. I don't seem to think Photoshop or image editors normally use good downsampling filters though. In the video world I'd use an app like Shake which has user selectable downsampling filters.
Graeme
-
Optical low pass filters (OLPF's) or Anti Alias filter don't come in different "strengths". The filter works in two passes, each layer splits the light in two either horizontally or vertically, so by combing them together, you get vertical and horizontal filtering. The distance of seperation of the two rays of light is governed by the thickness of that layer, so if you want, (or need to as you don't have square photosites) you can adjust the filter accordingly. You choose the thickness of the filter in relation to the spacing of the photosites on the sensor.
How do you do this? (the writting in bold)
-
Hi!
I guess that "you" is "you camera maker" and not "you photographer". :-)
How do you do this? (the writting in bold)
[a href=\"index.php?act=findpost&pid=154301\"][{POST_SNAPBACK}][/a]
-
Optical low pass filters (OLPF's) or Anti Alias filter don't come in different "strengths". The filter works in two passes, each layer splits the light in two either horizontally or vertically, so by combing them together, you get vertical and horizontal filtering. The distance of seperation of the two rays of light is governed by the thickness of that layer, so if you want, (or need to as you don't have square photosites) you can adjust the filter accordingly. You choose the thickness of the filter in relation to the spacing of the photosites on the sensor.
In dpreview's LPPH results following the test charts page, there's usually a discrepancy between vertical lines per picture height and horizontal lines per picture height. One is often quite significantly less or more than the other.
It is explained that both horizontal and vertical height are the same dimension, ie. a square, in order to avoid confusion due to different aspect ratios amongst cameras.
If the width is the same as the height, one wonders why the resolution discrepancy.
Could the above quote from Graeme be the explanation?
-
Hi!
I guess that "you" is "you camera maker" and not "you photographer". :-)
[a href=\"index.php?act=findpost&pid=154304\"][{POST_SNAPBACK}][/a]
As I was thinking, I thought I missed out on something when ordering my 1Ds MKIII! Being able to choose the thickness on the AA filter would be kind of good though.
-
Does anyone know how strong the AA filter is on the 1Ds MKIII compared to, 1DsMKII and 5D?
-
Being able to choose the thickness on the AA filter would be kind of good though.
[a href=\"index.php?act=findpost&pid=154321\"][{POST_SNAPBACK}][/a]
I hope you were joking about this too? I mean, if that choice were offered would you know how, or put otherwise, what would it take for anyone to know how to do so intelligently?
-
My "you" was referring to the camera maker. And in that respect, it's the specification you send off to the filter manufacturer. Once it's made, the OLPF is fixed.
Some sensors have non-square pixels. In that case indeed you'd need different OLPF spacing in the different dimensions.
Resolution measurements are usually in lines per picture height, to account for the varying aspect ratios of sensors.
Graeme
-
Resolution measurements are usually in lines per picture height, to account for the varying aspect ratios of sensors.
[a href=\"index.php?act=findpost&pid=154380\"][{POST_SNAPBACK}][/a]
Graeme,
I understand there are two figures. One for vertical lines and the other for horizontal lines. The horizontal distance is the same as the vertical distance, to account for varying aspect ratios. In other words, a square is taken from the sensor with the dimensions of the picture height.
Why the sometimes significant difference in number of resolved lines for height and width?
-
Experimental error? Lens effects? OLPF tolerances? Who knows?
Graeme
-
Experimental error? Lens effects? OLPF tolerances? Who knows?
Graeme
[a href=\"index.php?act=findpost&pid=154396\"][{POST_SNAPBACK}][/a]
Okay! I thought you might .
-
The only camera I have control over, and have the engineering specs of, measures the same either direction, so I've not had the problem to look into it further.
Graeme
-
Regarding the small pixel pitch of the new camera I wonder what is the impact of lens diffraction on moire and artefacts. Theoreticaly could you avoid moire by stopping down to f11?
Hey, my first post here, I am allowed to ask stupid questions
-
Stopping down to deliberately get diffraction to limit the resolution hitting the sensor would have the same effect as an optical low pass filter. If you always shot stopped down, removing the OLPF would make sense as then you'd not get double the effect.
Graeme
-
Why don't you guys go out and take some pictures? I own a 1ds2 and a 40D, plus some 10K$ of L lenses... so far the only limitations I see are my own skills! the rest is bullshit!
saludos!
-
Why don't you guys go out and take some pictures? I own a 1ds2 and a 40D, plus some 10K$ of L lenses... so far the only limitations I see are my own skills! the rest is bullshit!
saludos!
