Luminous Landscape Forum
Equipment & Techniques => Cameras, Lenses and Shooting gear => Topic started by: bluekorn on August 14, 2008, 09:19:55 pm
-
Hello All,
Ken Rockwell today declared the sharpness of the D300, the D700 and the D3 to be equal and he has always raved about the color rendition of all Nikons. Would anyone be willing to reflect for a moment, for the benefit of those of us amongst the laity, what factors besides sharpness (and excluding lenses and post processing), are to be considered in choosing between DX and FX regarding IQ? If DX and FX are equally sharp is there a gain to be had in IQ by spending for the FX? Thank you.
Peter Van Dyken
-
Uh oh, here is comes... you said the forbidden name...
-
what factors besides sharpness (and excluding lenses and post processing), are to be considered in choosing between DX and FX regarding IQ? If DX and FX are equally sharp is there a gain to be had in IQ by spending for the FX?
I don't understand the meaning of "DX and FX are equally sharp". The "sharpness" of a sensor depends on the AA filter and the sensel size, not on the size of the sensor.
However, an important consideration is, how good a given lens is with the sensor; this does depend on the sensor size. Many lenses excel with the cropping sensors, while they are soft with full frame; this is only natural. For example for me (dedicated to panoramas) the cropping cameras are the uncontested winners because of the edge sharpness.
Re the IQ: there can be no serious discussion about the superiority of the "pixel quality" of the D3 (= D700) compared to the D300, due to the size of the sensels. Nikon fanboys are in for a rude awakening with the high-pixelcount successor of the D3.
-
Hello All,
Ken Rockwell today declared the sharpness of the D300, the D700 and the D3 to be equal and he has always raved about the color rendition of all Nikons. Would anyone be willing to reflect for a moment, for the benefit of those of us amongst the laity, what factors besides sharpness (and excluding lenses and post processing), are to be considered in choosing between DX and FX regarding IQ? If DX and FX are equally sharp is there a gain to be had in IQ by spending for the FX? Thank you.
Peter Van Dyken
[a href=\"index.php?act=findpost&pid=215092\"][{POST_SNAPBACK}][/a]
Nice use of "laity". I've only heard used referring to parishioners, but you're correct.
Also, you've been warned now about *en %ockwell. $en @ockwell could post a real cure for cancer and people would still bash you if you praised him.
Finally, here's a short answer. And let's assume we're talking about the same number of pixels in either a larger or smaller space, with larger or smaller phostosites, respectively.
Smaller pixels may not be as sensitive to light as larger pixels. With larger photosites, because they are larger they catch more light more quickly. Think of putting a small bowl and a large bowl out in the rain. The large one collects more falling water more quickly, and can hold more water before it fills up.
The smaller the pixel the smaller number of photons it can hold before it fills up and spills over. That capacity is known as the photosite's "electron well." With the larger pixels/photosites that can hold more electrons you are going to have more dynamic range.
Moreover, given a certain amount of noise in a system, the larger pixels are going to suffer less. Let's say a small photosite holds 40,000 electrons, and a large photosite holds 100,000 electrons. Now this is an exaggerated number, but for the sake of making the point let's say you're going to get 1,000 electrons of noise per photosite. That means 2.5% of the data read off that site is going to be noise. With the larger photosite only 1% of the data is going to be noise. A much better ratio.
Now you know more than 99% of posters in camera forums. Once you can explain why diffraction limits the resolution of Nikon's current line of lenses and probably limits the number of photosites on a sensor in an FX body, you get bonus points.
-
Re the IQ: there can be no serious discussion about the superiority of the "pixel quality" of the D3 (= D700) compared to the D300, due to the size of the sensels. Nikon fanboys are in for a rude awakening with the high-pixelcount successor of the D3.
[a href=\"index.php?act=findpost&pid=215112\"][{POST_SNAPBACK}][/a]
Since a high MP FX DSLR is a primarily a landscape/studio camera and will mostly be used at base ISO, I have to wonder why I will be in for a rude awakening (this in spite of my taking issue with anyone calling me a "fanboy"). D300 pixels at base ISO or even pushed two or three stops are very good, and if I have 24 million of them in an FX sensor then I will be very happy as long as I have a couple of tilt/shift lenses to attain adequate DOF at f/11.
-
Since a high MP FX DSLR is a primarily a landscape/studio camera and will mostly be used at base ISO, I have to wonder why I will be in for a rude awakening (this in spite of my taking issue with anyone calling me a "fanboy"). D300 pixels at base ISO or even pushed two or three stops are very good, and if I have 24 million of them in an FX sensor then I will be very happy as long as I have a couple of tilt/shift lenses to attain adequate DOF at f/11.
[a href=\"index.php?act=findpost&pid=215132\"][{POST_SNAPBACK}][/a]
I too will be buying the D3x, or whatever they call it, along with the 24PC-E and probably the 85PC-E for, you guessed it, landscape and product stills.
-
Since a high MP FX DSLR is a primarily a landscape/studio camera and will mostly be used at base ISO, I have to wonder why I will be in for a rude awakening
These days the D3 is THE high-ISO camera. It is praised (rightly) for the high dynamic range (coming for the excellent pixel quality).
My comment was aimed at those, who expect a camera with twice the pixel count and the same image quality. There are quite a few of them.
this in spite of my taking issue with anyone calling me a "fanboy"
If you aren't one, then why do you think I called *you* a fanboy? (The fanboys' meeting place is DPReview.)
D300 pixels at base ISO...
I referred to the D3.
-
Hello All,
Ken Rockwell today declared the sharpness of the D300, the D700 and the D3 to be equal and he has always raved about the color rendition of all Nikons. Would anyone be willing to reflect for a moment, for the benefit of those of us amongst the laity, what factors besides sharpness (and excluding lenses and post processing), are to be considered in choosing between DX and FX regarding IQ? If DX and FX are equally sharp is there a gain to be had in IQ by spending for the FX? Thank you.
Peter Van Dyken
[a href=\"index.php?act=findpost&pid=215092\"][{POST_SNAPBACK}][/a]
Not so long ago Ken was going on about how 'full-frame' was inherently better than 'crop-frame' in terms of IQ. I guess his opinions change according to his mood. To be honest I simply do not trust his ability to put one foot in front of the other never mind perform a carefully thought out comparison between two cameras.
-
what factors besides sharpness (and excluding lenses and post processing), are to be considered in choosing between DX and FX regarding IQ?
High ISO capability, lower noise at all ISOs, and dynamic range.
Cartman has explained why the larger the pixel the better these things generally are. At the same megapixel count an FX camera will be more likely to be better in these areas than a DX camera, simply because each of its pixels is bigger. This is 'way over-generalized, but it serves as a starting point.
But the ball is not completely in the FX court. Panopeeper has mentioned one advantage that favours DX over FX. Another, at least for landscape, is the larger DOF implicit in using a shorter focal length for any given field of view. The larger DOF means more of the image will be resolved.
*
BTW: if some of the replies you've gotten and will continue to get are puzzling, it's because you've walked into a mine field simply by mentioning the name Ken Rockwell. Many photographic and imaging professionals frequent this forum. I'd venture that perhaps the only thing they all agree on is that Ken presents sensationalist disinformation in order to increase traffic to his web site for financial gain. IOW: it is clear to them that he has adapted to the web the tabloid trick of fabricating sensationalistic fiction and presenting it as fact in order to sell copy. If I were to stoop to the level of ad hominem that the anti-KR contingent unfortunately wallows in, I'd point out that I have it on excellent authority that Mr Rockwell is P. T. Barnum's great godson, but I strongly disapprove of ad hominems, so I won't mention it.
-
If you aren't one, then why do you think I called *you* a fanboy? (The fanboys' meeting place is DPReview.)
[a href=\"index.php?act=findpost&pid=215146\"][{POST_SNAPBACK}][/a]
Merely replying to your post implied that I was a "fanboy", which is akin to putting me in the awkward position of responding to the proverbial paradox "When did you stop beating your wife?"
-
Moreover, given a certain amount of noise in a system, the larger pixels are going to suffer less ... for the sake of making the point let's say you're going to get 1,000 electrons of noise per photosite. [a href=\"index.php?act=findpost&pid=215124\"][{POST_SNAPBACK}][/a]
Why would we assume that a larger device will produce an equal number of electrons of noise? It seems reasonable to expect instead that a increased photosite bulk produces an increase in noise sources. Also, larger well capacities require larger capacities for charge, voltage, or current all along the analog signal transport and processing chain, which seems to have the potential to increase the number of electrons of noise from sources like amplifiers (amp. noise is apparently the dominant noise source in the high end Canon's at low ISO).
This expectation is supported by data for Kodak CCDs, where total dark noise levels in electrons tend to increase with photosite size.
By the way, 1000 electrons is an absurdly high number, suggesting that you are not even vaguely aware of the real values, which run roughly from 3 to 30 electrons in spec. sheets and published measurements for SLR sensors.
-
The smaller the pixel the smaller number of photons it can hold before it fills up and spills over. That capacity is known as the photosite's "electron well." With the larger pixels/photosites that can hold more electrons you are going to have more dynamic range.
Would it then be correct to say that, (all other factors being equal):
The larger sensor with larger pixels, will provide wider dynamic range and tonality in a final print, independent of the size of the print?
-
Fanboy?
Should that not be Fan Person?
-
Merely replying to your post implied that I was a "fanboy", which is akin to putting me in the awkward position of responding to the proverbial paradox "When did you stop beating your wife?"
Well, I did not infer any such connection. In fact, I don't see many fanboys on LL. My comment was not aimed at any person here but at the worshipping attitude observable on DPRand and other, Nikon-specific forums. I saw repeatedly the assumption, that a D3 successor with twice the pixels will behave like the D3, not like the D300 in regard pixel quality, because Nikon has developed some "breakthrough" technology.
-
Fanboy?
Should that not be Fan Person?
Would not the correct form be Fanperson (like Chairperson)? Though if we start out with fanboy, then there should be fangirl, and the neutered/spayed variant should be fankid :-)
-
Since a high MP FX DSLR is a primarily a landscape/studio camera and will mostly be used at base ISO, I have to wonder why I will be in for a rude awakening (this in spite of my taking issue with anyone calling me a "fanboy"). D300 pixels at base ISO or even pushed two or three stops are very good, and if I have 24 million of them in an FX sensor then I will be very happy as long as I have a couple of tilt/shift lenses to attain adequate DOF at f/11.
[a href=\"index.php?act=findpost&pid=215132\"][{POST_SNAPBACK}][/a]
Quite true. However, users of the D3 or D300 don't necessarily have to go out and buy the D3x for landscapes since they can stitch together several D3 or D300 images and obtain resolution superior to the D3x with only one exposure.
Bill
-
Would not the correct form be Fanperson (like Chairperson)?
Quite right. Good point.
Though we could also really simplify.
Singular: Fan
Plural: Fannies
Awhile back when reasearching a camera purchase I stumbled upon DPR, and I am afraid I have developed somewhat of an addiction to it. Want to kick it really do. So many incredibly impolite people. But its kind of the gawking at an accident thing.
-
Alas, I stepped in something visiting the circus and now I've besmirched the landscape. I appreciate the heads up. I have much to learn in the field of digital psychology.
I was puzzled by my own naive assumptions gleened from the never to be mentioned again ringmaster regarding "equal sharpness" between the D700 and the D300. His declaration muddied my waters. Intuitively, I was assuming that full frame trumped cropped frame. I'm learning that application is part of the equation. Bottom line. I don't want to spend for bells and whistles yet I'm quite eager to spend for the current best combination of camera and lens (I realize there might be many) as a foundation for further development as a landscape photographer. Frames per second and high ISO shooting are not necessarily of concern. I want eventually to be able to produce seemingly crystal clear, softly luminescent prints that mimic the photos in the old Leica brochures. Good luck, you say? It's simply a distant goal that runs along the lines of personal taste. I'm not devoted to any camera system.
Peter
-
I don't see any reason why a higher pixel count D3 successor, given the few months more technological improvement, should not have at least equally low noise at high ISO.
There seems to be a confusion here between total image noise and individual pixel noise.
Canon's first full frame DSLR, the 11mp 1Ds was compared with the 6mp D60 at DPR. Despite its larger photosites, the 1Ds actually displayed slightly more noise than the D60, pixel for pixel (up to ISO 400). However, 1Ds images at all ISO settings, showed less noise than D60 images, comparing same size prints or same size images on the monitor.
