Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: Richowens on August 23, 2005, 07:46:58 pm
-
Jani,
Please explain your math for this, I don't follow your formula.
According to my "wetware" calculations, Pom is correct.
The 1.6 is simply a factor, not a measurement.
I could be wrong on this, the "wetware" is a little dry from age.
::
Rich
-
The cropped resolution is the same as aperture stops. Your taking a linear measure and extending it in two dimensions to get area. It's just that now we've got 1.6 instead of 1.4.
John
-
I'm still confused about the original subject.
Isn't the extra "reach" on a 1.5x sensor one of their biggest advantages (in certain situations)?
The extra "reach" on APS-C sized sensors comes from the desire to enlarge pictures taken with those sensors to the same sizes as that of 135-sized sensors.
Basically, you're changing the enlargement factor.
And yes, you're often helped along by a greater pixel density.
-
For another, it seems that few 35mm format lenses have enough resolution to exploit the potential resolution of a 30MP sensor; at least, probably not many zooms.
And even fewer lenses that can exploit the potential of a 30MP sensor in that part of the image greater than 14mm from the centre of the lens' image circle. However, as technology progresses, it is reasonable to presume that better lenses will eventually become available at an affordable price. (Maybe when the Chinese start making such lenses ). We know that Canon already makes such lenses. The expensive 200/1.8 has an impressively flat MTF response right out to the edges, at f4 as well.
Fields like low light and high speed action photography are likely to be best handled by DSLR's with somewhat smaller sensors, smaller pixels, and shorter lenses. Canon clearly thinks so: its two high frame rate DSLRs are 1.3x and 1.6x.
The introduction of the 5D could lead one to think that this is the beginning of the end for the cropped 35mm format. Whilst it's true there's a significant weight saving by using EF-S lenses on such cameras, there's no advantage in telephoto reach with standard EF lenses once the pixel density of the full frame reaches that of the cropped frame, as it's already done comparing the D60 with the 1Ds2, and comparing a future 25-30MP FF with the 20D.
There's a terrible dilemma here for both Canon and the consumer in view of the fact that the first film based APS format never caught on. Most people I know who are photography enthusiasts lust after cameras with the higher potential image quality. If they settle for something less, it's usually because they can't afford it, or justify the price, or because the equipment really is too bulky, too heavy and difficult to operate.
This has traditionally been the dichotomy between 35mm and Medium Format. The former has always been very attractive precisely because of everything that MF lacked; affordability, compactness, lightness, versatility, ease of use and an extraordinary degree of automation and choice of lenses.
MF essentially had just one advantage; superior image quality. (Not quite true I know. Being able to compose the image using both eyes and seeing outside the frame was also an advantage.)
It looked as though this dichotomy might have been preserved as film processes were gradually replaced, the APS-C format DSLR taking the place of 35mm film and FF 35mm taking the place of MF. It all sounds very plausible and reasonable and I believe Michael R also reinforced this view with earlier comments to the effect that FF 35mm DSLRs would always remain an expensive item aimed at the professional and well-heeled amateur.
Well, it's not plausible and reasonable to me. We're heading for exactly the same scenario that existed with film based APS and 35mm. The two formats are too close to be separate species.
-
What lenses do you think are missing for EF-S mount that are relevant to more than a small fraction of SLR users?
BJL,
Sorry! You've just shot yourself in the foot. :D
For the APS-C format to survive it needs more than a small minority of adherents. It has to be a full-system alternative to full frame 35mm with a range of high quality zooms and primes. We could start off with a standard EF-S 28mm F1.4 prime of better quality than the Canon 50/1.4, or a 20mm F1.4 of better quality than the Canon 35/1.4, or an EF-S 400/2.8 of better quality than the EF 600/4.
The range and quality of lenses available for any format is a major part of that format's success. The APS-C DSLR format is at present riding piggy back on the lenses of a larger format.
-
Ray,
of course sensor prices will come down a bit more, but what I seriously doubt is the dream of price reductions comparable to what is generally seen in electronics, because, as has been said many times, they depend to a large extent on reducing the size of electronic components: that size reduction is most of what drives "Moore's law", and is of course irrelevant when chip size is fixed at 24x36mm. Experience from the semiconductor industry is that prives for chips of a certain size plateau once production volume reaches a high enough level to produce good economies of scale, they do not keep getting sgnificantly lower.
That leaves mainly economies of scale, with a major thresholfd being making and selling enough of a chip to run a large, efficient chip fabrication line full time producing that one chip. This is not the case with low volume products like the 1D and 1Ds models, but it seems like that Canon is aiming at this sort of volume with the 5D.
If indeed the 5D reaches this production level, it has more or less reached the plateau; if you wish to predict differently, tell me what you know about the ecomomics of producing chips of a given large size.
-
- most photographers have price or weight limits which mean that where 200/2.8 is affordable/carryable with APS-C format, 35mm format will be limited to 300/4.
Not quite true. If we imagine a time when pixel density for both formats have reached a plateau and are at a level that makes the AA filter redundant, then a FF 200/f4 will do everything a DX or EF-S 200/2.8 can do and more, except it will be one stop slower. In situations where the FF 200/f4 is not long enough and the image has to be cropped to the same size as the APS-C image using the 200/2.8 lens, then resolution will be the same but the smaller format has that one stop advantage.
Both lenses will be about the same weight, but the FF 200/f4 will provide significantly better image quality if the image is not cropped. And of course (sorry for the dig :D ), if we are comparing an Olympus 4/3rds format to FF 35mm, then that 1 stop speed advantage will be completely eroded by the need to use a lower ISO setting to reduce noise on the Olympus.
Finally, diffraction is of no practical relevance to DSLR lens speed needs (it might be with the far higher sensor resolution needs of compact digicams.)
I wasn't making a distinction here between any particular format, rather just making an observation about the shift in the larger number f stops towards smaller f stops as the format decreases in size, with typically a smallest aperture of f90 for 8x10" format lenses, f32 for 35mm and f8 for the better quality P&S cameras. These f stops are smaller, the smaller the format, and the lenses are faster at these f stops for a given FOV and DoF as a result of diffraction limitations on image quality.
-
As a reality check, who has seen D2X prints, or even 1Ds prints, that reveal the resolution limits of their sensors? How large were those prints, and how close did you have to go to see the problems? (No test pattern viewing please!)
I might have in some A2 images, but that is far from obvious.