[a href=\"index.php?act=findpost&pid=154543\"][{POST_SNAPBACK}][/a]
Seems rather a waste of money to buy a 1Ds2 and 40D if you are only limited by your own skills. Surely a P&S camera would be sufficient.
-
Oooooooooooooo - that was "the unkindest cut of all" - (but I just stopped laughing)
Chiang Mai must be doing good things for you.
-
Mark,
I'm glad someone appreciates my humour .
I'm pretty relaxed here in Chiang Mai, probably because I have no deadlines to meet, can take my own time and change plans at will now I've reached retirement age. I might even pay another visit to Angkor Wat in the next few weeks .
-
I'm pretty relaxed here in Chiang Mai, probably because I have no deadlines to meet, can take my own time and change plans at will now I've reached retirement age. I might even pay another visit to Angkor Wat in the next few weeks .
[a href=\"index.php?act=findpost&pid=154577\"][{POST_SNAPBACK}][/a]
Sounds idyllic. One of these days again...........
-
I hope you were joking about this too? I mean, if that choice were offered would you know how, or put otherwise, what would it take for anyone to know how to do so intelligently?
[a href=\"index.php?act=findpost&pid=154377\"][{POST_SNAPBACK}][/a]
I don't see the problem? If the subjects that you shoot are less prone to moire then it could be in your best intrests to choose a weaker AA filter( If you could ever actually do this) And most people who are buying a camera of this level will have some idea about what they do and don't like about thier pics especially in regards to moire.
Should I interpret your statement as you would not know how to choose between light and heavy filtering?
-
Stopping down to deliberately get diffraction to limit the resolution hitting the sensor would have the same effect as an optical low pass filter. If you always shot stopped down, removing the OLPF would make sense as then you'd not get double the effect.
[a href=\"index.php?act=findpost&pid=154533\"][{POST_SNAPBACK}][/a]
It's more like redundancy, after a certain point. You can't really soften what is already soft by any significant amount, especially when the cut-off is fairly sharp.
-
I don't see the problem? If the subjects that you shoot are less prone to moire then it could be in your best intrests to choose a weaker AA filter( If you could ever actually do this) And most people who are buying a camera of this level will have some idea about what they do and don't like about thier pics especially in regards to moire.
Should I interpret your statement as you would not know how to choose between light and heavy filtering?
[a href=\"index.php?act=findpost&pid=154593\"][{POST_SNAPBACK}][/a]
I think the problem here is a Catch 22 situation. If Canon were able to manufacture a device so sophisticated that you could dial in AA filter strength settings, they could probably gives us an oversampling, low noise 50mp sensor that would have no need for an AA filter. Maybe with future developments in Nanotechnology this will eventually be possible.
Here's a schematic diagram of an AA filter from the Canon website. As you can see it's a pretty sophisticated device with many sandwiched layers and also acts as a protective layer to the sensor.
[attachment=3939:attachment]
-
I think the problem here is a Catch 22 situation. If Canon were able to manufacture a device so sophisticated that you could dial in AA filter strength settings, they could probably gives us an oversampling, low noise 50mp sensor that would have no need for an AA filter. Maybe with future developments in Nanotechnology this will eventually be possible.
Here's a schematic diagram of an AA filter from the Canon website. As you can see it's a pretty sophisticated device with many sandwiched layers and also acts as a protective layer to the sensor.
[attachment=3939:attachment]
[a href=\"index.php?act=findpost&pid=154616\"][{POST_SNAPBACK}][/a]
I can understand that. I was thinking along the lines of having say two sensors to choose from before you buy, one being only mildly filtered and the other lets say the same as what comes with the camera standard these days.
-
Should I interpret your statement as you would not know how to choose between light and heavy filtering?
[a href=\"index.php?act=findpost&pid=154593\"][{POST_SNAPBACK}][/a]
Not quite. Perhaps I was a bit "telegraphic". To explain the point more fully, there is at least one camera maker which does offer a choice of with/without the filter. In that case they can tell you quickly and easily about the implications of the choice. If it were to be more refined than that - say three options: heavy filter, light filter, no filter, we are entering new territory and along with that users - even sophisticated ones - would need to be more carefully educated about the extent to which problems are likely to occur or not occur and what to do about them as a function of the filter choice they make. Then having made that choice, once they see the results they may or may not be happy about it, which adds a layer of issues to after-sales support for the camera manufacturer - either exchanging the camera or replacing the filter, and at whose cost? I know we're just talking hypothetical stuff here, but once such an idea is raised, which may indeed not be a bad idea, it is also worthwhile trying to conjure the implications at the same time. That can shed light on whether the idea is likely to see the light of day (pun intended). Hope that helps.