-
...users of the D3 or D300 don't necessarily have to go out and buy the D3x for landscapes since they can stitch together several D3 or D300 images and obtain resolution superior to the D3x with only one exposure.
[a href=\"index.php?act=findpost&pid=215265\"][{POST_SNAPBACK}][/a]
Right, but then the D3x (or whatever this hypothetical high MP DSLR ends up being called) user can stitch too. My problem is that when I stitch I am often shooting dynamic scenes where either subject motion or changing light makes multiple exposures problematic; so I want to get enough detail in just two or three shots (or sometimes in just one) instead of half a dozen or more shots.
-
Hi,
I guess that FX and DX makes different demands on lenses. Some lenses may shine on FX an others on DX. DX has higher pixel density but uses a smaller area of the image. The relatively low pixel density on the D3 may hide some weakness in the lenses. Also the 3D probably has a quite strong AA (Anti Aliasing) filter compared to the D300. I guess that the D3 is somewhat better than the D300 with most lenses but the difference may be very small.
High ISO performance is probably better on the D3 than on the D300. The high res version of the D3 will be sharper than the present D3, unless a coke bottle is used instead of a lens.
The final question is how much sharpness is needed? Twelve megapixels are probably good enough for A3 and even A2 prints, if you print smaller I don't really think you can see any difference.
There has been a saying, "There is no substitute for square inches", and it still applies even if it may not hold under all circumstances.
In real life there are a lot of factors:
- Depth of field, improves for small apertures
- Diffraction, gets worse at small apertures
- Noise
- The ever discussed AA-filter
The postprocessing also plays a very important roll.
Erik
Hello All,
Ken Rockwell today declared the sharpness of the D300, the D700 and the D3 to be equal and he has always raved about the color rendition of all Nikons. Would anyone be willing to reflect for a moment, for the benefit of those of us amongst the laity, what factors besides sharpness (and excluding lenses and post processing), are to be considered in choosing between DX and FX regarding IQ? If DX and FX are equally sharp is there a gain to be had in IQ by spending for the FX? Thank you.
Peter Van Dyken
[a href=\"index.php?act=findpost&pid=215092\"][{POST_SNAPBACK}][/a]
-
Why would we assume that a larger device will produce an equal number of electrons of noise? It seems reasonable to expect instead that a increased photosite bulk produces an increase in noise sources. Also, larger well capacities require larger capacities for charge, voltage, or current all along the analog signal transport and processing chain, which seems to have the potential to increase the number of electrons of noise from sources like amplifiers (amp. noise is apparently the dominant noise source in the high end Canon's at low ISO).
This expectation is supported by data for Kodak CCDs, where total dark noise levels in electrons tend to increase with photosite size.
By the way, 1000 electrons is an absurdly high number, suggesting that you are not even vaguely aware of the real values, which run roughly from 3 to 30 electrons in spec. sheets and published measurements for SLR sensors.
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=215226\")
It seems that one's time on a forum and post count are directly related to their propensity to be an arse. At best, you have poor reading comprehension and missed the part where I explain the number is an exaggerated number, which I used to make the ratios significant to be noticed. At worst you're an A-hole who likes to mischaracterize what others have said and cast unwarranted aspersions against them. Alternatively, you could just be an idiot.
If you had at least said something like the noise in the signal to noise ratio is the square root of the number of electrons in the well and not insulted me, I could have simply noted you as someone who is trying to be a little more precise for the OP.
On the other hand, I don't think the OP was interested, nor at a level, to be worrying about the math.
And it appears that you're not aware that there is really no trend for read noise dependent on pixel pitch with modern sensors. For example, a Canon S70 has pixel spacing of 2.3 microns, a full well of 8,200 electrons, and a read noise of 3.2 electrons. A Canon 350D has corresponding numbers of 6.4 microns, 43,000 electrons, and 3.7 electrons respectively. While a Canon 5D has corresponding numbers of 8.2 microns, 80,000 electrons, and 3.7 electrons.
Moreover, I'm not impressed that you've seen some specs on Kodak sensors (are you an astrophotographer?). Why don't you explain to the OP what you mean by dark noise? And while you're at it why don't you tell him about shot noise, reset noise, output amplifier noise, white noise, and flicker noise? Or the noise generated by various ADC's?
You could do all that yet you would illuminate no more for the OP than I did and we would still be right where I left it -- that larger pixels have a higher signal to noise ratio than smaller pixels.
You sound like someone with just enough knowledge to make them dangerous and who thinks the laws of physics are going to be turned on their head and Nikon/Sony, Canon, Kodak, or whomever are going to release a 24 megapixel camera with the same noise levels and high ISO performance of the D3/D700 despite smaller pixels.
If you know how to make that work for them, I'm sure Nikon or Canon will not only pay you a fortune, but you may even win a Nobel.
If you want to know in greater detail why you're wrong in thinking that noise becomes an increasing problem as photosites get larger, this scientist has laid it out pretty well:
[a href=\"http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/index.html]http://www.clarkvision.com/imagedetail/dig...mary/index.html[/url]
http://www.clarkvision.com/imagedetail/doe...el.size.matter/ (http://www.clarkvision.com/imagedetail/does.pixel.size.matter/)
http://www.clarkvision.com/imagedetail/doe...l.size.matter2/ (http://www.clarkvision.com/imagedetail/does.pixel.size.matter2/)
Thanks for reminding me why so many decent people get turned off by internet forums. Maybe I'll spend the weekend telling people here who know what they're talking about they don't know what they're talking about so I can be as full of myself as you are.
-
Alas, I stepped in something visiting the circus and now I've besmirched the landscape
It's still luminous enough :-)
Frames per second and high ISO shooting are not necessarily of concern
High ISO capacity is very important in landscape shooting. Not, that you select high ISO, but because the high ISO capacity is the indicator of great dynamic range.
I often run into the DR limit while shooting landscapes. My ideal camera would be a 1.3x cropping, with relative large pixels, just like the Canon 1DMkIII, but it should weight and cost much less; I don't give a fig for the frame rate.
Btw, DR measurement: DPReview's reviews are generally useful, but their DR measurement is *totally worthless*, it is like a random number generator. For example my Canon 40D allegedly beats the D3 by half stop. In reality the D3 beats the 40D by more than one stop (actually, it is closer to two stops); and this belongs to their smaller blunders, some other measurements are plainly ridiculous.
-
There seems to be a confusion here between total image noise and individual pixel noise
The "total image noise" is a fiction. If you don't need the high pixel count, then buy a camera with less pixels but higher quality (like the D3). If you need many pixels, then don't compare your camera to one with low pixel count.
-
Awhile back when reasearching a camera purchase I stumbled upon DPR, and I am afraid I have developed somewhat of an addiction to it. Want to kick it really do. So many incredibly impolite people.[a href=\"index.php?act=findpost&pid=215266\"][{POST_SNAPBACK}][/a]
Simple solution, tell those idiots what you really think of them. Better still, start a few threads with titles like "This forum is full of fools and trolls", that will get Phil and his associates to kick you out soon enough. I've been booted from DPR, Nikonians and Photo.net -- I still don't know why I was kicked out of Nikonians, so it can be really easy sometimes. I finally had the good sense to kick myself off of Nikon Cafe by changing my email to "Bugoff@mindyourownbusiness.com".
There are just too many silly arguments founded on misconceptions (and one of the biggest purveyors of misconceptions is KR) and your reward for objecting to commonly held misconceptions is to be flamed and drawn into endless and pointless arguments. Even after you've demonstrated why someone is wrong, the next day they will post the same nonsense in another thread and it starts all over again. I say let people believe whatever they want to believe, let them believe that the problem is their camera and that buying a better one will overcome their lack of mastery of the technology and that their images will be sharper.
-
It's still luminous enough :-)
High ISO capacity is very important in landscape shooting. Not, that you select high ISO, but because the high ISO capacity is the indicator of great dynamic range.
I often run into the DR limit while shooting landscapes. My ideal camera would be a 1.3x cropping, with relative large pixels, just like the Canon 1DMkIII, but it should weight and cost much less; I don't give a fig for the frame rate.
Btw, DR measurement: DPReview's reviews are generally useful, but their DR measurement is *totally worthless*, it is like a random number generator. For example my Canon 40D allegedly beats the D3 by half stop. In reality the D3 beats the 40D by more than one stop (actually, it is closer to two stops); and this belongs to their smaller blunders, some other measurements are plainly ridiculous.
[a href=\"index.php?act=findpost&pid=215298\"][{POST_SNAPBACK}][/a]
"High ISO capacity is very important in landscape shooting. Not, that you select high ISO, but because the high ISO capacity is the indicator of great dynamic range."
I am confused..(what else is new)..My 16MP back with (fat pixels) has a lot of noise at ISO 400 but seems to have good DR with landscapes.
Steve
-
My 16MP back with (fat pixels) has a lot of noise at ISO 400 but seems to have good DR with landscapes
Most probably "your back" does not offer any true ISO selection (which is a good thing). Which back is that?
-
Most probably "your back" does not offer any true ISO selection (which is a good thing). Which back is that?
[a href=\"index.php?act=findpost&pid=215307\"][{POST_SNAPBACK}][/a]
Hasselblad CFV I. 4080 x 4080...Has ISO 50, 100, 200, 400.
Steve
-
Hasselblad CFV I. 4080 x 4080...Has ISO 50, 100, 200, 400.
I am not sure of this one, but I guess it is not different from most others.
Most MFDB shooters don't really know the digital characteristics of their back. If you are not lazy to make a test, you can learn yours: shoot anything with constant illumination from tripod, metered for ISO 50, then 100, 200 and 400. Pls upload the raw files (yousendit.com), and we will find out the truth.
Phase One's Plus models do have different ISO gains, the other models not, nor do Sinars. Selecting a higher ISO with those cameras means simply underexposing the shot and "pushing" it in raw conversion.
(I would like to have such a DSLR.)
Addendum
See http://forum.getdpi.com/forum/showthread.p...32563#post32563 (http://forum.getdpi.com/forum/showthread.php?p=32563#post32563) for a demonstration of what is happening.
(The Phase One P25 Plus does not have different ISO gains.)
-
I am not sure of this one, but I guess it is not different from most others.
Most MFDB shooters don't really know the digital characteristics of their back. If you are not lazy to make a test, you can learn yours: shoot anything with constant illumination from tripod, metered for ISO 50, then 100, 200 and 400. Pls upload the raw files (yousendit.com), and we will find out the truth.
Phase One's Plus models do have different ISO gains, the other models not, nor do Sinars. Selecting a higher ISO with those cameras means simply underexposing the shot and "pushing" it in raw conversion.
(I would like to have such a DSLR.)
Addendum
See http://forum.getdpi.com/forum/showthread.p...32563#post32563 (http://forum.getdpi.com/forum/showthread.php?p=32563#post32563) for a demonstration of what is happening.
(The Phase One P25 Plus does not have different ISO gains.)
[a href=\"index.php?act=findpost&pid=215312\"][{POST_SNAPBACK}][/a]
Thanks for the links...I will have to find the time to go through it. It was/is my understanding that with, for example, my D300, the "true" ISOs 200 to 3200 are "real" because the gain is due to electronic amplification and the "L" and "H" "ISOs" below and above 200 and 3200 are not real because they are the result of software exposure correction. So, for example a P-20 would have a fixed ISO and software exposure changes and a P-20+ would have hardware amplification for the ISO change?
Steve
-
The "total image noise" is a fiction. If you don't need the high pixel count, then buy a camera with less pixels but higher quality (like the D3). If you need many pixels, then don't compare your camera to one with low pixel count.
[a href=\"index.php?act=findpost&pid=215300\"][{POST_SNAPBACK}][/a]
Is your tongue in your cheek? What is "the need" for higher pixel count? Are you referring here to people who buy into the marketing ploy that sells the camera with the highest pixel count as the "best" camera? Or are there situations that truly need a higher pixel count?
In my original question about FX vs DX I didn't mean to focus on pixel count. I was trying to understand if the total configuration of the camera with a larger sensor (and the optimum lens and most skilled post processing), the D3, would ultimately give better results in landscape photography than a camera with a smaller sensor (and the optimum lens and most skilled post processing), the D300.
I want to assume the best for myself, that someday in the not too distant future I'll begin to increase my tech skills in this world of digital photography. And if that should happen, I don't want to be looking back saying Gee, I wish I had known to buy such and such. It sticks in my mind that I've heard it said that for landscape photography the 5D is far and away better than the D3 and by asking questions about what makes for the best potential IQ I thought I could discover for myself what camera to buy.