Although some images feel like it could have additional detail when compared to an Imacon scanned 4*5 slide, most of them feel very solid when viewed alone, even from closer than should reasonnably be done.
Then you have cylindrical or flat stitching to come to the rescue...
Regards,
Bernard
-
This is from Michael's comments about the 5D and natterers and naysayers and such.
"They'll also carry on about how 1.5X and 1.6X cameras give you greater reach with telephoto lenses (ya, right) ."
The comment in parens and italics sound dismissive, at least to me, of either the concept or the people expressing it.
I may well end up with the 5D for a variety of reasons, but I will surely continue to use the 20D for birding for a while longer. It has nothing to do with the size of the sensor and everything to do with the number of pixels per FoV, and the 20D remains at the top of that Canon foodchain. It is not "magical thinking" (or even wishful thinking) to say that the 20D has the highest linear resolution of the Canon chips - so Yes, it DOES outresolve the 1Ds2 in that regard ). In many situations, the bird is far enough away that it doesn't fill the frame even with my "longest" glass, so I'll be cropping and the more MP/FoV I can get, the better image I end up with - and I print quite a few. That long glass is heavy (500/4 + monopod or tripod/Sidekick) for hiking and a 1 series would add even more load. Sure, I'd love better AF and a brighter VF, but not at the expense of image size and backache! When the 1Ds2 was imminent, I had decided that if it were 22MP (i.e. a scaled up 20D chip at the same density) I'd have everything I wanted in one camera and a portable gym to boot. Maybe the 1Ds3 will be that camera. However, I'm quite good at the combo of AF and MF with the 20D VF so that the 45pt wouldn't improve my picture quality or keeper ratio much. (The 22MP also started to sound good enough (with a 4 shot stitch) to compete with my 4x5, but that's another story)
So, Michael , what exactly does (ya, right) mean? I know there are a lot of birders/wildlifers who like the MP/FoV of the 20D and who live with its limitations. Are we stupid or misguided or just have different priorities/goals - or am I really missing something????
Thanks
Andy
-
If you crop the picture from the 5D by 1.6 (like the 20D sensor) then you get exactly 8 megapixels. In other words you can have the best of both worlds.
No, you didn't do your math right.
While 12.8 Mpx divided by 1.6 is 8 Mpx, that's not how it works.
If you use the crop factor correctly, you should divide by 1.6*1.6, or 2.56.
That yields 5 Mpx.
-
Hi all,
Disregard the last post. The "wetware" is getting a little dry.
Thanks, Jdemott. I just went into PS and opened one of my 6 megapixel files, changed the width from 2000 to 3000 and guess what, 13.5 megapixels. Sometimes I'm just a little dense.
Thanks
::
Rich
-
The image is not cropped in the same way that you crop an image in an image editor by literally throwing away pixels.
Yes it is, if you are comparing two different formats with the same pixel density, such as the Canon D60 and 1Ds2.
Shoot the same scene with both cameras, from the same position using the same lens. The D60 image should be identical to the 1Ds2 image cropped in an image editor to the same size as the D60 image, ie. by literally throwing pixels away.
Note: I don't mean identical with regard to every characteristic, such as color hue and balance, and noise at high ISO etc., but identical with respect to resolution at base ISO.
-
I'm still confused about the original subject.
Isn't the extra "reach" on a 1.5x sensor one of their biggest advantages (in certain situations)?
Yes, essentially it is. There have always been technical arguments about whether or not there are "real" advantages to the crop factor vis a vis cropping to the same frame from a full frame sensor capture, but the bottom line is that with today's equipment there are indeed "real" advantages.
You will hear all types of technical discussions concerning pixel pitch, optical losses through using only part of the image circle and a host of optical physics discussions, but when you wade through it all and get to the kernel inside the nutshell you are left with greater enlargeability because in the end you are sampling with the full complement of sampling sites within a reduced (compared to 35mm frame size) field of view.
With present and soon to be released full frame sensors you must use longer focal length lenses to get the same enlargeability for that portion of the frame captured by the 1.5x/1.6x crop factor sensors. As an example, this means that the user who wishes to get similar telephoto results to what they might get with a Canon 20D (8 megaixel, 1.6x crop factor) by using a 16.7 megapixel 1DS Mark II must use a longer focal length lens which makes up the difference between about 6.68 megapixels and 8 megapixels or about 20 percent longer focal length. To put this into common perspective what you get with a 400mm frame with the 20D would take a 480mm lens to achieve with a 1DS Mark II.
The differences are much more revealing when you consider cameras like the 12 megapixel Nikon D2X and 16.7 megapixel Canon 1DS Mark II. Since the Nikon has a 1.5x crop, the Canon must make up about a 3.6 megapixel loss at the same field of view frame which is about a 45 percent difference. So what you would get with a 400mm lens in terms of the number of pixels on the subject with the Nikon would take 580mm with the 1DS Mark II, essentially requiring carrying a heavy 600mm lens with tripod, head, etc., versus a much lighter and smaller 400mm lens which could be hand held.
As someone who uses both full frame and crop factor sensors, I would never give up my crop factor sensors for full frame for wildlife use unless or until a full frame is available which has sufficient resolution to provide equal pixel count as my crop factor at the same crop percentage. I'm not holding my breath until this happens.....
Lin
-
Thanks Ray for moving this away from the marketing jargon of pixel counts to more traditional measures of sensor resolution, lp/mm, which can be roughly measured by pixel spacing. Quite simply, the relatively large pixel spacing of the 5D (over 8 microns) means that it can be expected to have lower resolution (lp/mm) than just about any other current DSLR other than the 1DMkII. Thus, to get an image of a given subject at a given distance with a given amount of detail (roughly, a given pixel count after cropping) the 5D and 1DMkII need to use longer focal lengths, while the D2X and E-300 need the shortest amongst DSLRs, and compact digicams are way ahead on "telephoto reach per mm of focal length", so long as you do not need a lot of speed.
Perhaps Michael was thinking of the fact that pixel size is not strictly tied to sensor size, so that in principle Canon could match the sensor resolution (lp/mm) of the D2X or E-300 in a 35mm format sensor of about 30MP. In practice, such a sensor is a long way off, if it ever comes. For one thing, digital signal processing speeds cannot yet handle 30MP at a reasonable frame rate: the 5D is already down to 3fps compared to the 20D's 5fps, probaly due to using the same DSP chip. For another, it seems that few 35mm format lenses have enough resolution to exploit the potential resolution of a 30MP sensor; at least, probably not many zooms.