-
Not quite. Perhaps I was a bit "telegraphic". To explain the point more fully, there is at least one camera maker which does offer a choice of with/without the filter. In that case they can tell you quickly and easily about the implications of the choice. If it were to be more refined than that - say three options: heavy filter, light filter, no filter, we are entering new territory and along with that users - even sophisticated ones - would need to be more carefully educated about the extent to which problems are likely to occur or not occur and what to do about them as a function of the filter choice they make. Then having made that choice, once they see the results they may or may not be happy about it, which adds a layer of issues to after-sales support for the camera manufacturer - either exchanging the camera or replacing the filter, and at whose cost? I know we're just talking hypothetical stuff here, but once such an idea is raised, which may indeed not be a bad idea, it is also worthwhile trying to conjure the implications at the same time. That can shed light on whether the idea is likely to see the light of day (pun intended). Hope that helps.
[a href=\"index.php?act=findpost&pid=154662\"][{POST_SNAPBACK}][/a]
As I suspected, your statement was one of general speaking. And I do believe if there was ever something like this , the uniformed would probably be the first to critisise others over there own bad decision.
As there are no pro canon bodies without the filter, a little bit of trial and error would come into play I suspect.
Anyway, I know where you are comming from, I just thought I should respond to get clarification, thanks.
-
Here's a schematic diagram of an AA filter from the Canon website. As you can see it's a pretty sophisticated device with many sandwiched layers and also acts as a protective layer to the sensor.
[attachment=3939:attachment]
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=154616\")
[a href=\"http://www.sensor-film.com/filter.html]Here's[/url] a close-up photograph of the beasts for the 20D/30D and the 5D. The thickness of the 5D's OLPFs is greater than that of the 20/30D, in roughly the proportion of their pixel spacing, suggesting the goal of the designers was to keep the "strength" related to the need to shift 1 pixel in each axis. BTW , that's why I'm always interested in how folks assess the "strength" of the effect... Approximating a 1 pixel shift in software (in PS for example ) last couple of posts here (http://www.naturescapes.net/phpBB2/viewtopic.php?t=116427) produces a very benign softening
Andy
-
Mark,
I'm glad someone appreciates my humour .
I'm pretty relaxed here in Chiang Mai, probably because I have no deadlines to meet, can take my own time and change plans at will now I've reached retirement age. I might even pay another visit to Angkor Wat in the next few weeks .
[a href=\"index.php?act=findpost&pid=154577\"][{POST_SNAPBACK}][/a]
Ray
All funnies aside, don´t take ´retirement age´ with too much complacency: it creeps up to you and then slowly, without you noticing a damn thing, it has you THINKING retirement age; once there it is virtually impossible to regain the previous innocence. I know.
Rob C
-
Ray
once there it is virtually impossible to regain the previous innocence. I know.
Rob C
[a href=\"index.php?act=findpost&pid=154754\"][{POST_SNAPBACK}][/a]
What makes you think he was innocent before retirement?
-
What makes you think he was innocent before retirement?
[a href=\"index.php?act=findpost&pid=154765\"][{POST_SNAPBACK}][/a]
Mark,
I was born innocent. It seems that some people are direct descendants of Adam and Eve and are born with lots of original sin, and some are direct descendants of ape-like creatures who lived in Africa about 5 million years ago and are born innocent.
Phew! Talk about getting off topic .
-
The thickness of the 5D's OLPFs is greater than that of the 20/30D, in roughly the proportion of their pixel spacing, suggesting the goal of the designers was to keep the "strength" related to the need to shift 1 pixel in each axis. BTW , that's why I'm always interested in how folks assess the "strength" of the effect... Approximating a 1 pixel shift in software (in PS for example ) last couple of posts here (http://www.naturescapes.net/phpBB2/viewtopic.php?t=116427) produces a very benign softening
Andy
[a href=\"index.php?act=findpost&pid=154741\"][{POST_SNAPBACK}][/a]
Interesting! I didn't realise there was another forum discussing the same issue more or less contemporaneously. I guess I spend too much time on LL and have no time to look at other forums. I haven't visited Photo.net in a year or so and my subscription has probably expired.
I noticed one comment from 'Jeff' that seemed very counter-intuitive, but I didn't see anyone correcting him on the NatureScapes forum.
The problem is that no AA or a weak AA filter causes some pretty serious moire, particularly as pixel size decreases. For this reason, the AA filters tend to get somewhat stronger as pixel size decreases, creating a bit of diminishing returns at least with regard to detail at very small spatial frequencies.