-
It was/is my understanding that with, for example, my D300, the "true" ISOs 200 to 3200 are "real" because the gain is due to electronic amplification and the "L" and "H" "ISOs" below and above 200 and 3200 are not real because they are the result of software exposure correction
Already 3200 is a numerical derivative of 1600 (the pixel values get doubled).
[/quote]So, for example a P-20 would have a fixed ISO and software exposure changes and a P-20+ would have hardware amplification for the ISO change?[/quote]
Is there a P20 Plus?
I know for sure, that the P45 Plus does have different gains, and I guess (based on nothing) that the P30 Plus has too. The P25 Plus is single ISO.
The difference between the fake ISOs of the DSLRs and the MFDBs is, that the DSLRs create raw images, which look like those with true ISO gains, at least at the first sight.
Take a look at the overall histograms, ISO 200, 1600 and 3200. They don't show anything special. However, the "zoomed" histograms are revealing (the next three captures). Here each pixel value is represented by one bar (a column in the histogram) (only 512 values can be shown at once). ISO 200 shows gaps in the red and blue; the meaning is, that the effective number of levels is less than 16384, and the values are "stretched" (without any good reason). ISO 1600 exhitibs the same schema. However, look at ISO 3200; if you load the images in PS and magnify them, you can easily count the effective pixel levels.
The MFDBs I have analyzed do not play this unnecessary (and for the DR counterproductive) game; the true, low (while underexposed) pixel values are stored with higher ISO and the raw converter is instructed by metadata to scale those values differently.
Anyway, the consequence is (if your MFDB is doing the same), that you should regard higher ISO as plain underexposure. If the back delivers good result even when vastly underexposed, that's great (some MFDBs are astonishingly good at it); however, there is no truly equal substitute for proper exposure.
The DR of these MFDBs is great at base ISO, because they "put everything into it", unlike the DSLRs, which show the total DR only in several shots with different ISOs. However, increasing the ISO by one stop decreases the DR of those MFDBs by one stop, while good DSLRs lose one stop in the highlight but can "gain back" perhaps 1/3 stop in the shadows.
-
Is your tongue in your cheek? What is "the need" for higher pixel count?
First of all, Ray compared 6 Mpix with 11 Mpix. I don't think you would qualify 11 Mpix as unnecessarily high.
Second, I don't pretent to know the market better than the camera makers. If Hasselblad at al come out with 40, 50, 60 or how many Mpix machines, then I assume that there is a market for that.
Plus, I assume that those buying these high-end cameras with so many pixels are not doing this in order to be able to downres the image to the half.
In my original question about FX vs DX I didn't mean to focus on pixel count. I was trying to understand if the total configuration of the camera with a larger sensor (and the optimum lens and most skilled post processing), the D3, would ultimately give better results in landscape photography than a camera with a smaller sensor (and the optimum lens and most skilled post processing), the D300
I would not hesitate buying the D3 over the D300 (and over the high-pixel count successor of the D3), if I could justify the cost (my special concern for edge sharpness for stitching is certainly not a general consideration).
It sticks in my mind that I've heard it said that for landscape photography the 5D is far and away better than the D3
I would say this is BS.
I saw only one issue re image quality with the D3: banding. However, this does not affect landscaping; it occurs only when very strongly underexposed and at the same time some areas are oversaturated. Not all copies exhibit this to the same degree, and it disturbs only a few people, who are shooting for example concert in extreme low light, with lamps in the view causing oversaturation.
-
Not so long ago Ken was going on about how 'full-frame' was inherently better than 'crop-frame' in terms of IQ. I guess his opinions change according to his mood.
No, technology changes. But an idiot is still and idiot.
He is fun to read, though. I check his site every day to see if he's written some more idiocy worth laughing at.
-
And it appears that you're not aware that there is really no trend for read noise dependent on pixel pitch with modern sensors.
If you want to know in greater detail why you're wrong in thinking that noise becomes an increasing problem as photosites get larger, this scientist has laid it out pretty well:
http://www.clarkvision.com/imagedetail/dig...mary/index.html (http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/index.html)
[a href=\"index.php?act=findpost&pid=215297\"][{POST_SNAPBACK}][/a]
Looking at the first of your sources, figure 8a shows the 1DMkII with its big CMOS pixels to have 30 electrons of observed dark noise at ISO 100, 16 at ISO 200, etc., while figure 8b shows the S70 with its small CCD pixels to have 3 or 4 electrons of total noise. The 3 or 4 electrons of nosie that Clark mentions is an upper estimate of noise form the photosite itself, but clearly there are substantial additional sources of noise in the final signal. It seems that with Canon's DSLR sensors, the great majority of the noise is not from the photosite itself but from later in the analog stage. The main source seems to be the main ISO gain amplifier, according to an analysis I have seen. Whatever the source, undeniably the total noise is electrons is far greater for the larger photosites in this comparison. (And I do not buy the story that most of the noise comes from Canon using inadequate ADCs in its high end DSLR's.)
By the way, Clark's figures for electrons of noise are inaccurate as a measure of noise arising in the output of the ISO gain amplifier because his correction for the ISO gain factor does not apply to noise arising after the gain. ADU's are a better bottom line measure, and dark noise in ADU's increases with ISO for the 1DMkII (and for any sensor, I would expect.)
This total observed noise has to be taken into account when considering overall S/N ratio. Of course larger photosites will still in general have better S/N ratio than smaller ones (I never suggested otherwise), but the gain with increasing pixel size is not as great as one would conclude by assuming equal total dark noise level.
By the way, the Kodak sensors I was referring to are the CCD's used in various medium format backs, the Leica M8 and R digital back, and early Four Thirds models, not astrophotography sensors. I use Kodak for information about CCDs because Kodak publishes far more details about its CCD's than most other suppliers of CCD's for digital cameras (Sony tells us little about its DSLR sensors). Again, the main dark noise source in these CCDs might be the amplifier between photosite and ADC. For example Kodak has talked of reducing noise in its forthcoming MF sensors by reducing the amp. noise by lowering operating frequency, in turn done by having more output channels and more amps working in parallel.
Likewise Panasonic has talked of reducing noise in recent versions of its MOS sensors by reducing amplifier noise. So amp. noise seem repeatedly to be indicated as a major sensor noise source, at least when the S/N ratio of the photosites is high enough to put high demands on amp. performance.
-
The "total image noise" is a fiction. If you don't need the high pixel count, then buy a camera with less pixels but higher quality (like the D3). If you need many pixels, then don't compare your camera to one with low pixel count.
[a href=\"index.php?act=findpost&pid=215300\"][{POST_SNAPBACK}][/a]
Maybe some people want high resolution at low ISO speeds (for highly detailed stationary subjects like landscapes), while accepting less high resolution at higher ISO (for moving subjects). [We used to use different films for that!]
If a high pixel count sensor has good enough noise levels at lower ISO speeds, it becomes interesting to ask how its "image noise level" at high ISO compares to that of a lower pixel count alternative: suitably processed prints of the same size is my preferred comparison.
Anytime that higher and lower res. sensor alternatives are about equal in "total image noise" or "visible noise on same-sized prints", the one that also offers higher resolution without visible noise problems at lower ISO speeds has an overall IQ advantage. The idea in some digicams of having higher ISO settings only available at lower pixel counts is maybe not so stupid.
-
[]
[/quote]
"The MFDBs I have analyzed do not play this unnecessary (and for the DR counterproductive) game; the true, low (while underexposed) pixel values are stored with higher ISO and the raw converter is instructed by metadata to scale those values differently."
Can the Back manufacture call a "ISO" ISO if the effect is accomplished using software vs hardware amplification. That is the explanation that I was given for Nikon using the "L" and "H" notation.
For a hypothetical, if I assume that my CFV back has a fixed "hardware" ISO of 50 and the preview and histogram are based on an instruction for the "future" RAW conversion, should I push the histogram to the right using a fraction of or a full stop for several shots? I'll look as some shots that were pushed way to the right and see if the highlights are really blown. Or am I going out into the "weeds" here?
Steve
Steve
-
Anytime that higher and lower res. sensor alternatives are about equal in "total image noise" or "visible noise on same-sized prints", the one that also offers higher resolution without visible noise problems at lower ISO speeds has an overall IQ advantage. The idea in some digicams of having higher ISO settings only available at lower pixel counts is maybe not so stupid.
The wish is understandable. However, this situation exists only in fairy tales; two half-pixels don't make one whole, and one whole pixel does not make two halves.
-
Can the Back manufacture call a "ISO" ISO if the effect is accomplished using software vs hardware amplification
They are not in the habit of asking for permission before ignoring and violating standards. At the present, they can do anything and much of the crowd will be cheering.
For a hypothetical, if I assume that my CFV back has a fixed "hardware" ISO of 50 and the preview and histogram are based on an instruction for the "future" RAW conversion, should I push the histogram to the right using a fraction of or a full stop for several shots?
Speculations are a recipe for disaster. It is the question of simple fact, which can easily be determined.
The essence is, again: *if* the ISO gain is real, you can count on some fraction gain by increasing the ISO. The light is not enough, you increase the ISO; you lose one stop in the highlights, and this loss may be "empty" if there are none, but anyway you gain something in the shadows.
However, with the fake ISOs you don't gain anything in the shadows. If you lose something in the highlights depends on the implementation: with DSLRs you alway lose one stop with every ISO stop, because clipping on the level of raw data occurs even if no pixel saturation occured. With the "ISO-less" MFDBs the loss is nominal: the raw data is there, but the raw converter may discard it; this depends on how the camera passes this information along, and how the raw converter reacts. The only certain aspect is, that there is nothing to save with such DSLR raw files, but the MFDB raw files can be manipulated into "recovering" that, what is there but the raw converter believes it is not there.
An example would be much more useful for demonstration than talking about hypothetical situations.
-
It sticks in my mind that I've heard it said that for landscape photography the 5D is far and away better than the D3 and by asking questions about what makes for the best potential IQ I thought I could discover for myself what camera to buy.
[a href=\"index.php?act=findpost&pid=215333\"][{POST_SNAPBACK}][/a]
I do not even think the 5D is "far and away better" than the D300; that tidbit you heard sounds exactly like the kind of trash KR peddles. Most of what distinguishes the image quality coming from these cameras at or near base ISO is the lens you put on it. Understanding exposure, utilizing optimum shooting discipline and using appropriate post processing will deliver exceptional photographs with all the high end DSLRs, to get appreciably more than that you should skip the smaller formats and graduate to MFDBs. This all comes back to people wanting to make up for their lack of understanding and skills by buying a "better" camera, and that's precisely the crowd KR is preaching to.
-
Hi,
Fake ISO isn't that like pushing B&W film?
Erik
They are not in the habit of asking for permission before ignoring and violating standards. At the present, they can do anything and much of the crowd will be cheering.
Speculations are a recipe for disaster. It is the question of simple fact, which can easily be determined.
The essence is, again: *if* the ISO gain is real, you can count on some fraction gain by increasing the ISO. The light is not enough, you increase the ISO; you lose one stop in the highlights, and this loss may be "empty" if there are none, but anyway you gain something in the shadows.
However, with the fake ISOs you don't gain anything in the shadows. If you lose something in the highlights depends on the implementation: with DSLRs you alway lose one stop with every ISO stop, because clipping on the level of raw data occurs even if no pixel saturation occured. With the "ISO-less" MFDBs the loss is nominal: the raw data is there, but the raw converter may discard it; this depends on how the camera passes this information along, and how the raw converter reacts. The only certain aspect is, that there is nothing to save with such DSLR raw files, but the MFDB raw files can be manipulated into "recovering" that, what is there but the raw converter believes it is not there.
An example would be much more useful for demonstration than talking about hypothetical situations.
[a href=\"index.php?act=findpost&pid=215364\"][{POST_SNAPBACK}][/a]
-
Fake ISO isn't that like pushing B&W film?
It is like pushing the exposure in raw conversion. However, there is an important difference: when for example the D300 creates an ISO 3200 file, the multiplication of the ISO 1600 values leads to cutting off anything in the top EV; this can not be reversed any more. In contrast, if the raw data is recorded as ISO 1600, then you can increase it in raw conversion exactly as much as required, while the highlights can be saved, if there are any.
-
Merely replying to your post implied that I was a "fanboy", which is akin to putting me in the awkward position of responding to the proverbial paradox "When did you stop beating your wife?"
[a href=\"index.php?act=findpost&pid=215225\"][{POST_SNAPBACK}][/a]
But Tony, the answer to that was always simply: "the day before I started."