The strengths and weaknesses of the 5D seem likely to reinforce the trend that larger-than-mainsteam digital SLR formats having a similar role to larger-than-35mm film SLR formats: mostly, better imaqe quality (dynamic range, resolution etc.) when used at low ISO speeds, while being close to parity with smaller formats when used with higher ISO settings and telephoto focal lengths.
Fields like low light and high speed action photography are likely to be best handled by DSLR's with somewhat smaller sensors, smaller pixels, and shorter lenses. Canon clearly thinks so: its two high frame rate DSLRs are 1.3x and 1.6x.
-
I am puzzled why some people ignore all trends in order to cling to the the belief that larger DSLR formats will someday offer something for nothing: sensors of more than twice the area, with the same pixel pitch and so more than twice as many photosites, and lenses with more than twice the image circle area yet with the same resolution (lp/mm), all at no significant price increase.
JBL,
No. That's not what I'm saying. There will always be a price difference between the larger format and the smaller format for equal pixel density.
I'm saying that the reason I don't currently own a full frame DSLR is the same reason that 10 million other photographic enthusiasts don't own one. It's not because my D60 and 20D are better tools for my purposes, although, if I were a keen bird photographer, that would be a reason, it's principally because the 1Ds2 costs as much as a small motor car.
Drop the price to something reasonable and I'll buy one for the sake of the higher over all picture quality. If I have a choice between a $500 future cropped format and a $1500 future 25MP FF 35mm, I'll definitely choose the latter, and I bet I'll not be the only one.
Notice that the price differential ($500 as opposed to $1500) in my example, is actually greater than the current price difference between the 20D and the 5D, although the resolution of the 20D is greater in terms of lp/mm. However, at some point, little purpose will be served by increasing pixel density. The cropped format will likely reach that useful limit before FF and then will have nowhere to go with regard pixel count.
If the cropped format is to survive, Canon will have to provide a full range of high quality EF-S lenses. I'm not convinced the public at large will want to commit themselves to investing in a range of lenses that cannot be used on a FF body when there's an expectation and a clear trend that's already in place, for full frame 35mm to become affordable.
-
- the price difference between the 20D and 5D is about US$1800, well over the $1000 in your $500 vs $1500 comparison. Sensor price difference seems to be the main factor in this US$1800 difference.
BJL,
You're old enough to remember the production costs of the first audio CDs and the quality control problems that, at the time, were beyond the manufacturing capabilities of Philips.
There should be no doubt that whatever the costs and difficulties that are currently associated with 35mm sensor production, they will prove to be a temporary state of affairs. That's the nature of technological progress.
You seem to be premising your arguments on the current situation rather than current trends.
I used the term 'price differential' to indicate a percentage difference in price rather than an absolute difference. In any case, the 'value' of items is generally perceived in terms of the price of alternative 'desired' items. Automobiles and houses are major and necessary purchases in life. The price of luxury items (for the non-professional) such as cameras have to be brought into perspective. If I need a new car, then I can't afford a 1Ds2.
Those who are not photography enthusiasts are amply served with a plethora of P&S digicams. Those who are photography enthusiasts will gravitate towards the camera that produces the higher image quality. It's really quite simple.
Compare a 200/2.8 to a 300/2.8 for price (and weight), for example.
There's a weight advantage in this example that is offset by an image quality disadvantage. The smaller format generally has poorer performance at higher ISOs. The lenses need to be faster, not only to overcome higher noise but also diffraction. There's no ultimate advantage here except weight and there's an unavoidable, ultimate image quality disadvantage. Image quality usually takes precedence over weight, provided the wieght difference is not very significant, and sometimes even if it is significant.
The weight difference between my Sony DSC T1 and my 20D is actually very significant. I bought this extremely compact camera because I could carry it around everywhere. But I don't use it nearly as often as I anticipated because the ultimate image quality and noise factor is significantly worse than my 20D.
Presumably your prediction about the smaller DSLR formats (please, can we avoid the inaccurate and derogatory term "cropped format"?) needing "a full range of high quality EF-S lenses" only refers to Canon's EF-S mount DSLRs; the survivial of DX format, Four Thirds, etc. do not rely on EF-S lenses!
The term 'cropped format' is not derogatory but precisely meaningful and provides an indication as to why it will probably not survive. A 20D with an EF-S lens is an APS-C format. No cropping involved. However, a 20D with an EF lens is indisputably a cropped format.
Will the Nikon D2x continue the line? Probably not. 12MP on that size sensor is getting close to the useful limit. When the limit is reached and higher pixel count full frame cameras offering greater over all picture quality become available at a similar price, even though slightly higher price, Nikon will be forced to go full frame.
Economics is about adapability and versatility. Rigid policies don't work.
-
Your assumption has at least one fundamental flaw.
The same advances that will make 35mm sensor production easier/cheaper will also benefit smaller sensors.
Jani,
Maybe my assumption has at least one major flaw, but you haven't explained it to me yet.
I don't buy a camera because it is cheap but because I want what it can do, because I intend to use it and because I consider what it can do to be relatively good value at a certain price point. I'm image quality orientated, as are many others on forums such as this. Present me with a choice of a 20D at $1500 or a 1Ds2 at $8000 and I'll reject the 1Ds2 because of its insane price and choose the 20D. Cut the price of both cameras by 2/3rds, $500 for the 20D and $2600 for the 1Ds2, then I'll choose the 1Ds2. Where's the flaw in that?
-
Ray, your post (see jani's quote) seemed to imply that you believe the prices of Full Frame will come to be close, perhaps not the same, but close to the same price for Reduced frame.
I believe that this is what jani is responding to.
----
My 2 cents on ISO performance: compared to film, digital is already pretty amazing. How great does it even need to be? :)
Great thread, good reading.
-
Compare a 200/2.8 to a 300/2.8 for price (and weight), for example.
There's a weight advantage in this example that is offset by an image quality disadvantage. The smaller format generally has poorer performance at higher ISOs. The lenses need to be faster, not only to overcome higher noise but also diffraction.
Ray,
firstly, given that smaller format SLR sensors will always be somewhat less expensive, it is the larger format which needs to establish a compensating advantage; it is enough for the lenses used with a smaller format to have rough parity in size/weight/cost/speed/DOF trade-offs.