I take it he's got this backwards?
-
What makes you think he was innocent before retirement?
[a href=\"index.php?act=findpost&pid=154765\"][{POST_SNAPBACK}][/a]
On the assumption that you and Ray are being funny - not too easy to tell without facial expessions to modify message - I am being serious here. There is a stage in life when everything feels possible and then, for one reason or another, that starts to slide away from you and is replaced with its opposite number. Once that bugger is there, he won´t let go; he does seem to come to visit when time becomes too available. Just a little message to warn about him was all...
Of topic yes, but common to all of us sooner or later.
Rob C
-
Why don't you guys go out and take some pictures? I own a 1ds2 and a 40D, plus some 10K$ of L lenses... so far the only limitations I see are my own skills! the rest is bullshit!
saludos!
[a href=\"index.php?act=findpost&pid=154543\"][{POST_SNAPBACK}][/a]
Naturally, you are being hit with whatever comes to hand - to mind - but I´m afraid that the actual thinking behind what you wrote is pretty accurate.
There is no doubt whatsoever that the advent of digital photography has brought to life a very exaggerated version of the film and developer syndrome which left me bemused for much of my lfe. That´s not to say that people who have this thing were not also capable photographers at times, and I have no reason to doubt that some of our resident pixel peepers are also as hot as hell with the production of really deep, insightful and incisive images. No, the problem for me is that it all seems so goddamn pointless: you get what the makers sell you and I do not believe they give a hoot about what you think. As an example, take our host´s battle to get a direct MU device from Canon: no greater fan can be found, I suspect, on the web, but have they paid the slightest attention to him?
As I have said so often, to the point of boring you even, were photography back in the 50s as it is now, I would never have dreamed of spending my life working in it. Photography was direct -a rapid, simple means of getting to a state where a visual idea was down on paper, for better or for worse, your soul was on show. I liked that sense of directness, the fact that once you had learned the simple techniques of maintaining temperatures constant and solution strengths up to snuff, that was it: the rest was in your eye, in your mind and in your natural sense of what looked right. Simple, quick and personal
I have played about with digital for some time now, the main reason being that after a heart attack I discovered that trying to empty large trays of fluid back into bottles through a funnel was a little beyond my strength to manage safely; equally, living in a country with endemic water problems, it didn´t seem clever to wash prints for an hour and RC papers (plastics?), though les wasteful, never pleased me that much.
So, yes, I had to learn to print via a computer. And it has not led to all that much glee. Ironically, the buzz now comes when I get something which resembles a good black and white, something which was a given for over fifty bloody years!
In that sense of the thing, Jorge, you are absolutely on the money. The new photography is not about image; the new photography is about computer technique and toys.
It is strange how so much has improved without anything getting any better.
(NOT an attack on anybody, so don´t take it as such - simply my feelings on the photo-world of now.)
Hasta la vista, hombre - Rob C
-
Naturally, you are being hit with whatever comes to hand - to mind - but I´m afraid that the actual thinking behind what you wrote is pretty accurate.........
.....No, the problem for me is that it all seems so goddamn pointless: you get what the makers sell you and I do not believe they give a hoot about what you think. As an example, take our host´s battle to get a direct MU device from Canon: no greater fan can be found, I suspect, on the web, but have they paid the slightest attention to him?......
.......It is strange how so much has improved without anything getting any better.
[a href=\"index.php?act=findpost&pid=154909\"][{POST_SNAPBACK}][/a]
Well, Rob, I guess we could explain that attitude very briefly by quoting from your previous post.
There is a stage in life when everything feels possible and then, for one reason or another, that starts to slide away from you and is replaced with its opposite number. Once that bugger is there, he won´t let go; he does seem to come to visit when time becomes too available. Just a little message to warn about him was all...
Some people get what the makers sell them, others get what they choose to buy, after careful consideration and analysis of the features on offer.
The reason why Canon have not included a dedicated MLU button on their latest models is probably precisely because they do listen to the consumer. I don't think that Michael R is a typical Canon DSLR consumer. The average buyer of Canon DSLRs would probably consider a dedicated MLU button a waste of real estate. Most strangers I meet on my travels who happen to be using a Canon or Nikon or Fuji DSLR and whom I get talking to, don't even shoot RAW.
MLU is for those who are concerned about getting the maximum image quality their system can deliver. It's for people who are likely to be also concerned about the deleterious effects of an AA filter.