Rob C
-
It seems that one's time on a forum and post count are directly related to their propensity to be an arse. [a href=\"index.php?act=findpost&pid=215297\"][{POST_SNAPBACK}][/a]
Catman,
I'm surprised that BJL has deigned to reply to you. It saddens me that such abusive language is directed to our senior and longstanding members of this admirable forum.
Your lack of respect is of no advantage to you.
-
Simple solution, tell those idiots what you really think of them. Better still, start a few threads with titles like "This forum is full of fools and trolls", that will get Phil and his associates to kick you out soon enough.
Not a bad suggestion.
I pretty much write in forums the same way i talk in "meat space", so I generally simply ignore trollish or impolite posts directed at my comments. I used to argue for a living. If some wants me to argue, they better pay me,
My addiction is not to the argument, I think it is to the weirdness...
-
The answer that I take away from your responses is that FX does not necessarily offer better image quality than DX so my ambiguity about which format to purchase lingers on. Thanks to all who responeded here. I appreciate your reflections.
Peter
-
Would anyone be willing to reflect for a moment, for the benefit of those of us amongst the laity, what factors besides sharpness (and excluding lenses and post processing), are to be considered in choosing between DX and FX regarding IQ? If DX and FX are equally sharp is there a gain to be had in IQ by spending for the FX?[a href=\"index.php?act=findpost&pid=215092\"][{POST_SNAPBACK}][/a]
That depends on what your shooting constraints are, particularly as regards ISO -- at high ISO Nikon's current FX DSLRs win hands down. Likewise, for shallower DOF the FX format is better, but a good, fast lens can get you a pretty darn shallow DOF even on a DX sensor. FWIW, I'm more interested in landscapes and I will purchase 24mm and 45mm tilt/shift lenses before I purchase a D700. In fact, I will skip the D700 and wait for an affordable higher MP FX DSLR, which is just as well as I will be spending nearly $4K on those lenses.
-
Tests on DP review of the D3 (and so D700) appear that the IQ is identical to the 5D and low iso.
I have a D700 and all I can say is that it doesn't come close to imacon scanned 6x7. It will not see much landscape use.
-
The "total image noise" is a fiction. If you don't need the high pixel count, then buy a camera with less pixels but higher quality (like the D3). If you need many pixels, then don't compare your camera to one with low pixel count.
[a href=\"index.php?act=findpost&pid=215300\"][{POST_SNAPBACK}][/a]
Gabor,
I wouldn't call it a fiction, but rather a perception. I see a distinction between pixel noise and total image noise.
When comparing noise levels on our monitor to see which camera produces lower noise at which ISO, we sometimes forget that the degree of magnification of the crops we are examining is representative of a huge print, probably a far bigger print than we would ever make from a single frame.
When comparing, for example, a 1Ds3 crop with a D3 crop, both at 100%, the 1Ds3 crop is representative of a significantly larger print than is the D3 crop. The print would be roughly 1.33x longer in each dimension.
Now, I shan't argue that one should therefore view the larger print from a distance that is 1.33x greater, because one can't dictate what viewing distance should be. People will view a print from any distance they like, and in any case, people's acuity of vision varies so much it's not meaningful to have a fixed rule about viewing distance.
But I shall argue that, when comparing noise, it's more meaningful for the viewer to examine same size prints or same size image detail in the crops on the monitor.
Let's consider the dpreview comparison of the D3, 1Ds3, D300 and 5D at ISO 3200, below.
[attachment=7963:attachment]
Each image (crop) is comprised of the same number of pixels. It's a 'pixel for pixel' comparison of noise.
The D3 image in this comparison definitely looks cleaner than the 1Ds3 image. In particular, there is less chroma noise in the D3 image. I understand these are all jpegs straight out of the camera.
But what happens if we equalise the amount of image data in each crop by interpolating the D3 image so that the Queen's face is the same size as in the 1Ds3 crop? The noise in the D3 image then becomes more obvious, but it still has the advantage of lower chroma noise because that's already been removed in-camera. That's certainly an advantage for jpeg shooters, (which I'm not.)
If we pass the 1Ds3 image through a program like Noise Ninja which can specifically address chroma noise without applying the luminance NR which tends to soften resolution, we find that the noise in the 1Ds3 image becomes roughly on a par with that in the D3 image, but the 1Ds3 image is clearly more detailed.
Some have suggested that you can do the same for the D3 image. Pass it through a noise reduction program. You can, but you can't remove chroma noise twice. The only improvement that can be made to the D3 image (according to my tests) is to apply luminance NR and soften the image even more, resulting in the 1Ds3 image being even sharper by comparison.
-
I have a D700 and all I can say is that it doesn't come close to imacon scanned 6x7. It will not see much landscape use.
[a href=\"index.php?act=findpost&pid=215705\"][{POST_SNAPBACK}][/a]
If you consider adding stitching to the equation, then my view is that the D700 can easily outdo the Pentax for landscape usage.
Unless you do B&W, the D700 is already superior in terms of DR for single images compared to Imacon scanned slide film, with stitching you will also get superior results in terms of detail.
I won't even mention convenience, since it isn't even close.
Granted, stitching isn't always easy, but shooting slide film with grad filters isn't super easy either, is it?
Back to the original question, my view remains that APS sensors remain today the best option for those willing to do classical landscape with a lot of DoF. It is also superior for people needing to reach far, like most landscape shooters. The only area where FX has a very clear lead that will never go away is in shallow DoF situation.
Now the problem is that we don't know how far manufacturers will be able to go/want to go in terms of detail and DR with APS sensors. I don't expect Canon, Nikon and Sony to try to work hard on this, since they will want to preserve the differentiation of their FX high end bodies.
My view is that this decision is not driven mostly by the technological limitations of APS, it is mostly driven by the fact that these manufacturers know they can make more money with FX.
This uncertainty led me to invest in a D3 and matching lenses that I have been shooting happily with for 9 months. I have just sold my D2x yesterday (it was still in a very good condition actually) at a very low price on Yahoo Auction, following on most of my DX lenses, and my only remaining footprint in DX today is a D80.
Do I like the D3 for landscape work? It is a wonderful camera delivering very nice tonalities and colors. Detail is also excellent, but so was the detail of the D2x. Do I suffer from the lack of DoF? Yes I clearly do and use a lot DoF stacking a lot more than I used to.
Does it mean that I will not invest anymore in DX in the future? Probably not. I would probably buy a video enabled D90 with the new 18-105 VR lens for casual shooting (I would keep one in my bag all the time). I might even consider getting a D300 at some point of time as a back up since it does take the same batteries as the D3.
Now, generally speaking, I expect manufacturers like Pentax to keep pushing the DX enveloppe as far as they can. This is probably why they have chosen to work with Samsung for their sensors. They have understood that Sony had given up on high end APS...
Cheers,
Bernard
-
If you consider adding stitching to the equation, then my view is that the D700 can easily outdo the Pentax for landscape usage.
...
I won't even mention convenience, since it isn't even close.
Granted, stitching isn't always easy, but shooting slide film with grad filters isn't super easy either, is it?
Cheers,
Bernard
[a href=\"index.php?act=findpost&pid=215736\"][{POST_SNAPBACK}][/a]
I tend to agree, and moreover such is the ability of stitching software to stitch even hand held images seamlessly these days, a stitched D700 image would be a lot easier and faster to produce than a scanned transparency. It won't look identical because the processes are too different and stitching cant be used for everything, but most landscapes and even architecture would work. Add a pano head and HDR in to the mix, and you are way ahead of scanned film. Taken to its logical conclusion, then a D700 becomes an alternative to MF backs and cameras for many purposes, as well as film.
Quentin
-
I tend to agree, and moreover such is the ability of stitching software to stitch even hand held images seamlessly these days, a stitched D700 image would be a lot easier and faster to produce than a scanned transparency. It won't look identical because the processes are too different and stitching cant be used for everything, but most landscapes and even architecture would work. Add a pano head and HDR in to the mix, and you are way ahead of scanned film. Taken to its logical conclusion, then a D700 becomes an alternative to MF backs and cameras for many purposes, as well as film.
Quentin
[a href=\"index.php?act=findpost&pid=215747\"][{POST_SNAPBACK}][/a]
I've tried stitiching a few times and without pano gear it is often pretty difficult in post processing. I will spend more time with it since many people have good success.
Here is my issue with multi shot techniques:
Take a typical landscape image: foreground interest, high brightness difference between the foreground/sky, desire for high quality. You decide to shoot 4 frames alonge and two high, a total of 8 frames. Close foreground focus can be handelled by focus blending (say 3 shots) and the contrast by HDR (maybe 2 shots) right? Now you need to shot a minimum 8 x 3 x 2 shots which gives 48 shots all of which need to be focused the same on each pass or the stitching software throws it's toys out the pram (ok, some of the 'sky only' shots may not need to be duplicated).
Add to this the fact that framing and visulising the composition becomes very difficult and that the light may be fast changing and stiching suddenly isn't quite to appealing. Last time I was out I was glad to get back to using film. With my LF camera (shooting 6x7 with roll film back or 4x5 b&w sheet film) I can plonk it on the tripod (it's non-folding), compose on the groundglass, 10 secs to focus with lens tilt, exposue and job done.
Film goes off to the lab, I can forget about it. It comes back a few days later and I send a few frames off for scanning. Or for b&w I dev the film which I don't mind doing since I'm no pro with deadlines and send the frame(s) off to be hand printed.
Since I've only just got a digital camera perhaps people more familiar with using small format digital can help me better use my D700 in the field for landscapes.
-
I've tried stitiching a few times and without pano gear it is often pretty difficult in post processing. I will spend more time with it since many people have good success.
Here is my issue with multi shot techniques:
Take a typical landscape image: foreground interest, high brightness difference between the foreground/sky, desire for high quality. You decide to shoot 4 frames alonge and two high, a total of 8 frames. Close foreground focus can be handelled by focus blending (say 3 shots) and the contrast by HDR (maybe 2 shots) right? Now you need to shot a minimum 8 x 3 x 2 shots which gives 48 shots all of which need to be focused the same on each pass or the stitching software throws it's toys out the pram (ok, some of the 'sky only' shots may not need to be duplicated).
Since I've only just got a digital camera perhaps people more familiar with using small format digital can help me better use my D700 in the field for landscapes.
[a href=\"index.php?act=findpost&pid=215767\"][{POST_SNAPBACK}][/a]
That's a worst-case scenario. The real world is a lot more forgiving. I'm using a Canon Eos-1Ds III, and I find that with absolutely perfect technique I can occasionally shoot something so sharply it can withstand enlargement up to 24x36" and still look really great. The ability of a digital file to withstand sharpening is far greater than scanned film, where you generally end up sharpening the film grain. If I need more resolution, especially for a pano, I start stitching. But just a moment or two to level the camera makes post-processing pretty simple. The hardest thing I've done is HDR followed by stitching for a big panoramic; but you can't even begin to do something like that using film. I always hated the obvious dark line across film images using a graduated neutral density filter; too many darkened tree-tops. As far as focus blending goes- with medium format film, depth of field issues rapidly become a major PITA compared with 35 mm or APS-sized sensors. You may need to bite the bullet and go to a view camera to deal with that. And the drudgery of cleaning and scanning film, not to mention the difficulty finding really good processing anymore...
Plus it's all a moving target. I can safely predict that within 5 years or so one of the major D-SLR makers will introduce a body that automatically captures HDR frames, making the whole thing a lot easier.
Just my 2¢.
-
Deleted
-
That's a worst-case scenario. The real world is a lot more forgiving. [a href=\"index.php?act=findpost&pid=215778\"][{POST_SNAPBACK}][/a]
Even then, it isn't too hard to deal with such extreme situations when they show up:
1. HDR pano sample
(http://farm3.static.flickr.com/2415/2480215400_83008210fb_o.jpg)
2. Infinite DoF with moving objects
(http://farm4.static.flickr.com/3088/2692286765_999934449c_o.jpg)
3. Panorama shot in the rain with moving objects
(http://farm4.static.flickr.com/3163/2630287125_2e7e86ac3f_o.jpg)
All shot with a D3.