Secondly, why did you ignore my mention of the 200/2 DX shortly afterwards? The scenario is this:
- most photographers have price or weight limits which mean that where 200/2.8 is affordable/carryable with APS-C format, 35mm format will be limited to 300/4. Conversely, for the tiny minority who can afford and carry a 300/2.8 for 35mm format, they can equally afford a 200/2 for DX. Either way, similar budgets get similar usable shuter speed and DOF, just with different combinations of f-stop and ISO speed.
This is because, as I think I have said before, it is the effective aperture diameter that determines light gathering speed (and thus high speed/low light image quality), DOF and diffraction, when photographing a given scene (same FOV, illumination, etc.). Effective aperture diameter is also a rough guide to front element size, and thus to lens size, weight and construction cost. I see little advantage to either larger or smaller formats in this respect; at least, not enough to cancel the smaller sensor's price advantage.
Finally, diffraction is of no practical relevance to DSLR lens speed needs (it might be with the far higher sensor resolution needs of compact digicams.) Surely you know that diffraction limitation only comes into play at f-stops around f/11 and smaller, and maybe from f/8 with the smaller pixels of the current DSLR sensor resolution leaders, the D2X and E-300. Even lenses limited to f/4 or f/5.6 (which covers almost every recent Canon lens) have no diffraction problems when used at their larger apertures.
-
The obvious reality is instead that this middle ground is now the most profitable sector of the digital camera market, serving the many photography enthusiasts who want a DLSR with interchangable lenses, greater speed that small digicam sensors can provide, and so on, and are looking for bodies at under $1000, or at most about $1,500 and use lenses not costing thousands of dollars, and so are far away from the 35mm format DLSR market sector.
The key word is 'now'. The middle ground is now the most profitable sector of the digital camera market. Have you noticed that situations have a habit of changing. As the Lord Buddha is reputed to have said, "Nothing is permanent in this world".
As prices of FF 35mm come down (no doubt gradually) and as peoples' desires and expectations shift ground, then FF 35mm is likely to become that profitable middle ground.
I've just been reading a review of the Olympus 8MP E-300 on dpreview. What seems quite amazing is that the image from the P&S Olympus C-8080 WZ, which has a tiny sensor just 8.8x6.6mm compared with the E-300's 17.3x13mm sensor which is almost exactly 4x the size, is actually slightly sharper. However, the C8080 is much noisier than the E-300 as one would expect. But interestingly, the E-300 is also much noisier than the 20D which is also slightly sharper than the E-300.
I would think that a P&S such as the C8080 with improved speed, bigger buffer to handle a number of RAW images without the usual long delays, and reduced noise, would be ample camera for the non-enthusiast, the vast majority of happy snappers and as a second camera for the enthusiast.
-
Therefore the 20D with a pixel pitch of 6.42 microns gives us a 28% better resolution than the 5D with a pixel pitch of 8.24 microns for any same sensor size.
But that is only useful/true when one doesn't have sufficiently long glass to compose the subject properly in the viewfinder. In the majority of situations, the 5D's additional pixels can be used to capture additional subject detail in spite of the fact that they are spaced a bit further apart on the silicon. It's the same reason why my 1Ds got a lot more use than my 10D while I had it.
-
Will the Nikon D2x continue the line? Probably not. 12MP on that size sensor is getting close to the useful limit.
Ray,
how do you conclude that the 5.4 microns pixel pitch of the D2X is anywhere close to the limit? Even the lower quality interline CCD technology of inexpensive digicam sensors can get to less than half that spacing: around 2.5 microns is now common, which would lead to well over 50MP in DX format. (Not that almost anybody needs 50MP).
-
I just thought I'd resubmit something that I got from a Norwegian photo discussion group, regarding the cost of Canon's sensors:
Original source: http://news.techwhack.com/1878/canon-5d-is-finally-unveiled/ (http://news.techwhack.com/1878/canon-5d-is-finally-unveiled/)
Canon manufactures its sensors from 8 inch wafers (200mm dia). One 8 inch wafer yields 20 FF sensors(if there are no defects) but presently yields are at 25 %. Seventy two 1.6x (22.7x15.1mm) sensors can be made from one 8 inch silicon wafer. Yields for 1.6x sensors per silicon wafer are at 70+ %. So therefore, with each LSI batch, there are about 5 usable FF sensors for every 50 1.6x sensors. At the time the 1Ds-Mk2 was released, yields were 10 % per wafer. With fewer usable FF sensors per wafer, production costs per sensor were significantly higher and this reflected on the 1Ds's price when it was released. Canon sells far fewer 1D series bodies than the 1.6x 20D and Kiss n Digital's. Optimizing and increasing the yields is a slower process for FF sensors.
For those who don't quite get what a "wafer" might be, and can't visualize this, try the image of a wafer on this page (and if you're curious, read the article, too, although it's certainly tangential):
http://computer.howstuffworks.com/euvl.htm (http://computer.howstuffworks.com/euvl.htm)
-
Ray,
a) to some extent, designing a lens of the same focal length for a smaller image circle allows improvements, but once the angular coverage is distinctly smaller than "normal", as with telephoto lenses, there is very little in it. Anyway, as I have said before, if and when there is an advantage to be had by designing a lens of a given focal length specifically for a smaller image circle, then such lenses can be designed for Canon EF-S, Nikon DX, Olympus Four Thirds etc. (Canon has done so at up to 60mm for their EF-S macro lens.)
Photodo scores are based on measurements all the way to the edge of the 24x36mm format image circle (about 21mm radius), but for use on DX or EF-S formats of a bit less than 16x24mm, only measurements to about 15mm radius are relevant. As you have often observed, MTF is better nearer the center: see the noticable edge fall-off in Canon's MTF graphs for their 400/2.8.
Thus, if Photodo scores were recomputed for 15mm image circle radius only, the advantage for the 400/2.8 with this image circle compared to 600/4 with 24x36mm image circle would probably increase.
c) As to your price/weight comparison: given the clear cost advantages of a smaller format in other respects, a 400/2.8 for "16x24mm formats" only needs to break even on price and weight compared to a 600/4 for 24x36mm to leave the overall cost/weight advangage with the smaller format. So if a 400/2.8 is in fact a bit lighter and cheaper, this clearly gives the edge to the smaller format.
d) The above comparions are all irrelevant to overall DSLR market share predictions unless 24x36mm format bodies drop vastly in price, and so far you have provide no evidence that this will happen, only vague analogies that indicate it might not be impossible.