If you look at Canon's latest model, the 40D, you'll see a number of new features there that seem to appeal to popular demand, such as anti-dust sensor, a live-view LCD screen, a 14 bit A/D converter plus an extra 2 megapixels to at least give the psychological comfort of expected image quality improvement.
Perhaps most revealing of all, which really shows that Canon is listening to the problems of the consumer, is a system for reducing the chance of blown highlights.
Most people who make the transition from film to digital, as well as those who are buying a digital camera for the first time, will often accidentally blow highlights because of the lack of that very broad shoulder that film has. I bet it's been the subject of thousands of threads on photographic forums. "Why can't Canon implement an auto-exposure system that recognises there will be blown highlights and take automatic action to reduce the exposure.?"
Well, Canon have listened and they've introduced such a system in the 40D. It's not perfect and not quite what was asked for. As I understand, there's no analysis of the scene to determine if highlights are going to be blown as a result of the user's choice of settings. (Someone correct me if I'm wrong on this point.) There's a simple automatic reduction of ISO setting by one stop, followed by a brightening of the review which gives the impression the image has been correctly exposed. It might indeed have been correctly exposed by this reduction in ISO, thus protecting the consumer from her own incompetence. On the other hand it might have been drastically underexposed because the user's setting would already have produced an underexposed image or at best a proper exposure to the right.
Got to rush off now and take some photos. There some sort of parade passing by; part of the Loi Krathong festivities.
See you later, or as they say in Thai, Pop Gan Mai!
-
The new photography is not about image; the new photography is about computer technique and toys.
It is strange how so much has improved without anything getting any better.
[a href=\"index.php?act=findpost&pid=154909\"][{POST_SNAPBACK}][/a]
Rob, No. In a fundamental sense there is no new photography. It was, is and will be about making good phtoographs, however defined. Without that the rest is useless. What is new is the technology. And this technology has opened a plethora of creative potential that never existed before or only existed with great difficulty and far less predictable outcomes. And it has reached the stage where the quality of B&W images is at least equal to and arguably exceeds the B&W of the film era. As for colour, there is simply no contest.
-
The reason why Canon have not included a dedicated MLU button on their latest models is probably precisely because they do listen to the consumer. I don't think that Michael R is a typical Canon DSLR consumer. The average buyer of Canon DSLRs would probably consider a dedicated MLU button a waste of real estate. Most strangers I meet on my travels who happen to be using a Canon or Nikon or Fuji DSLR and whom I get talking to, don't even shoot RAW.
MLU is for those who are concerned about getting the maximum image quality their system can deliver. It's for people who are likely to be also concerned about the deleterious effects of an AA filter.
If you look at Canon's latest model, the 40D, you'll see a number of new features there that seem to appeal to popular demand, such as anti-dust sensor, a live-view LCD screen, a 14 bit A/D converter plus an extra 2 megapixels to at least give the psychological comfort of expected image quality improvement.
[a href=\"index.php?act=findpost&pid=154929\"][{POST_SNAPBACK}][/a]
Ray, this is disturbing. It would never have occurred to me that anyone investing over 8000 dollars in a top-flight DSLR system would be using it to shoot JPEGs exclusively and would NOT be focused on getting the best image quality they can. Gosh those folks should donate their wasted money to charities and buy a good quality P&S for a fraction of the price.
I think the anti-dust feature is a good idea for all of us, and the live-view screen is also great for helping to anticipate and prevent blown highlights. I agree that the extra 2 MP would go largely unnoticed in a print, and as for the 12 vs 14 bit business - the effect of that is less obvious - it will need some pretty careful analytical testing to say anything informed about it.
-
Mark,
I was born innocent. It seems that some people are direct descendants of Adam and Eve and are born with lots of original sin, and some are direct descendants of ape-like creatures who lived in Africa about 5 million years ago and are born innocent.
Phew! Talk about getting off topic .
[a href=\"index.php?act=findpost&pid=154855\"][{POST_SNAPBACK}][/a]
LOL! Love it!
Proud to be among the second group.
Perfect accompaniment to the Darwin-Fish.
-
On the assumption that you and Ray are being funny - not too easy to tell without facial expessions to modify message - I am being serious here. There is a stage in life when everything feels possible and then, for one reason or another, that starts to slide away from you and is replaced with its opposite number. Once that bugger is there, he won´t let go; he does seem to come to visit when time becomes too available. Just a little message to warn about him was all...
Of topic yes, but common to all of us sooner or later.
Rob C
[a href=\"index.php?act=findpost&pid=154904\"][{POST_SNAPBACK}][/a]
Is any assuming needed,...what with the facial expressions of smiley-faces and ape-like creatures all over the place?