Cheers,
Bernard
-
That's a worst-case scenario. The real world is a lot more forgiving. I'm using a Canon Eos-1Ds III, and I find that with absolutely perfect technique I can occasionally shoot something so sharply it can withstand enlargement up to 24x36" and still look really great. The ability of a digital file to withstand sharpening is far greater than scanned film, where you generally end up sharpening the film grain. If I need more resolution, especially for a pano, I start stitching. But just a moment or two to level the camera makes post-processing pretty simple. The hardest thing I've done is HDR followed by stitching for a big panoramic; but you can't even begin to do something like that using film. I always hated the obvious dark line across film images using a graduated neutral density filter; too many darkened tree-tops. As far as focus blending goes- with medium format film, depth of field issues rapidly become a major PITA compared with 35 mm or APS-sized sensors. You may need to bite the bullet and go to a view camera to deal with that. And the drudgery of cleaning and scanning film, not to mention the difficulty finding really good processing anymore...
Plus it's all a moving target. I can safely predict that within 5 years or so one of the major D-SLR makers will introduce a body that automatically captures HDR frames, making the whole thing a lot easier.
Just my 2¢.
[a href=\"index.php?act=findpost&pid=215778\"][{POST_SNAPBACK}][/a]
How many frames are you stitching?
I shoot 6x7 (and 4x5 b&w) on an Ebony LF camera due to the dof issue. This is what drove me mad with my mamiya 7 which my D700 now replaces as it is a more workable travel/wedding/mountain hikes camera.
So far the dynamic range the D700 can deal with has impressed me so needing HDR may only be required on occasion, processing RAW files twice with different settings might be enough or simply using auto-bracketing to shoot the other exposures.
Bernard, impressive shots under difficult conditions for stitching. Looks like a lot can be accomplished after figuring out the process some more.
-
I might even consider getting a D300 at some point of time as a back up since it does take the same batteries as the D3.
Now, generally speaking, I expect manufacturers like Pentax to keep pushing the DX enveloppe as far as they can. This is probably why they have chosen to work with Samsung for their sensors. They have understood that Sony had given up on high end APS...
[a href=\"index.php?act=findpost&pid=215736\"][{POST_SNAPBACK}][/a]
The D300 only uses the same batteries as the D3 if you use the MB-D10.
I believe that Nikon, Sony and Canon will defend DX and APS-C markets against competitors -- they will not likely neglect the much more widely adopted smaller format and give smaller players an opening to chip away at overall market share which might allow those competitors to eventually compete for the higher end, larger format users by establishing a base of consumers invested in their lenses and accessories. Here's a very notable piece of evidence of Nikon's commitment to DX, their newest top of the line Speedlight has a DX mode.
-
Take a typical landscape image: foreground interest, high brightness difference between the foreground/sky, desire for high quality. You decide to shoot 4 frames alonge and two high, a total of 8 frames. Close foreground focus can be handelled by focus blending (say 3 shots) and the contrast by HDR (maybe 2 shots) right? Now you need to shot a minimum 8 x 3 x 2 shots which gives 48 shots all of which need to be focused the same on each pass or the stitching software throws it's toys out the pram (ok, some of the 'sky only' shots may not need to be duplicated)
There certainly are some settings not suitable for panorama technique. OTOH, focus blending can often be avoided by targeted framing. Bernard's second image (the pod) is an excellend demo of this: it could be shot in several raws, so that every frame is well-focused on its own.
Add to this the fact that framing and visulising the composition becomes very difficult and that the light may be fast changing and stiching suddenly isn't quite to appealing
Changing light can be a real PITA even with only a few frames (what about moving clouds?). This would not ruin the pano, but the pre-stitching preparation would become more laborous.
On the other side is, that the DR of the entire scenery may be so high, that it can not be captured even by the best MFDB, but often it works well with several smaller frames (using variable exposure, again lots of pre-stitching preparation).
-
this situation exists only in fairy tales; two half-pixels don't make one whole, and one whole pixel does not make two halves.
[a href=\"index.php?act=findpost&pid=215361\"][{POST_SNAPBACK}][/a]
What situation are you referring to? Processes like downsampling can give output pixels with higher S/N ratio and so higher DR than that of the input pixels, so the mere fact that the individual pixels of a 24MP sensor have lower S/N ratio that those of an equally large 12MP sensor do not tell us how the S/N ratio will be in the pixels of, for example, a 12MP TIFF produced by downsampling of that 24MP output.
By the way, it is also fairly clear that downsamping 24 million Bayer CFA pixels to a 12MP RGB file (e.g. TIFF) gives more resolution that a 12MP Bayer CFA sensor. There are many examples of the fact that downsampling to X MP from a higher MP sensor gives more resolution than a Bayer CFA sensor of X MP. (The basic reason is that the 12MP RGB file has 12 million values for each of R, G, and B derived from 12 million G, 6 million each of R and B, while the 12MP sensor has only 6 million G, 3 million each of R and B.)
So if anything, downsizing output of a 24MP sensor to match the resolution of a 12MP sensor for a visible noise and dynamic range comparison might involve downsampling the 24MP file to somewhat less than 12MP.
This sort of complication is why I prefer the idea of just comparing equal sized prints, relying on the higher PPI used with the higher MP file to reduce the visible effects of the per pixel S/N ratio.
[Edit] P. S. I do not expect such "down-ressing" to completely eliminate the noise level advantage of a sensor with fewer, bigger photosites, so those who never want more than X MP are probably better off with a camera of X MP rather than 2X MP. But the processing or printing procedures could reduce the gap in visible noise levels and DR to the point that a photographer who sometimes has a use for higher resolution can benefit from the higher pixel count sensor without losing much sleep over how much worse high ISO images will look.
-
So if anything, downsizing output of a 24MP sensor to match the resolution of a 12MP sensor for a visible noise and dynamic range comparison might involve downsampling the 24MP file to somewhat less than 12MP.
This sort of complication is why I prefer the idea of just comparing equal sized prints, relying on the higher PPI used with the higher MP file to reduce the visible effects of the per pixel S/N ratio.
[a href=\"index.php?act=findpost&pid=215881\"][{POST_SNAPBACK}][/a]
Indeed! And that extra resolution from the sensor with the higher pixel count can be traded for lower noise by passing the image through a noise reduction program in situations, or for print sizes, where low noise is of greater priority than high resolution.
These factors when combined will tend to ensure that, for example, no 1Ds3 image will be be at a noise disadvantage compared with a D3 image. However, it is certainly true that there will be circumstances when a D3 image will be at a resolution disadvantage compared with a 1Ds3 image.
I'm also not convinced by Panopeepers's argument that one should get a camera with a pixel count that's just sufficient for one's purposes. That might involve carrying many cameras for different size prints and for situations that cannot be predicted. For example, I don't find it ideal that, in the interests of best image quality and longest reach using my longest telephoto lens, I feel it necessary to always travel with a 20D in addition to a 5D. Both purposes could be served with one full frame 35mm camera with the pixel pitch of the 20D.
However, the 1Ds3 is both too expensive and too heavy for me .
-
The D300 only uses the same batteries as the D3 if you use the MB-D10.
I believe that Nikon, Sony and Canon will defend DX and APS-C markets against competitors -- they will not likely neglect the much more widely adopted smaller format and give smaller players an opening to chip away at overall market share which might allow those competitors to eventually compete for the higher end, larger format users by establishing a base of consumers invested in their lenses and accessories. Here's a very notable piece of evidence of Nikon's commitment to DX, their newest top of the line Speedlight has a DX mode.
[a href=\"index.php?act=findpost&pid=215802\"][{POST_SNAPBACK}][/a]
I am not saying that Nikon/Canon will not continue investing in DX, I am saying that they will probably not release very high end DX bodies able to compete with the high end FX models in terms of pixel count. This is why I am personnally giving up on DX although I feel that it is probably the best option for landscape shooting.
As far as the D300 goes, yes I am aware that the MB10 is required to use the batteries of the D3.
Cheers,
Bernard
-
There certainly are some settings not suitable for panorama technique. OTOH, focus blending can often be avoided by targeted framing. Bernard's second image (the pod) is an excellend demo of this: it could be shot in several raws, so that every frame is well-focused on its own.
[a href=\"index.php?act=findpost&pid=215817\"][{POST_SNAPBACK}][/a]
Yes on the principle, but with this particular image each of the individual frames is itself a DoF stacking made up of 3 or 4 images.
Cheers,
Bernard
-
So, how did we progress from a topic relating to equality of sharpness between FX and DX, to very convoluted procedures on image stitching?
I thought I was the master of 'straying off topic'. Not that I disapprove. Everything (everybody) is related to everything, to some degree.
The issue of cropped format versus full format, is really one of shadow noise, dynamic range and availability of high quality wide-angle lenses.
I don't believe you could get, with a D300, the sort of image you can get with a D3 using the 14-24/2.8 Nikkor lens. There's no equivalent quality 9mm lens for the D300.
I guess this is where stitching comes in.
-
I am not saying that Nikon/Canon will not continue investing in DX, I am saying that they will probably not release very high end DX bodies able to compete with the high end FX models in terms of pixel count. This is why I am personally giving up on DX although I feel that it is probably the best option for landscape shooting.
Cheers,
Bernard
[a href=\"index.php?act=findpost&pid=215917\"][{POST_SNAPBACK}][/a]
I don't see how a future camera that is more diffraction limited would have any bearing on the relative image quality of a D300 versus a D700, or which one you would choose to use for landscape photography.
-
I don't believe you could get, with a D300, the sort of image you can get with a D3 using the 14-24/2.8 Nikkor lens. There's no equivalent quality 9mm lens for the D300.
I guess this is where stitching comes in.
[a href=\"index.php?act=findpost&pid=215920\"][{POST_SNAPBACK}][/a]
(http://photos.imageevent.com/tonybeach/mypicturesfolder/august2008/huge/_3A20115-22_web.jpg)
Yes, and you couldn't get this image with an FX sensor and a 14-24 without stitching, just as I did with that lens and the D300.
-
I don't see how a future camera that is more diffraction limited would have any bearing on the relative image quality of a D300 versus a D700, or which one you would choose to use for landscape photography.
[a href=\"index.php?act=findpost&pid=215925\"][{POST_SNAPBACK}][/a]
The question is whether or not you want to stay longer on the DX track that does - per my explanation above - not have a very bright future for high end.
I have personnally chosen to make the jump quickly, but it is of course not a problem to keep using DX.
Cheers,
Bernard
-
Yes, and you couldn't get this image with an FX sensor and a 14-24 without stitching, just as I did with that lens and the D300.
[a href=\"index.php?act=findpost&pid=215927\"][{POST_SNAPBACK}][/a]
Interesting image, but I get a sense there's something cropped off the lower half at the bottom. Would you have preferred an FX sensor?
-
Interesting image, but I get a sense there's something cropped off the lower half at the bottom. Would you have preferred an FX sensor?
[a href=\"index.php?act=findpost&pid=215936\"][{POST_SNAPBACK}][/a]
I find foregrounds on 180° or greater FOV images problematic because the center becomes pinched. Also, in that particular image the foreground would have become much deeper and disrupted the balance to my eye, and at some point my own shadow (rather distracting) would have ended up in the frame. Therefore, I would have simply zoomed the 14-24 out more -- perhaps a little wider, like 18mm. I will mention DOF issues between formats later in this post, but in this image that was not an issue. I find though that I need to stop down to f/11 on DX with my copy of the 14-24 shooting deep DOF images or I get peculiar edge softness at infinity -- I do not know what would be required to avoid the same phenomenon on an FX sensor, but it may well be f/16.
Now as to whether I would prefer to use an FX sensor, I would say that depends. I would see no advantage in that image using a D700 or D3, but if I had a 24 MP FX DSLR then perhaps. However, there is a question of DOF and subject motion. If I stick to f/11 I lose DOF (although that's not an issue in this image) and the edges might start to become unacceptably soft; but if I go to f/16 then I end up with blur introduced by the wind blowing the grass and from the people walking across the scene. So there you go, it's all about trade-offs between formats; there's no substitute ("magic bullet") for understanding how to maximize what you can get from your gear, and the biggest limitations to getting improved photographs in descending order are: vision, technique, support, lens, format, and camera.
Going back to your earlier post where you compared an FX sensor using a 14mm focal length to a DX sensor using a non-existing rectilinear 9mm focal length, and the need to stitch 14mm DX images together to equal the FOV of a single 14mm FX image, well if both cameras (D300 and D700) start with 12 MP, then the DX image will have more resolution and will have been taken at twice the shutter speed or twice the ISO for an equivalent DOF. Applying this scenario to the OP, the answer as to which image will be sharper becomes obvious.