Optimism is fine, but please
i) say that it might possibly happen, not that it is inevitable
ii) spare a little optimism for continuing progress in sensors and lenses for smaller DSLR formats: selective optimism about the progress of your favorite format and pessimism about the far better selling alternatives makes no sense to me.
-
If we get back to my hastily penned comment which didn't express what I was attempting to say, namely;-
but I have no doubt that a 600/4 on a 1Ds2 would deliver higher resolution than a 400/2.8 on a Nikon D2X.
, the staement was supposed to be, I have no doubt that a 600/4 on a future full frame Canon camera with the same pixel densitiy as that of the D2X (ie. about 28MP total), would deliver higher resolution than a 400/2.8 on the Nikon D2X.
From the reviews I've read, the D2X almost has the same resolving power as the 1Ds2, but in the final analysis those extra pixels of the 1Ds2 do count for something.
Since I'm not a Nikon owner and have little knowledge of the relative merits of their lenses, I'm assuming they have lenses of equivalent type and quality to Canon's, so I'm ignoring the fact that in practice one would not be using a Canon IS lens on the D2X.
Bearing this in mind, I think it would be reasonable to state that a D2X with a lens of the quality of the Canon 400/2.8 would produce very similar results to the current 1Ds2 with Canon's 600/4. The slightly lower resolving capacity of the D2X would be compensated by the slightly higher resolving capacity of the 400/2.8, as compared with the 600/4.
But what's the advantage here, of the D2X/400mm combination. The two lenses weigh about the same (the 400/2.8 is just a few grams lighter). They cost about the same, (the 400/2.8 might be a few dollars cheaper depending on where you do your shopping). The DoF with the 400 at 2.8 is about the same as that of the 600 at f4 with the larger sensor.
There are two relatively modest advantages that I see,
(1) the 400/2.8 is a faster lens and should allow the use of a faster shutter speed with the D2X for the same DoF, FoV and resolution. But this advantage is eroded if the D2X has more noise (which it certainly does at higher ISO's).
(2) the D2X body costs less than the 1Ds2. However, anyone who is really into getting fine results from high quality lenses will find such cost savings relatively small compared to the total cost of their lenses.
If we now do a bit of crystal ball gazing, and predict a time in the near future when we have 30MP full frame 35mm cameras, then such a camera attached to a 600/4 lens should according to my reckoning should produce a more detailed image than the current D2X with the 400/2.8.
Of course, in the meantime the D2X has not stood still. Perhaps we will by then have a 22MP D2X successor. My point is, there's a 'law of diminishing returns' at work here. At some level of pixel density, there's no more detail and resolution to be had without improving lenses. The pixels then just represent over-sampling to remove the need for an AA filter.
So I repeat, if the 'derogatory' cropped format is to survive, we will need a crop of DX and EF-S lenses that are off the Photodo scale of 1-5. There's going to be a squeezing from both ends. Whilst full frame comes down in price, P&S simultaneously goes up in quality. APS-C is not in an enviable position.
-
I don't see how you could be misguided when you are producing beautiful images that you and others are enjoying. It doesn't matter how you get there as long as you are happy with your photos.
T
-
Why? are there more megapixels at the edges or something?
-
On the issue of pixel density....
The 20D sensor is 22.5mm x 15mm with ~8.2M pixels. This equates to a total of 337.5 sq mm sensor size (22.5 x 15) with a pixel density of 24,252.87 pixels/sq mm.
The 5D sensor is 35.8mm x 23.9mm with ~12.8M pixels. This equates to a total of 855.62 sq mm sensor size (35.8 x 23.9) with a pixel density of 14,865.96 pixels/sq mm. Therefore, if you crop to the same sensor size as the 20D (pixels/sq mm x 337.5), you get ~5.02 megapixels.
The 1Ds Mark II sensor is 36mm x 24mm with ~16.7M pixels. This equates to a total of 864 sq mm sensor size (36 x 24) with a pixel density of 19,228.44 pixels/sq mm. Therefore, if you crop to the same sensor size as the 20D (pixels/sq mm x 337.5), you get ~6.49 megapixels.
This means that for telephoto shooting in the Canon world, the 20D is king of extracting resolution from distant subjects (provided the lens provides the resolution to do so).
As an additional point of reference, the Nikox D2x has a pixel density of 32,820.6 pixels/sq mm, and the Canon Powershot S70 has a pixel density of 183,745 pixels/sq mm. Imagine if you could mount a 600mm f/4 L IS lens on the front of that! I think that equates to a 5x focal length "multiplier", or an equivalent of a 3000mm lens - and that's before you even break out the 1.4x Extender!
There is ultimately a point for every lens where more pixels doesn't help you out with additional resolution, but based on the high quality of most telephoto lenses and the pixel density seen in the point and shoot cameras, I think we have a lot of room.
Sheldon
-
To calculate the area reduction for any crop factor exactly, divide 1 by the crop factor squared:
1.6x crop factor = 1 / (1.6 * 1.6) = 0.390625 = 39.0625% of the area of a full-frame sensor. That's close enough to 0.4 for most calculation purposes.
-
Thats all correct what Lin and Jonathan said, BUT the pixel size on both sensors are not the same. That also has to be accounted for if we compare both sensors. For the 20D the pixel lenght is 6.42 microns. The 5D has a pixel lenght of 8.24 microns, which is 28.35% larger.
If we devide 12.8 megapixels by 2.56 = 5 Megapixels (as jani explained), we also have to allow 28.35% to adjust for pixel size. Hence 5 megapixel become 6.4 megapixel.
-
I'm still confused about the original subject.
Isn't the extra "reach" on a 1.5x sensor one of their biggest advantages (in certain situations)?
-
After reading the article again, I'll add another quote to the first "Other than lower cost, there isn't much advantage".
I guess the advantages/disadvantages weigh in heavier/lighter for different people doing different work.
Lots of cool information in this thread, very interesting read.
-
For another, it seems that few 35mm format lenses have enough resolution to exploit the potential resolution of a 30MP sensor; at least, probably not many zooms.
However, as technology progresses, it is reasonable to presume that better lenses will eventually become available at an affordable price.
Fields like low light and high speed action photography are likely to be best handled by DSLR's with somewhat smaller sensors, smaller pixels, and shorter lenses. ...
The introduction of the 5D could lead one to think that this is the beginning of the end for the cropped 35mm format. ... there's no advantage in telephoto reach with standard EF lenses once the pixel density of the full frame reaches that of the cropped frame, as it's already done comparing the D60 with the 1Ds2, and comparing a future 25-30MP FF with the 20D.