As a retiree who appreciates that time is the greatest luxury, I feel that the early "anything is possible" innocence has at last been freed of the impositions, obligations, and dictates of pre-retirement life, and can at last go out to play in this world of wonders. (Not to mention that a pocket-more-full of accumulated assets adds to that freedom.)
True, the "Life is too short" demon comes to sit on a shoulder now and then, but anyone immersed in living, rather than contemplating dying, pays him little attention.
When retirees complain of "nothing-to-do", purposelessness, etc., I can't help but think that they should have gotten-a-life (beyond their job) along the way, which they could now be enjoying immensely. This view prompted by the reality of a fellow retired firefighter, who came back from retirement to ride a medic-unit (an often less-than-pleasant job) not out of any unquenchable desire to "help", but out of boredom and habit. Contrasted with my own list of things to hopefully do, this firmly establishes that yes,...people are different indeed.
-
Ray, this is disturbing. It would never have occurred to me that anyone investing over 8000 dollars in a top-flight DSLR system would be using it to shoot JPEGs exclusively and would NOT be focused on getting the best image quality they can.
[a href=\"index.php?act=findpost&pid=154970\"][{POST_SNAPBACK}][/a]
Absolutely agree!
If those buying expensive, state of the art "Pro" equipment are not those seeking "best image quality", ...then who is?
I'm with Michael on this (and appreciate his continued "prodding" of the manufacturers).
Why have we "progressed" in so many ways, ...but left 40+ years in the past a feature that is undeniably useful in some situations (which is all that can be said of a great many "features")?
All I want is the simple complete MLU control I had in the early 70's on my Minolta SRT's.
C'mon Canon!
-
Ray, this is disturbing. It would never have occurred to me that anyone investing over 8000 dollars in a top-flight DSLR system would be using it to shoot JPEGs exclusively and would NOT be focused on getting the best image quality they can. Gosh those folks should donate their wasted money to charities and buy a good quality P&S for a fraction of the price.
[a href=\"index.php?act=findpost&pid=154970\"][{POST_SNAPBACK}][/a]
Mark,
Why are you surprised. We had a review recently on LL of the Nikon D3 from James Russell who seems to have used the camera in jpeg mode for most of his shots. I bet there'll be lots of owners of that camera who will use jpeg most of the time as well as auto-exposure and the various programmed picture modes.
Isn't the trend towards making cameras that do all the thinking and automate everything that can possibly be automated? We've even got a feature in the latest Canon models that can recognise faces, auto-face detection .
-
Isn't the trend towards making cameras that do all the thinking and automate everything that can possibly be automated?
[a href=\"index.php?act=findpost&pid=155023\"][{POST_SNAPBACK}][/a]
In a way - yes - and there are clearly situations where this really helps. But there remains - or one would hope - that steady clientele for excellence, which most of the time requires custom control image by image - and raw files.
-
Mark,
Why are you surprised. We had a review recently on LL of the Nikon D3 from James Russell who seems to have used the camera in jpeg mode for most of his shots. I bet there'll be lots of owners of that camera who will use jpeg most of the time as well as auto-exposure and the various programmed picture modes.
Isn't the trend towards making cameras that do all the thinking and automate everything that can possibly be automated? We've even got a feature in the latest Canon models that can recognise faces, auto-face detection .
[a href=\"index.php?act=findpost&pid=155023\"][{POST_SNAPBACK}][/a]
I believe my camera has more modes than manual and aperture priority ae:) I do like autofocus though
Mike
-
In a way - yes - and there are clearly situations where this really helps. But there remains - or one would hope - that steady clientele for excellence, which most of the time requires custom control image by image - and raw files.
[a href=\"index.php?act=findpost&pid=155027\"][{POST_SNAPBACK}][/a]
Mark & Wolfy,
As readers and contributers to the LL forum we probably get a skewed impression of the (camera buying) general public's concern for maximum image quality.
Often only the top of the range 35mm film cameras used to boast an MLU feature, although with some models you could get around that by using the delayed action timer.
In a sense we're into the same territory as the suggestion for an adjustable AA filter, or the option of a choice of camera with AA filter, with weak AA filter or no AA filter. It requires a certain level of expertise to even know which you might prefer.
The benefits of MLU are equally nebulous. I use it with slow shutter speeds on tripod just to be sure. But when I've taken the trouble to actually test the benefits of MLU the results are very variable and the benefits seem to depend very much on other factors such as stability and design of tripod, total mass of camera and lens, presence or lack of the slightest breeze if outdoors, and of course the actual shutter speed being used.