-
Going back to your earlier post where you compared an FX sensor using a 14mm focal length to a DX sensor using a non-existing rectilinear 9mm focal length, and the need to stitch 14mm DX images together to equal the FOV of a single 14mm FX image, well if both cameras (D300 and D700) start with 12 MP, then the DX image will have more resolution and will have been taken at twice the shutter speed or twice the ISO for an equivalent DOF. Applying this scenario to the OP, the answer as to which image will be sharper becomes obvious.
[a href=\"index.php?act=findpost&pid=216044\"][{POST_SNAPBACK}][/a]
Tony,
I'm not at all sure this is the case, that the D300 would be sharper. If one considers the situation where maximum DoF is required (without softening the image too much with diffraction effects), then one is comparing, say, F16 with the D3 and F10 or F11 with the D300. Because the D300 has greater pixel density, it needs the higher MTF of a lens at F11. Therefore, the D3 should deliver approximately the same resolution at F16 as the D300 at F11, as well as the same DoF.
If one moves to the other end of the range of apertures, and compares the D300 at F2.8 with the D3 at F4.5, then it's likely that the lens used will be sharper at F4.5 than it is at F2.8. In this situation, the D300 needs a lens at a sharper aperture but is in fact getting the opposite, a lens which is less sharp at the aperture required for equivalent DoF.
However, it might be the case that, irrespective of lens quality at particular apertures, the wider aperture that the D300 can always use for an equivalent DoF, allows for use of a faster shutter speed, and as we all know, shutter speed is often critical for a sharp result.
On the other hand, it seems quite clear that the D3 has lower noise at high ISO than the D300, just as the 5D has lower noise perhaps to a lesser extent (because it's older technology), than the 40D and 450D, and not just at high ISO.
It would be interesting to compare a D300 image at F2.8 and ISO 200 with a D3 image at F4 and ISO 400, using the same lens. Shutter speed and DoF should be the same, but what about resolution? My bet is, the FX sensor would deliver better results.
-
Thank you Ray for bringing us back to the topic raised by the OP.
Tony,
I'm not at all sure this is the case, that the D300 would be sharper. If one considers the situation where maximum DoF is required (without softening the image too much with diffraction effects), then one is comparing, say, F16 with the D3 and F10 or F11 with the D300. Because the D300 has greater pixel density, it needs the higher MTF of a lens at F11. Therefore, the D3 should deliver approximately the same resolution at F16 as the D300 at F11, as well as the same DoF.[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=216068\")
Well, if you are comparing single frames then they should be about the same, but if you stitch three images together from the D300 (recall that we were comparing 14mm on FX to 14mm on DX and I'm making an allowance for necessary overlap) then you will have more resolution from the DX camera by virtue of the stitching. Now you can zoom out and stitch with the FX, but I think the images will mostly be indistinguishable from one another except at some extreme level of tonal detail that probably exceeds what is printable.
If one moves to the other end of the range of apertures, and compares the D300 at F2.8 with the D3 at F4.5, then it's likely that the lens used will be sharper at F4.5 than it is at F2.8. In this situation, the D300 needs a lens at a sharper aperture but is in fact getting the opposite, a lens which is less sharp at the aperture required for equivalent DoF.
Yes, there is no doubt that at the other end of the aperture spectrum the FX sensor has the advantage. However, at what price and are you further ahead buying the more expensive camera and using cheaper lenses or buying more expensive lenses and using the cheaper camera? My bet is that a cheaper lens on an expensive camera delivers less than an expensive lens on a cheaper camera (considering that were comparing the D300 to the D700 or D3).
However, it might be the case that, irrespective of lens quality at particular apertures, the wider aperture that the D300 can always use for an equivalent DoF, allows for use of a faster shutter speed, and as we all know, shutter speed is often critical for a sharp result.
On the other hand, it seems quite clear that the D3 has lower noise at high ISO than the D300, just as the 5D has lower noise perhaps to a lesser extent (because it's older technology), than the 40D and 450D, and not just at high ISO.
I believe this gets to the crux of the OP's question. Are you really gaining much if you end up dialing in a higher ISO? Perhaps you are and that will depend on where you start and the quality of the light -- the D3 and D700 seem optimized for artificial lighting, the D300 also seems to be but not as much so. If you want to shoot in lowlight situations, have excellent lenses, and can afford it -- get the D700 or D3; if reach is a consideration, and related to that money -- then a D300 might be better, especially if there are improvements that can be purchased by spending on lenses.
It would be interesting to compare a D300 image at F2.8 and ISO 200 with a D3 image at F4 and ISO 400, using the same lens. Shutter speed and DoF should be the same, but what about resolution? My bet is, the FX sensor would deliver better results.
Zoom or prime? Using the same focal length and moving forward or back to maintain FOV but change perspective? Using the same focal length and standing in the same spot and changing FOV to maintain perspective? You see, now you have opened up a real can of worms -- comparing primes in different formats are apples to oranges comparisons. Here's a thought though, lets compare the same lens standing in the same spot with the same perspective -- in other words, lets compare zoom lenses. Start with a very nice moderate tele-zoom: [a href=\"http://www.photozone.de/nikon--nikkor-aps-c-lens-tests/247-nikkor-af-s-70-200mm-f28-g-if-ed-vr-review--test-report?start=1]Nikkor 70-200/2.8 VR.[/url] Wide open on DX compared to f/4 on FX (with adjustments to focal length to get equal perspective and FOV), FX has an advantage with this lens, but that pretty much goes away comparing f/4 to f/5.6. Looking at the premier mid-range zoom: Nikkor 24-70/2.8 (http://www.photozone.de/nikon--nikkor-aps-c-lens-tests/236-nikkor-af-s-24-70mm-f28g-ed-review--test-report?start=1) the two formats are pretty much indistinguishable even wide open, but that lens is just ridiculously good.
Sharpness we can measure, noise too; although measuring these can become more complicated than most assume by just looking at published charts. There are intangible qualities too, and there the larger format usually excels. At the extremes there are trade-offs that make the formats comparable, with an advantage to the larger format at the wide end and an advantage for the smaller format at the longer end, and in the middle I would choose FX. Yesterday though, Thom Hogan wrote at DPR that he would choose two D300 cameras and a GX-100 for a trip to Africa, that's his choice and YMMV. There are a lot of considerations that make a blanket FX format is always better than DX format both simplistic and sometimes just wrong.
-
Tony,
Rather than reply to every point you've just raised, I'll just make a balnket statement which I beleive covers everything.
A D3 upgrade with a D300 pixel pitch (and facility to use DX lenses) will give you the best of both worlds, and perhaps even more with a bit of technological improvement.
-
A D3 upgrade with a D300 pixel pitch (and facility to use DX lenses) will give you the best of both worlds, and perhaps even more with a bit of technological improvement.
[a href=\"index.php?act=findpost&pid=216087\"][{POST_SNAPBACK}][/a]
It is the camera I'm waiting for; but I would prefer a smaller body and I will have to wait for one that is affordable (perhaps late next year). However, I am reasonably certain that even with technological improvements we will not get D3/D700 pixels from a D300 pixel pitch DSLR in the near future, but all we can do is speculate about what we will or will not get from any future camera. Again, it's about trade-offs and blanket statements are simply too broad to always be applicable in all circumstances.
One thing I've been reading about the DX crop mode on the D3 is increased noise. It's something peculiar and unexpected, and if it happens in Nikon's future DSLRs it will make the DX crop mode useless to me. Other than that, I would like the option of switching to cropped modes (even with non-DX lenses) to keep files smaller for more casual shots and when I don't need or want 24+ MP files.
-
OK. I'm stumbling along behind you guys trying to garner as much lingo and knowledge as I can.
"D3 upgrade with a D300 pixel pitch" The pitch is the distance between the pixels? Does this mean that your ideal camera would have a full size sensor with the pixels the same size as those in the D300 and the same distance apart as the pixels in the D300? Thereby you would maintain the resolution offered by the D300 and.....?
Peter
-
The question is whether or not you want to stay longer on the DX track that does - per my explanation above - not have a very bright future for high end
The cropped FF cameras have at least as bright future beside FF, as the FF cameras have beside MF backs. I don't see any reason to make a sharp division between the sensors of P&S and the MFDBs (fakes all together) just through the middle of the 1.5x or 1.6x croppers.
In fact, I have *never* seen a solid reasoning, how the problem of the lack of overall sharpness could be solved on FF, as opposed to the cropping cameras using the cream of the lens.
-
One thing I've been reading about the DX crop mode on the D3 is increased noise. It's something peculiar and unexpected
It is not "unexpected"; it is a ridiculous statement. I am saying this without having any proof of it, and I am ready to eat my crow publicly if anyone can prove that there is such a connection.
-
In fact, I have *never* seen a solid reasoning, how the problem of the lack of overall sharpness could be solved on FF, as opposed to the cropping cameras using the cream of the lens.
[a href=\"index.php?act=findpost&pid=216147\"][{POST_SNAPBACK}][/a]
I might not have been clear enough. I do not believe that technical reasons will limit the developement of high end DX bodies with more pixels. What will kill these developements is the willingness of Nikon and Canon to focus on higher margin FX bodies and to keep them differentiated enough compared to DX.
Cheers,
Bernard
-
One thing I've been reading about the DX crop mode on the D3 is increased noise. It's something peculiar and unexpected, and if it happens in Nikon's future DSLRs it will make the DX crop mode useless to me.
[a href=\"index.php?act=findpost&pid=216100\"][{POST_SNAPBACK}][/a]
Well, at a given print size yes, since you have 2.25 times less pixel in DX mode than in FX mode, you need to "enlarge" more, and the noise will therefore be more visible.
If you are talking about pixel quality, then the DX and FX modes are of course identical.
Cheers,
Bernard
-
OK. I'm stumbling along behind you guys trying to garner as much lingo and knowledge as I can.
"D3 upgrade with a D300 pixel pitch" The pitch is the distance between the pixels? Does this mean that your ideal camera would have a full size sensor with the pixels the same size as those in the D300 and the same distance apart as the pixels in the D300? Thereby you would maintain the resolution offered by the D300 and.....?
Peter
[a href=\"index.php?act=findpost&pid=216125\"][{POST_SNAPBACK}][/a]
Yes. This sort of leap-frogging between Canon cropped formats and full frame has been going on for a while. The first Canon full frame, the 1Ds, had more pixels than the then current 6mp cropped formats, the D60 and 10D, and a wider pixel pitch.
The successor to the 1Ds, the 16mp 1Ds2 had about the same pixel pitch as those earlier cropped formats, but in the meantime Canon cropped formats had moved up to 8mp in the form of the 20D and 30D.
The latest Canon full frame, the 21mp 1Ds3, has the same pixel pitch as the now discontinued 20D & 30D, but in the meantime the bar has been raised and Canon now have both 10mp and 12mp cropped format models.
The trend would indicate that the successor to the 1Ds3 will be around 31mp if it achieves the same pixel pitch as the 450D.
-
Sorry. Never mind.
-
It is not "unexpected"; it is a ridiculous statement. I am saying this without having any proof of it, and I am ready to eat my crow publicly if anyone can prove that there is such a connection.
[a href=\"index.php?act=findpost&pid=216148\"][{POST_SNAPBACK}][/a]
It is something Thom Hogan wrote, but I need to confirm that with him as I cannot find where he posted it at DPR. I will update this post when I (hopefully) hear back from him; then one of us will be eating crow unless you reject Thom's observation.
I got Thom's response and according to him the problem did exist but has apparently been resolved.
-
Well, at a given print size yes, since you have 2.25 times less pixel in DX mode than in FX mode, you need to "enlarge" more, and the noise will therefore be more visible.
If you are talking about pixel quality, then the DX and FX modes are of course identical.
Cheers,
Bernard
[a href=\"index.php?act=findpost&pid=216166\"][{POST_SNAPBACK}][/a]
As I recall, Thom Hogan wrote that there is a noise problem with both the DX and 5:4 crop modes on the D3.
-
As I recall, Thom Hogan wrote that there is a noise problem with both the DX and 5:4 crop modes on the D3.
[a href=\"index.php?act=findpost&pid=216186\"][{POST_SNAPBACK}][/a]
Tony,
Ah, OK. Sorry, I wasn't aware of that. I really haven't used the DX mode of the D3 a lot (don't see much value to it), but never noticed any problem.
Regards,
Bernard
-
It may be due to the kicking-in of the electronics required to 'crop' the sensor (for lack of a better term) for DX or 5:4 (boosting the denominator of the S/N ratio).