I seriously doubt that there is a lot of room for resolution improvement in 35mm format lenses; that lens technology is too mature. The example of the discontinued 200/1.8 is not very relevant: such a telephoto lens with its narrow angular field of view is far easier to correct for abberations and such than normal to wide lenses.
And the evidence offered for the idea that larger formats will be able to match the sensor resolution (lp/mm, pixel spacing) of smaller formats instead suggests the opposite, through its time lag and price gap. The 1DsMkII matches the pixel size of the D60, but only several years later and at a far higher price; meanwhile the 20D (and D2X) again jump far ahead of the 1DsMkII for sensor resolution, and at lower prices.
What is worse, the 5D has the advantages over the 20D of being newer and far more expensive, and yet has distinctly lower sensor resolution (and lower frame rate.)
I am puzzled why some people ignore all trends in order to cling to the the belief that larger DSLR formats will someday offer something for nothing: sensors of more than twice the area, with the same pixel pitch and so more than twice as many photosites, and lenses with more than twice the image circle area yet with the same resolution (lp/mm), all at no significant price increase. And all this is meant to happen despite the funding handicap for the larger formats of having far lower sales revenues than the smaller DSLR formats from which to finance all the required R&D.
-
... at some point, little purpose will be served by increasing pixel density.
I just deleted a longer reply in favor of this: far before DSLR formats reach any technological limits on pixel density, they will be beyond the resolution/sharpness that the great majority of SLR users will bother to pay extra for, and larger 35mm format sensors will always cost more by enough to discourge those customers from buying them. Indeed, I suggest that the Nikon D2X is already beyond that level of image quality, and the same resolution will soon be possible in a lower level DX format body at far less than the cost of the 5D, for the same reasons that the 5D can cost so much less than the 1DsMkII.
So the pursuit of "ultimate pixel count limits" with larger formats will be the domain only of a small minority of SLR users who crave such extremes enough to actually pay for them, not just to fantasize about them in internet forums.
As a reality check, who has seen D2X prints, or even 1Ds prints, that reveal the resolution limits of their sensors? How large were those prints, and how close did you have to go to see the problems? (No test pattern viewing please!)
-
Maybe my assumption has at least one major flaw, but you haven't explained it to me yet.
Well, the explanation was further down in the same post that you so delicately snipped.
-
What lenses do you think are missing for EF-S mount that are relevant to more than a small fraction of SLR users?
BJL,
Sorry! You've just shot yourself in the foot.
For the APS-C format to survive it needs more than a small minority of adherents.
Ray,
you must be totally out of touch with the current patterns of lens usage if you think that all but "a small minority" demand fast wide to normal prime lenses like 24/1.4 or 20/1.4.
Or otherwise, you are right, but Canon, Nikon, Olympus, Pentax, etc. are all wrong, which I very much doubt. Because NONE of those companies are offering any such fast short focal length primes for their mainstream DSLR formats, even though such lenses are quite feasable to make; such lenses are lacking not only for EF-S, but also for Nikon's DX, which is solidly aimed at covering the professional level, and are also lacking for Four Thirds, Pentax DA and Konica-Minolta DT.
That leaves only your example of the 400/2.8: but Canon and Nikon do both have 400/2.8 lenses that are perfectly usable with EF-S and DX bodies! Surely you know that at such long focal lengths, nothing is gained by designing for a smaller than 35mm format image circle? Or do you believe once again that you are right about lens needs and every major camera/lens maker is wrong, in that every one of them is failing to produce new smaller image circle lenses at focal lengths beyond 60mm (60mm minimum in the case of zoom lenses.)
P. S. See my forthcoming reply on the "cropping myth" to learn why it makes no sense to produce lenses beyond about 60mm with smaller than 35mm format image circle.
-
Fascinating thread(though a bit complicated to follow when english is not your mother language...)
Something that has been on the back of my head now, and has been "insinuated" here:
No doubt, APS sensors have their advantages, specially when it comes to long reach and sport-wildlife shooting(my max. length is a 300mm and very often I find it wanting for my water-sports shooting on my 20D).A lot has been said about pixel count/FOV so..
I suspect if Canon released an APS sized camera, with a pro spec body(not necesarily so big as the 1 series), weather sealed, 100% viewfinder (and bigger if possible..), RGB Hystogram, spot-meter,in the range of the 10-12MP, and a high frame rate, we would have....a D2X!!! (and a #### of a machine, very, very, very apealling to many..).If only I didnt have all that money invested in Canon Glass!!!
Maybe, maybe...
Fascinating times for Techno-freaks!!
Happy shooting.
Erik
-
such lenses are lacking not only for EF-S, but also for Nikon's DX, which is solidly aimed at covering the professional level, and are also lacking for Four Thirds, Pentax DA and Konica-Minolta DT.
Indeed they are lacking, which seems a bit odd considering the one stop (or more) shallow DoF disadvantage of the smaller format and its need for higher resolving lenses because of its (usually) higher pixels density.
I can only speculate on the reasons, but I think high quality primes in general are only affordable when there's a large market for them. How many times have you read posts on this forum and others about the indecision of committing to the purchase of an EF-S lens which will be no use with FF 35mm? The impression I get is that most owners of APS-C Canon DSLRs bought their camera because they couldn't afford a 1Ds or 1Ds2, or at least justify the purchase to their spouse. I just don't feel that most of them, if money were no object, would choose an APS-C format in preference to FF 35mm.
That leaves only your example of the 400/2.8: but Canon and Nikon do both have 400/2.8 lenses that are perfectly usable with EF-S and DX bodies! Surely you know that at such long focal lengths, nothing is gained by designing for a smaller than 35mm format image circle?
No, I didn't know that. Could you explain why. I recall reading that NASA considered the Zuiko 300/2.8 designed for the Olympus 4/3rds format, the finest lens it had ever tested. I assumed that part of the reason this lens is so good is because it was optimised and corrected for a smaller image circle.
-
Essentially all lenses that give a field of view less than about "normal" crop the image formed by the lens, with the degree of crop depending mainly on how narow the desired angular FOV is, independent of format. Roughly the FOV gathered by most such lenses is about 40º to 50º, projecting an image circle far larger than needed for the format, and the hidden "in-the-lens" crop factor is the ratio between that image circle angular coverage and the smaller angular coverage of the sensor/film.
BJL,
This is an interesting piece of arcane information but also a fine example of a specious argument (well done! ).