And I forgot perhaps the most important factor of all; proper dampening of mirror slap. If Canon were to attach more importance to the MLU feature, in effect raising the status of the feature, it would be tantamount to admitting they had not done a good job in dampening mirror slap. This not going to improve sales.
-
Ray
I don´t think that it needs Canon to admit to any sort of failing if they provide a MU facility; I think it would simply be another gadget for those who know no better and a Godsend for those who do. Win - win?
Wolfy
Retirement equating with time to do something you kept on the backburner rings a little hollow when you are a pro photographer - at least, one of those who didn´t just do it because he failed at everything else! It´s been my life, still is my only love outside family, but lots of other external factors kick in, some which never enter your head and are sometimes not even of your own direct making. Let´s hope you never encounter them.
Rob C
-
Ray
I don´t think that it needs Canon to admit to any sort of failing if they provide a MU facility;[a href=\"index.php?act=findpost&pid=155179\"][{POST_SNAPBACK}][/a]
Rob,
I think they've always provided an MU facility except on their cheapest cameras, and even there indirectly with the auto timer. That's not in question. What some people are asking is that it be more instantaneously available, either have its own dedicated button with flashing lights or a more readily accessible main heading on the LCD menu.
-
Rob,
I think they've always provided an MU facility except on their cheapest cameras, and even there indirectly with the auto timer. That's not in question. What some people are asking is that it be more instantaneously available, either have its own dedicated button with flashing lights or a more readily accessible main heading on the LCD menu.
[a href=\"index.php?act=findpost&pid=155190\"][{POST_SNAPBACK}][/a]
Sorry, Ray, I already knew that there was a route to it, but I was referring to a straight, one button device, which is what Michael has been campaigning for - shall try to keep the old mind more focussed on what it authorises the fingers to type...
Rob C
-
Rob,
I think they've always provided an MU facility except on their cheapest cameras, and even there indirectly with the auto timer. That's not in question. What some people are asking is that it be more instantaneously available, either have its own dedicated button with flashing lights or a more readily accessible main heading on the LCD menu.
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=155190\")
Guys,
Because we have drifted considerably from the OT, ands because a wider readership may have interest in the MLU issue, I am posting this in the digital camera forum as well.
[a href=\"http://luminous-landscape.com/forum/index.php?showtopic=21124]http://luminous-landscape.com/forum/index....showtopic=21124[/url]
--------------------------------------------------------------------------------------------------------------
NOTE TO MODERATORS - If it is decided that the post duplication is unacceptable, please leave the link in the above paragraph here, and give preference to the post in the other forum.
--------------------------------------------------------------------------------------------------------------
Those with little use for MLU (or who BELIEVE that there is little use for it,) might consider these points:
1. A programmed camera is not always an effective substitute for a thinking photographer. This means that when a "self-timer" is controlling the moment of capture (2 sec. delay?, 10 sec. delay? 30 sec. shut-down?), any sort of peak-moment capture is reduced to a wild guess. Hardly a "professional" feature. IMO, this is no "work-around. It is a poor imitation of user-control.
[ Here are four (A-D)of many possible scenarios;
A. You are using a tripod mounted long tele with remote release pre-focused on a distant offshore rock upon which waves break spectacularly. You wish the rocks(mermaid, whatever,) to be as sharp as possible, with whatever softness to the water you may create by your choice of shutter speed. You wish to capture the moment of peak splash height, like ...,...,...NOW! Oops, ...the self timer missed it by a mile (maybe because it "timed-out", un-noticed after some designer-decided 30 sec. limit).
B. Same set-up, different target. The osprey/eagle/pterodactyl is about to land on the edge of the nest, fish-in-beak, wings spread and claws out-stretched for touchdown, matching the outstretched necks of the fledglings straining to receive dinner. Another "NOW!" moment. Is the blind self-timer going to guess it correctly? My Donkey!]
C. Target - birds at feeder. Anyone know how quickly flitting birds can move from the "perfect pose" shot to a wasted one? The camera doesn't, that's for sure.
D. Two wild mountain rams at the moment of butting impact, dust flying from foreheads, some hooves off ground? A distant, very "Now!" moment. Perhaps capturable by YOU (especially with 8-10 FPS), ...but not with programmed interference from a self-timer.
Control means control!
2. Opinions differ as to MLU usefulness at any given shutter speed. However, anytime there is a POTENTIAL benefit from MLU, there is no arguing this fact:
Reduced or "damped" mirror vibration is not the same thing as ELIMINATED mirror vibration. Canon's providing real MLU would "admit" nothing except this incontrovertible fact. [ As to the nebulosity of MLU usefulness -- For an in-depth study of MLU effectiveness under various conditions, timings, focal-lengths, etc. done by a serious, careful and skilled photographer, see first link below.]