That and/or an effect the cropping mechanics has on the effectivity of in-camera NR (vs 24x36 FF)
Tony,
Ah, OK. Sorry, I wasn't aware of that. I really haven't used the DX mode of the D3 a lot (don't see much value to it), but never noticed any problem.
Regards,
Bernard
[a href=\"index.php?act=findpost&pid=216194\"][{POST_SNAPBACK}][/a]
-
It may be due to the kicking-in of the electronics required to 'crop' the sensor (for lack of a better term) for DX or 5:4 (boosting the denominator of the S/N ratio).
That and/or an effect the cropping mechanics has on the effectivity of in-camera NR (vs 24x36 FF)
[a href=\"index.php?act=findpost&pid=216222\"][{POST_SNAPBACK}][/a]
I heard back from Thom this morning, and he wrote:
"I saw additional hot pixel production and a slight buildup in noise on my original D3. Curiously, I don’t see it now with updated firmware."
It appears that the problem has been resolved.
-
The cropped FF cameras have at least as bright future beside FF, as the FF cameras have beside MF backs. I don't see any reason to make a sharp division between the sensors of P&S and the MFDBs (fakes all together) just through the middle of the 1.5x or 1.6x croppers.
In fact, I have *never* seen a solid reasoning, how the problem of the lack of overall sharpness could be solved on FF, as opposed to the cropping cameras using the cream of the lens.
[a href=\"index.php?act=findpost&pid=216147\"][{POST_SNAPBACK}][/a]
I agree there's no doubt that the cropped format is generally sharper at the edges when using lenses designed for full frame. This has always been one of the two main advantages of APS-C, the other being the longer reach with telephoto lenses.
As the pixel density of full frame models reaches that of the cropped format, the 'longer reach' advantage no longer applies. In fact, in one sense, the full frame with equal pixel density begins to have the advantage when using long telephoto lenses. It's much easier to locate your target (a small bird in the foliage) when using (for example) a 400mm lens as opposed to a 640mm lens.
However, the greater resolution fall-off at the edges with 'full frame' might not be as significant as you think. Whilst every part of the composition is important to some degree, the corners and edges tend to be away from the central focus of interest. But I admit, there are always exceptions.
The other factors affecting edge resolution are the quality of the lens and the choice of aperture. For equal DoF, you are using the equivalent lens with APS-C at one stop wider aperture, or more. Edge resolution is generally worse the wider the aperture.
Consider also the generally mediocre performance of Canon wide-angle lenses compared to their better lenses. Whilst it is true you could expect better edge performance from a 50/1.4 on a 40D at F8, than you would get from the same lens on a 5D at the same aperture, this is not a sensible comparison. The composition, FoV and DoF would be different.
The comparison should be between a 28mm (or 30mm) lens at F5 (on the 40D) and the 50mm lens at F8 on the 5D. Then, the differences in edge sharpness between the two images might not be significant. It would be interteresting to do the test.
-
However, the greater resolution fall-off at the edges with 'full frame' might not be as significant as you think. Whilst every part of the composition is important to some degree, the corners and edges tend to be away from the central focus of interest. But I admit, there are always exceptions.[a href=\"index.php?act=findpost&pid=216347\"][{POST_SNAPBACK}][/a]
Ray, I don't mean to be argumentative, but I think the edges of the frame are more important than you are suggesting -- at least to me.
(http://photos.imageevent.com/tonybeach/mypicturesfolder/january2008/websize/C1-2_AB04869_web.jpg)
(http://photos.imageevent.com/tonybeach/mypicturesfolder/march2008/websize/_AB20273_web.jpg)
(http://photos.imageevent.com/tonybeach/mypicturesfolder/july2008/websize/C1_1A58954%20web.jpg)
(http://photos.imageevent.com/tonybeach/mypicturesfolder/august2008/large/C1_1A59817%20web.jpg)
The other factors affecting edge resolution are the quality of the lens and the choice of aperture. For equal DoF, you are using the equivalent lens with APS-C at one stop wider aperture, or more. Edge resolution is generally worse the wider the aperture.
As has already been discussed, shutter speed is a factor. If you are stopping down one stop more, you are either losing shutter speed or increasing ISO -- it's a trade-off and not nearly as clearcut an advantage for the larger format as many assume.
-
Ray, I don't mean to be argumentative, but I think the edges of the frame are more important than you are suggesting -- at least to me.
As has already been discussed, shutter speed is a factor. If you are stopping down one stop more, you are either losing shutter speed or increasing ISO -- it's a trade-off and not nearly as clearcut an advantage for the larger format as many assume.
[a href=\"index.php?act=findpost&pid=216374\"][{POST_SNAPBACK}][/a]
Those are stunning images, Tony. I agree, there are exceptions. Edge resolution is defintely important in some situations and my main gripe with the Sigma 15-30 zoom is that it is plainly not sharp in the corners, even at F16, using the 5D.
However, I doubt that the Canon EF-S 10-22 is sharp at the corners either.
Below is an image I took a few weeks ago on a trip down the Daintree River, North Queensland, using the 40D. Lens was the Canon 100-400 at 400mm and F8. Shutter speed was 1/500th and ISO 1600.
Notice how sharp the corners are .
[attachment=8008:attachment]
-
Tony
Love the black and white; biased opinion (of course) but still think it jumps off the screen. In a good way.
Rob C
-
Edge resolution is defintely important in some situations
I admit, there are always exceptions
Ray, I don't know your world, but in my one corner to corner sharpness is a requirement in the normal case.
I doubt that the Canon EF-S 10-22 is sharp at the corners either
Below is an image I took a few weeks ago on a trip down the Daintree River, North Queensland, using the 40D
I don't understand what you want to communicate. The issue was, that FF lenses, which are not sharp edge to edge on FF, are sometimes stellar on cropping cameras.
You have just demonstrated, that the above consideration is valid (although I do not see that as an argument for the continued existence of cropping cameras).
The comparison should be between a 28mm (or 30mm) lens at F5 (on the 40D) and the 50mm lens at F8 on the 5D
It is amusing, how you are trying to transform everything on some strange ways in order to make a comparison for an outcome to your expectation. I am waiting for your way of turning a painting into something in order to prove, that a P&S is better for landscapes than Monet.
-
Being a Kentucky transplant, married to a Kentucky woman and all as I am, I am not averse to horseflesh (no wife jokes please); except when the pummelled carcass has begun to ripen, as perhaps it has done with this thread....
Geesh folks. Do you not all have jobs or something?
-
Being a Kentucky transplant, married to a Kentucky woman and all as I am, I am not averse to horseflesh (no wife jokes please); except when the pummelled carcass has begun to ripen, as perhaps it has done with this thread....
Geesh folks. Do you not all have jobs or something?
[a href=\"index.php?act=findpost&pid=216484\"][{POST_SNAPBACK}][/a]
Thanks Mike,
I was going to suggest a forum specifically designated for the aggrieved but lacked the nerve. And again, I'm very appreciative. I've been able to gleen a beginner's understanding of sensors and of the relationship between sensors and the various uses of lenses. Best of shooting to all.
Peter
-
Ray, I don't know your world, but in my one corner to corner sharpness is a requirement in the normal case.
Gabor,
As one who has taken many thousands of shots with cropped format DSLRs before using a 5D, it's my experience that the advantages of full frame, compared with the cropped format, outweigh the disadvantages, although I accept that the 40D is better than the 5D in some respects because it's new technology.
I recall only one occasion when I returned to a scene to retake the shots with a 20D as a result of the 5D shots being spoiled due to vignetting and poor resolution at the edges. On that occasion I'd used a TS-E 24mm for stitching purposes and had ignored the red marks on the lens which warn of unacceptable sharpness when shifting too far.
I don't understand what you want to communicate. The issue was, that FF lenses, which are not sharp edge to edge on FF, are sometimes stellar on cropping cameras
That's true. My point is, the lenses that are not acceptably sharp from corner to corner tend to be wide-angle lenses used at wide apertures. Wide angle capability also happens to be the main disadvantage of the cropped format, so one advantage is somewhat mitigated by another disadvantage.
It is amusing, how you are trying to transform everything on some strange ways in order to make a comparison for an outcome to your expectation. I am waiting for your way of turning a painting into something in order to prove, that a P&S is better for landscapes than Monet.
If I've written anything in this thread which you think is not factual, or at least a reasonable opinion, please mention it specifically. Perhaps I can learn something.
-
As the pixel density of full frame models reaches that of the cropped format ...
Is that happening? Is it going to happen?
What I see over the last five or six years is little or no change in the pixel spacing ratio, with roughly a doubling of pixel count in a wide variety of SLR formats:
- 4/3 from 5MP then to 10MP (and that was reached 18 months ago while APS-C was still at 10MP, so another bump is probably due);
- APS-C from 6MP then to 10MP and 12MP (and up to 14.6MP from Pentax/Samsung);
- 35mm from 11MP (and Kodak's 13.5MP) then to 12MP and 22MP (and soon Sony's 24.7MP).
So for sensor resolution in l/mm as indicated by pixel spacing, at best the highest resolution 35m model at a certain time matches most but not all APS-C models (like the Kodak 14/N and 6MP DX models both at 8 microns.)
Actually one gap is closing in the opposite sense: the gap in pixel count between the highest in APS-C models and that of the lower priced 35mm models. That gap is currently "negative" due to a push in the low light performance direction in 35mm format.
One sensor resolution gap that shows no sign of disappearing is that between 35mm and 4/3; these very different formats seem to serve quite different purposes, especially when it comes to telephoto lenses.
-
Is that happening? Is it going to happen?
What I see over the last five or six years is little or no change in the pixel spacing ratio, with roughly a doubling of pixel count in a wide variety of SLR formats:
[a href=\"index.php?act=findpost&pid=216573\"][{POST_SNAPBACK}][/a]
Good point! This has certainly been the case with Canon, although the time frame between the introduction of the D60 and the 450D (representing a doubling of pixel count) is slightly greater than the time frame between the introduction of the 1Ds and 1Ds3 which also represents a doubling of pixel count (almost).
The incremental jumps in pixel count from Canon I find a bit annoying, especially when they appear so close together, as with the announcement of the 450D not long after the 40D was available in the shops.
Perhaps Nikon will change this paradigm. Is it not a fairly safe bet that the successor to the D3 will be 24mp? That respresents not quite the same pixel density as the D300, but as close as matters. I don't think anyone could reasonably quibble about the difference between 24mp and 27mp. One has to remember that the Nikon DX format is slightly larger than the Canon cropped format.
One should also bear in mind that as pixel counts increase in both formats, such increases become less relevant without a corresponding increase in lens quality.
I think the survival of the cropped formats will therefore depend on the development of DX and EF-S lenses which actually do have a higher MTF response than most FF 35mm lenses, as is the case with the Zuiko lenses.
I'm very impressed with the quality of the Canon EF-S 17-55/2.8, except for its autofocussing accuracy.
-
Perhaps Nikon will change this paradigm. Is it not a fairly safe bet that the successor to the D3 will be 24mp? That represents not quite the same pixel density as the D300, but as close as matters.[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=216601\")
Is the 1Ds MkIII the successor to the 1D MkIII? The D3 competes with the 1D MkIII; the inevitable Nikon 24MP FX DSLR (which we have good indications of based on an accidental firmware upgrade released for the D3 that had JPEG quality settings based on 24MP files) will compete with the 1Ds MkIII and the Sony A900; it will not replace the D3, instead the "D3s" will replace the D3, but I'm fairly certain they will both use the same sensor.
I expect that this hypothetical 24MP FX DSLR will undoubtedly cost $5000 or more -- after all, Nikon isn't going to sell it for less than the D3 -- so this is not the camera I'm waiting for unless my finances improve dramatically. No, I'm waiting for another year or so when I expect a consumer version will appear, not unlike what the D700 represents vis-a-vis the D3 -- Nikon (and Canon) will be compelled to release lower cost high MP DSLRs in order to compete with Sony's less expensive A900 which is currently projected to cost $3000.
The problem with resolution using these high pixel pitch DSLRs will not be the lenses, my Nikkor 14-24 does very nicely on my D300 and I'm sure it will do just as nicely on a high MP FX DSLR that has slightly larger pixel pitch (edge performance notwithstanding). Likewise, the Nikkor 24-70 has very impressive MTF numbers and I'm sure Nikon designed these lenses with a higher MP future in mind. The problem is that you will not see any improvements in resolution on a 24MP FX DSLR past f/11 over a 16 MP FX DSLR due to [a href=\"http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm]diffraction limitations.[/url] For landscape and product photography this will be a huge barrier and will not require higher MTF then many top end Nikkor lenses can already deliver, it will require lenses that have tilt capability in order to expand the DOF. Not coincidentally, Nikon has introduced three lenses with this soon to be essential tilt capability.