If one wishes to be completely literal about this, it's rarely possible with any standard camera smaller than 4x5 to take a shot without cropping the image circle. A rectangle cannot fit into a circle without cropping parts of the circle. Whether or not the cropping is done internally through the lens design, the image circle must be cropped.
However, 'cropped format' has a different meaning, and not derogatory. It's the cropping of the image circle to a greater degree than was anticipated during the lens design.
-
BJL,
This is an interesting piece of arcane information but also a fine example of a specious argument ...
Ray, if you are going to accuse me of specious arguments, please at least make a pretence at showing the fault in my argument. Or are you now going to suggest that "specious argument" is not derogatory?
I repeat: a 200mm lens in DX format and a 300mm lens in 35mm format each take an image of about 40º angular field of view and record a portion of it with about 8º field of view, which measn that they use abot the same fractio of the image. Please explain how one format is cropping more than another.
Or to get straight to practical issues, please explain how the image quality of the 300mm, 35mm format option should be expected to be noticably better than that of the 200mm, DX option?
Let me reapt anothe pont that yopu have ignored: longer telephoto focal lengths are almost always achieved essentialy but putting a teleconverter into a lens of shorter focal length, so that the resolution in lp/mm decreases in proportion to the increase in focal length, giving about the same resolution in lp per picture height. (Your low opinion of the effects of teleconverters on resolution has been noted already in these forums.)
For example, compare the MTF curves of the Canon 600/4 to those of their 400/2.8 over the central part out to about 16mm, which is what would be used in a 1.5x crop. The 400/2.8 looks far better, rather than suffering from the crop.
If that is so, how is using a longer telephoto focal length and larger format expected to give a significant resolution advantage?
-
I repeat: a 200mm lens in DX format and a 300mm lens in 35mm format each take an image of about 40º angular field of view and record a portion of it with about 8º field of view, which measn that they use abot the same fractio of the image. Please explain how one format is cropping more than another.
Well, I'll answer this with a few other points from your previous posts.
(1) That leaves only your example of the 400/2.8: but Canon and Nikon do both have 400/2.8 lenses that are perfectly usable with EF-S and DX bodies! Surely you know that at such long focal lengths, nothing is gained by designing for a smaller than 35mm format image circle?
(2)Let me reapt anothe pont that yopu have ignored: longer telephoto focal lengths are almost always achieved essentialy but putting a teleconverter into a lens of shorter focal length, so that the resolution in lp/mm decreases in proportion to the increase in focal length, giving about the same resolution in lp per picture height. (Your low opinion of the effects of teleconverters on resolution has been noted already in these forums.)
My view has been for a long time that whatever the focal length, if a smaller image circle is the design criterion, then a higher resolution is possible. A 300mm lens designed for the 35mm format can produce significantly higher resolution (in terms of lp/mm) than a standard 300mm lens for an 8x10" field camera.
I don't understand precisely how this is possible, except in very layman's type jargon, ie. robbing the quality of the outer edges to provide greater resolution in the centre of the lens. According to basic principles, a 300mm lens throws an image circle 300mm in diameter. But modern lens design has changed that result considerably.
I've checked the Photodo results for the Canon 400/2.8 and 600/4 and it's clear the 400/2.8 has a higher MTF, is a better lens; but only marginally. At f8 we have a rating of 84 for the 400/2.8 and 82 for 600/4. This is the sort of difference between the Canon 50/1.8 and 50/1.4, but I have no doubt that a 600/4 on a 1Ds2 would deliver higher resolution than a 400/2.8 on a Nikon D2X.
Now, if you have some insights into lens design that can shatter some accepted fallacies, then I'm all ears. Please explain them clearly without hand waving, ambiguity and smokescreens :D .
-
How many times have you read posts on this forum and others about the indecision of committing to the purchase of an EF-S lens which will be no use with FF 35mm?
Many times, in this forum and others. Meanwhile, Canon's EF-S mount cameras and EF-S lenses are selling very well. What this tells me (yet again) is that the pattern of opinions and preferences expressed in internet forums is a very, very poor predictor of actual market trends.
-
Thanks T
Very nice sentiment - and I, and others, do indeed enjoy my pictures, so I'm not anxious.
One thing you can't get from something written is the intonation of a phrase such as "ya, right" :D
-
Why? are there more megapixels at the edges or something?
A brief example (using round numbers for simplicity) may make it clear. Image a full frame sensor with dimensions of 4500 by 3000 pixels, or 13.5 MP. If you want to compare that camera to a camera that has a 1.5x multiplication factor, then you have to crop the field of view to 2/3 of its original dimensions, i.e., you crop both horizontally and vertically. That would yield dimensions of 3000 by 2000 pixels or 6MP, the same result you would get if you divide 13.5 by the product of 1.5 times 1.5. Put another way, if you have a 500mm lens and a 6MP camera with a 1.5 multiplication factor, then you would have to buy a full frame camera with at least 13.5 MP in order to provide an equivalent level of detail with the same lens.
-
None of this math makes sense to me either. While you might be using a smaller area of the lens, depending on the size of the sensor being used, the image is still focussed on the entire surface area of the sensor providing you with all of the megapixels available. The image is not cropped in the same way that you crop an image in an image editor by literally throwing away pixels.
T
-
While 12.8 Mpx divided by 1.6 is 8 Mpx, that's not how it works.
If you use the crop factor correctly, you should divide by 1.6*1.6, or 2.56.
That yields 5 Mpx.
Exactly. A good reason for 20D owners to wait for the next version of the 5D, presumably 16-18Mp, presumably in 2007.
-
It's simpler to understand if you think of it this way:
A 1.6x crop factor means that you have approximately a 60 reduction in frame content. Since sampling sites (pixels) are distributed evenly, this means if you take a full frame sensor image and crop it to the dimensions of a 1.6x "crop factor" sensor you throw away 60 percent of the image, i.e., 60 percent of the pixels. That leaves 40 percent of the original pixels.
So whatever the original pixel count in megapixels is reduced by 60 percent which is the same as multiplying by .4.
11 megapixel 1DS cropped 60 percent = 11x.4= 4.4 megapixels
12 megapixel 5D cropped 60 percent = 12x.4= 4.8 megapixels
15.7 megapixel 1DS Mark II cropped 60 percent = 6.68 mp, etc.