3. Real user-controlled MLU was not limited to only the "top" models, at least in the case of Minolta in the early 70's (late 60's?). The SRT line was below their briefly offered "pro" model (can't remember the model designation).
There have been numerous studies of MLU usefulness. Here is a link to what I consider one of the most thorough (The entire site is well worth visiting, ...and there is some intriguing Nikon vs. Canon opinion from this experienced pro.)
Fritz Polking index (See Workshop I - "Sharp Photographs"): http://www.poelking.com/index_e.htm (http://www.poelking.com/index_e.htm)
Sadly, this renowned photographer/author is now deceased. Notice is here:
Polking Passing (July 23, '07)-
http://www.digitalphotopro.com/news/master...lking-dies.html (http://www.digitalphotopro.com/news/master-nature-photographer-fritz-polking-dies.html)
Bottom line for me;
Give us back real user controlled (and simple) MLU. "Flip",...it's up. "Flip" it's down. I'll decide when.
Canon, this user will trade you one facial-recognition "feature" for one real MLU anyday!
Opinion. Correction welcome as always,
-
A. You are using a tripod mounted long tele with remote release pre-focused on a distant offshore rock upon which waves break spectacularly. You wish the rocks(mermaid, whatever,) to be as sharp as possible, with whatever softness to the water you may create by your choice of shutter speed. You wish to capture the moment of peak splash height, like ...,...,...NOW! Oops, ...the self timer missed it by a mile (maybe because it "timed-out", un-noticed after some designer-decided 30 sec. limit).
[a href=\"index.php?act=findpost&pid=155220\"][{POST_SNAPBACK}][/a]
Sorry Wolfy, that just won't wash. Mermaids never keep still. They're always wiggling their tail and preening themselves. It would not be a good idea to use a slow shutter speed on them.
I've got no argument against the usefulness of MLU. I always use it with shutter speeds of less than 1/60th on tripod, just to be sure to be sure.
I also agree completely that an auto timer has serious disadvantages for MLU purposes, but it's better than nothing. All Canon's professional and prosumer cameras have full MLU. The 300D didn't and probably the 400D doesn't. I haven't checked.
I personally would prefer to see the MLU option say under the auto-bracketing option on the first page of the menu rather than custom function #12. But it's no big deal for me, maybe because I don't use a tripod often. I'm always excited by new ways I can avoid using a tripod, such as CS3's excellent auto-alignment feature which opens up the opportunity for hand-held bracketing.
I have to assume that Canon does its own market research and I can only make a wild guess at the percentage of DSLR owners who use MLU, but it could be something like 1% on a regular basis; 5% occaisionally and 94% never.
I don't expect the world to organise itself around my wishes, you know.
ps. copied to other thread
-
The preening I can do without,...but the tail wiggling sounds good. I like it on terramaids!
Please describe "full" MLU with respect to what I called "real" MLU (Up when I say, down when I say, shoot when I say).
[I do not have a Canon "pro" camera. Was set to jump on the 1DIII, when all the AF discussion broke out,...still waiting for the dust to settle.]
If "full" MLU is the equivalent of old-time user-controlled MLU, for shots such as the examples I gave, ...then what is Michael always going-on about? (Maybe the other forum is the place to respond?)
Thanks,
[a href=\"index.php?act=findpost&pid=155239\"][{POST_SNAPBACK}][/a]
Full MLU, I suppose is what I've got on my D60, 20D and 5D. Having set the custom function to 'enable' MLU, one presses the shutter button once to flip up and lock the mirror; one waits until the vibrating tripod legs have stabilised and then presses the shutter a second time which begins the exposure. When the exposure has finished, the mirror flips back down automatically and one is ready for the next shot.
-
But none of this is a simple, honest, press-me-now-to-engage feature; there is something far wrong if you consider that messing about in menus is as efficient or as user-friendly as simply sliding a single catch or pressing a buton!
Okay, you want to ignore this, as is your right, but to so do does not make it so, if you see what I mean.
Also, I have to disagree with your notion of mermaid behaviour, Ray. I have shot one in a studio (she once became earth-bound as a Miss Scotland) but as the rock was wood and papier maché (sorry about circumflex - can´t make it work) she didn´t move at all on the basis of it being too much of a hazard. Now there´s a thought: a papier maché world to keep everything still long enough for menu-linked options to be kicked in!
Terramaids: now that´s an interesting take on birds!
Rob C