-
Is the 1Ds MkIII the successor to the 1D MkIII? The D3 competes with the 1D MkIII; the inevitable Nikon 24MP FX DSLR (which we have good indications of based on an accidental firmware upgrade released for the D3 that had JPEG quality settings based on 24MP files) will compete with the 1Ds MkIII and the Sony A900; it will not replace the D3, instead the "D3s" will replace the D3, but I'm fairly certain they will both use the same sensor.
I'm looking at this from the aspect of format size. Canon have doubled the pixel count of their full frame sensors over a period of approximately 6 years.
Nikon seem about to double the pixel count of their FF sensors a year or so later than their first FF model was announced, perhaps a bit longer. I've got no idea if they will duplicate that (no doubt expensive upgrade) with a cheaper version like the D700, but it wouldn't surprise me if they were to.
I'm hoping that the Sony 24mp version will be the budget model because I already have a few Minolta lenses, but I don't have any Nikon lenses, so that's a problem for me.
The problem with resolution using these high pixel pitch DSLRs will not be the lenses, my Nikkor 14-24 does very nicely on my D300 and I'm sure it will do just as nicely on a high MP FX DSLR that has slightly larger pixel pitch (edge performance notwithstanding). Likewise, the Nikkor 24-70 has very impressive MTF numbers and I'm sure Nikon designed these lenses with a higher MP future in mind. The problem is that you will not see any improvements in resolution on a 24MP FX DSLR past f/11 over a 16 MP FX DSLR due to diffraction limitations. (http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm)
As I understand, sensors with a higher pixel density but equal pixel count compared with a larger format, always need a sharper lens in order to deliver the same image detail.
When maximum DoF is sought and one is shooting at apertures that begin to be noticeably affected by diffraction, it's not an issue. Any good lens at F16 with a 5D will deliver approximately equal total image resolution to an equivalent focal length lens on a 450D at F10. The 450D needs higher resolution from the lens, because of its greater pixel density, and it gets it, when used within the range of apertures affected by diffraction.
The problems begin when a wide aperture and shallow DoF are required for the composition. From the Canon perspective, you are then comparing (for example) a 450D with 50mm lens at F1.8, with a 5D and 85mm lens at F2.8. In these circumstances instead of the 450D getting the required higher resolution from the lens, that it needs, it's getting lower resolution.
-
I think the survival of the cropped formats will therefore depend on the development of DX and EF-S lenses which actually do have a higher MTF response than most FF 35mm lenses, as is the case with the Zuiko lenses.
[a href=\"index.php?act=findpost&pid=216601\"][{POST_SNAPBACK}][/a]
Indeed! My guess is that eventually lens resolution will be the main limit to the resolution in any given format, so it comes down to the extent that lens designs scaled down from 35mm format by 1.5x, 1.6x or 2x can reduce the length scale of aberrations and so increase l/mm of resolution. There does seem to be some l/mm advantage to smaller format, smaller image circle lens designs.
And as we talk, there is a leak from the Canon China web-site of Canon going to 15MP with a new EF-S 50D body. So we will soon have 14MP+ from Canon, Sony and Pentax, and I am predicting 14MP+ in the first Micro 4/3 bodies, to match up on MP with high end compacts.
To match the pixel pitch of that 50D's 15.1MP in 14.9x22.3mm, the bar has been raised for 35mm format to almost 40MP. And 14MP in 4/3 would push it to over 50MP.
-
IMO the survival of the cropping design has nothing to do with lens development, nor with pixel density. The different formats (mobile phone, P&S, cropping, FF, MFDB) represent
- different sizes and weights,
- different prices,
- different image qualities.
These are reason enough to have all of them around, until a large format camera will be as small and light as a credit card (including the lens or its equivalent, of course).
-
IMO the survival of the cropping design has nothing to do with lens development, nor with pixel density. The different formats (mobile phone, P&S, cropping, FF, MFDB) represent
- different sizes and weights,
- different prices,
- different image qualities.
These are reason enough to have all of them around, until a large format camera will be as small and light as a credit card (including the lens or its equivalent, of course).
[a href=\"index.php?act=findpost&pid=216673\"][{POST_SNAPBACK}][/a]
Gabor,
You've missed one important factor. The cropped format is not just a different format, it's a cropped format, the same as current MFDBs are cropped formats. They are in a different category to mobile phones, P&S and even Olympus 4/3rds.
In my view, the Olympus 4/3rds is the genuine, non-cropped format, with lenses to match, that can fill this niche. But it's okay with me if Canon and Nikon wish to continue competing with Olympus. I just wish that Olympus had struck up an agreement with Canon instead of Kodak. A Zuiko lens on a Canon cropped format body, would have been really something.
-
IMO the survival of the cropping design has nothing to do with lens development, nor with pixel density. The different formats (mobile phone, P&S, cropping, FF, MFDB) represent
- different sizes and weights,
- different prices,
- different image qualities.
These are reason enough to have all of them around, until a large format camera will be as small and light as a credit card (including the lens or its equivalent, of course).
[a href=\"index.php?act=findpost&pid=216673\"][{POST_SNAPBACK}][/a]
In practice you are quite likely right. The difference in lens resolution is just one most certainly persistent difference, along with other differences that are so far quite persistent: sensor and body cost, sensor resolution (l/mm again) and the minimum size and weight of bodies available in a given format. At least when one compares sufficiently different formats, which is why I often use the somewhat extreme example of 4/3 vs 35mm. Comparisons of relatively nearby formats like 4/3 vs EF-S are murkier.
But as you have probably noticed, there are many people who expect or at least hope for the current differences in sensor cost and resolution and body size and weight to be reduced to negligible levels once the 35mm DLSR makers work hard enough on it. It is harder to expect lens resolution differences to be eliminated, any more than it is likely that noise level differences at equal ISO speed for different pixel sizes will be eliminated.
Another question though is why SLR makers would want to work so hard on getting most or all of their customers to move to products that are inherently more expensive to make (i.e ones with larger sensors). The more common approach to increasing profitability is to get the mass market customers to accept products that are less expensive to make, and then pick the deeper pockets by also offering more expensive high end options at high profit margins.
P. S. To Ray: the distinction of EF-S and DX as "cropping" rather than just being a different, smaller format depends on how well suited the lens system available for them is. With a sufficient range of EF-S/DX lenses, along with longer focal length lenses that work quite well with both those formats and 35mm, "cropping" just become another word for "also compatible with some lenses designed for another, larger format". Pentax 35mm film cameras can also use Pentax MF lenses, but no-one calls them "cropping" on that basis. On the other hand, all currently available DMF systems are to varying degrees "cropping" as their lens systems have at best been minimally adapted to the FOV needs of sensor formats smaller than the MF film formats.
-
I just wish that Olympus had struck up an agreement with Canon instead of Kodak.
[a href=\"index.php?act=findpost&pid=216681\"][{POST_SNAPBACK}][/a]
If you had not noticed, Kodak has been dumped, and in fact that was already in progress with the design of the first consumer level 4/3 body, the E-300, which adopted the porro-prism VF needed for the subsequent rangefinder-shaped Panasonic/Leica models. Unlike Kodak, the electronics giant Panasonic seems to be becoming a roughly equal partner, and might even be the lead player with products like video-capable Micro FourThirds.
-
If you had not noticed, Kodak has been dumped, and in fact that was already in progress with the design of the first consumer level 4/3 body, the E-300, which adopted the porro-prism VF needed for the subsequent rangefinder-shaped Panasonic/Leica models. Unlike Kodak, the electronics giant Panasonic seems to be becoming a roughly equal partner, and might even be the lead player with products like video-capable Micro FourThirds.
[a href=\"index.php?act=findpost&pid=216736\"][{POST_SNAPBACK}][/a]
You are right. I haven't been following the corporate developments behind the 4/3rds system. My comment was directed at a perceived shortcoming of the Olympus 4/3rds system with regard to total image resolution and noise at high ISO.
If that latest rumour about the Canon 50D is factual, then such a camera with sensor cropped to a 4:3 aspect ratio and fitted with the Zuiko range of lenses would be superb.
-
P. S. To Ray: the distinction of EF-S and DX as "cropping" rather than just being a different, smaller format depends on how well suited the lens system available for them is. With a sufficient range of EF-S/DX lenses, along with longer focal length lenses that work quite well with both those formats and 35mm, "cropping" just become another word for "also compatible with some lenses designed for another, larger format". Pentax 35mm film cameras can also use Pentax MF lenses, but no-one calls them "cropping" on that basis. On the other hand, all currently available DMF systems are to varying degrees "cropping" as their lens systems have at best been minimally adapted to the FOV needs of sensor formats smaller than the MF film formats.
[a href=\"index.php?act=findpost&pid=216735\"][{POST_SNAPBACK}][/a]
There is also a distinction to be made between a format that can incidentally accept lenses designed for a larger format, and a format that, at least initially, totally relies upon lenses designed for the larger format. It took a while before Canon brought out their first EF-S lens and it wasn't a particularly good lens.
We should not forget that the difference in sensor area between the Canon cropped format and 35mm is almost as great as the difference in area between 35mm film and the smallest MF film, 6x4.5 (comparing actual exposed film area).
I don't think that 35mm film for still photography would have been so successful if the users had to rely upon lenses designed for 6x4.5.
If Canon and Nikon were to design top quality lenses for their cropped formats, which were actually sharper (on average) than the full frame equivalent focal lengths, then the 'cropped format' system would become almost as expensive as a full frame system, for the serious photographer. It's the lenses which are the major cost. As I recall, my 17-55/2.8 zoom cost as much as the 40D body.
In fact, for the serious photographer, it's the availability of good lenses which is a major drawcard when choosing a system, at least for me.
-
... a format that, at least initially, totally relies upon lenses designed for the larger format. It took a while before Canon brought out their first EF-S lens and it wasn't a particularly good lens.
[a href=\"index.php?act=findpost&pid=216784\"][{POST_SNAPBACK}][/a]
Agreed about the early days of DSLRs: clearly the early DSLRs were all cropping, and even the first few DX, DA and EF-S lenses did not change that much. But today the measure is the adequacy of current lens offerings. I would rate EF-S as quite good for most of the amateur market (but with weaknesses like no 180º fish-eye and a gap in suitable f/2.8 zoom lens coverage between the 17-55 and 70-200); Nikon DX is maybe a bit more complete, but still not at the level of what is available for FX format (e.g. no f/2.8 wide zoom, and that same gap between f/2.8 zoom options).
In a sense the Pentax DA format system is the most complete and self-sufficient APS-C lens system, as Pentax has realigned almost all of its lenses to the format, including the pairing of 16-55/2.8 DA and 55-135/2.8 DA, and recently even long focal length Pentax lenses have all been designed specifically for DA format.
-
I do not agree that 'all' full frame DSLR sensors, in themselves, are less sharp at the edges - Nikon's D3 sensor technology has proved this. This notion is still banded about as Canon sensors do take a nose dive at the edges of the frame, even when using prime telephoto lenses.
[a href=\"index.php?act=findpost&pid=216878\"][{POST_SNAPBACK}][/a]
Nick,
You've raised some interesting concerns about the edge performance of the sensor itself. After perusing the real-world MTF charts at Photodo it becomes very clear that just about all lenses take a dive in MTF response towards the corner of the full 35mm frame, but not necessarily at the middle of the short edge, which is only 18mm from the centre. Some lenses are still very good up to 18mm from the centre, at least at F8. The Canon 50/1.4 is one such lens.
I recall some months ago there was a thread comparing the resolution of the 20D with the 40D, using the same 50/1.4 lens supposedly focussed on the same spot in the same scene. I've failed to find the thread after a few searches, but as I recall there was a strange anomaly in the test results. Although focussing appeared to be the same in both images and over-all resolution was very similar, one image was clearly less sharp at the edges, but I can't remember which camera was soft at the edges.
For some time I've been casting an envious eye at a D700/14-24mm combination, but have held off because I've got an order in for a Canon/Nikkor adapter.
I now wonder if the excellent corner to corner resolution of this lens might be compromised with a 5D sensor.
-
I dont think sensors ar less sharp at the edges than in the center. I suggest you try to mount an old Hasselblad medium format lens to your DSLR with an adapter. All over the image is not sharper than a 35mm lens, though you most likly dont see the same sharpness fall off with the MF lens.