The figures are not exact so substitute the actual effective megapixel rating and multiply by .4
For a 1.5x crop multiply by .5
For a 1.7x crop multiply by .3
Makes it a lot easier to quickly calculate...
Lin
-
I'm still confused about the original subject.
Isn't the extra "reach" on a 1.5x sensor one of their biggest advantages (in certain situations)?
Ya, right!!
-
Hummm - how do I delete a post?
-
- larger chips/sensors waste more of a silicon wafer
- larger chips/sensors have a higher probability of defects
I see this as an explanation for the price difference between the larger and smaller sensor, not a flaw in my argument. I've never suggested that the larger format would be the same price as the smaller format, everything else being equal, such as waterproofing, high frame rate, f8 focussing etc. That would not likely happen until the Chinese enter the market.
Whether the price differential will become substantially lower is mostly guesswork.
Maybe I've confused the issue by using the term 'price differential'. The price differential can remain the same for all I care. Just lower the price :D .
-
Those who are not photography enthusiasts are amply served with a plethora of P&S digicams. Those who are photography enthusiasts will gravitate towards the camera that produces the higher image quality. It's really quite simple.
I wil take your response one fallacy at a time.
Here you are using a false dichotomy, pretending that there is no middle ground of photographers between P&S digicam users and those who are willing to pay thousands of dollars for DSLR bodies and thousands of dollars more for the heavier lenses needed to get the much hope for high shutter speed advantage. The obvious reality is instead that this middle ground is now the most profitable sector of the digital camera market, serving the many photography enthusiasts who want a DLSR with interchangable lenses, greater speed that small digicam sensors can provide, and so on, and are looking for bodies at under $1000, or at most about $1,500 and use lenses not costing thousands of dollars, and so are far away from the 35mm format DLSR market sector.
-
Is 12 mpxl's (give or take) enough?? Most will say so.
Volume production of FF chips will help to reduce the cost/price of the 1DsII and likely make it more competitive with the D2x so I can forsee price reductions/rebates of the 1DsII or more likely a 1DsIII as this is more canon's style.
I heartily agree that somewhere around 12MP (or even less) is the end of the MP race for the great majority of SLR users.
I also agree that the cost/price reductions seen with the 5D could at some stage flow through to the 1D/1Ds line. Not yet, because the 5D has to use a different, lower res., lower read-out speed sensor, but once the cost of higher read-out speeds comes down ("DIGIC III"?), Canon could perhaps use the same high resolution, high speed 35mm format sensor in a range of bodies, from "enthusiastic amateur" to top of the line "medium format eater", getting sensor costs down to about 5D levels.
Then maybe we will see a 1Ds model for under $5,000. The same technological progress would probably also lead to a 12MP plus "enthusiastic amateur" DX body for under $2,000. I wonder how close the much rumored D200 will come to this?
-
of course sensor prices will come down a bit more, but what I seriously doubt is the dream of price reductions comparable to what is generally seen in electronics, because, as has been said many times, they depend to a large extent on reducing the size of electronic components: that size reduction is most of what drives "Moore's law", and is of course irrelevant when chip size is fixed at 24x36mm.
BJL,
This is not a good reason for the price of FF 35mm to remain high, but rather an aplogy for the insanely high prices. At best it's an explanation for prices not dropping as dramatically and quickly as other electronic components.
The sensor is a major component in a camera, but not the only component, and the fact that the 5D with its 24x36mm sensor will have a retail price somewhat lower than that of the Dx2 which has a sensor less than half the area, is testament to this fact.
There's an anlogy here with regard to CRT monitor prices. 11 years ago when I bought my first computer, I really wanted a 20" Sony monitor with a maximum resolution of 1280x1024 at 70Hz, but couldn't justify paying the A$3500 asking price. In terms of current dollars, allowing for inflation over the 11 year period, that's around $4500.
I find I can now buy a 20" monitor with a resolution of 2048x1536 at 70Hz for A$700 and a top of the range LaCie 22" with a res of 2048x1536 at 85Hz for around A$1600.
Now this is not as spectacular a drop in price that we are used to seeing with regard to hard drive capacity, RAM and processor speed, but nevertheless a very substantial price drop despite the obvious fact that a 20" CRT monitor cannot be miniaturised whilst still remaining a 20".
-
BUT the pixel size on both sensors are not the same
We don't know the pixels size. That information is not generally released by the camera manufacturer. The figure we get is the pixel pitch, the distance between the centre of one pixel and the centre of the adjacent pixel. For example, I believe the size of the D30 pixel is around 5.25 microns, but the pixel pitch is (from memory) something like 9 microns. This figure of 9 microns is loosley bandied around as being the pixel size, which is incorrect.
-
Apart from the obvious advantages of weight and cost, there are two clear advantages to the cropped format that affect image quality. (1) A usually greater pixel density which is able to resolve more detail provided the lens is capable of delivering it, (2) A usually superior edge performance to the image.
These advantages should be weighed against the advantages of the larger format, namely a usually greater number of pixels in total on the larger sensor.
The smaller format provides greater detail per square millimetre, or more line pairs per mm, whereas the larger format provides more detail and resolution per picture.
Can't make it any clearer than that :D .
-
how do you conclude that the 5.4 microns pixel pitch of the D2X is anywhere close to the limit?
Just hearsay, BJL :D .
Wasn't this the 'raison d'etre' for the creation of the 4/3rds format; ie. 35mm lenses cannot resolve smaller than 5 microns.
-
how do you conclude that the 5.4 microns pixel pitch of the D2X is anywhere close to the limit?
Just hearsay, BJL .
Wasn't this the 'raison d'etre' for the creation of the 4/3rds format; ie. 35mm lenses cannot resolve smaller than 5 microns.
Firstly, why would DX format be limited by the resolution of 35m format lenses? It already uses new DX format lenses for the wider angles where resolution is most problematic, and Nikon is of course free to introduce as many new DX format lenses as are needed to keep up with future higher resolution sensors. Nikon has indicated a 2 micron resolution goal for DX lenses.
Secondly, the main reason for creation of both Four Thirds and DX format seem to be the cost and size advantages of smaller formats and shortrr focal lengths; the same ones that enabled 35mm film format to drive medium format out of the mainstream, and even out of the professional mainstream. A possible secondary reason is that digital sensors with microlenses work better with a ratio of back-focus distance to sensor size far greater than is the case for 35mm SLR formats. Nikon achieves that by keeping the same back-focus distance and reducing sensor size; Four Thirds reduces both dimensions, but the sensor size more than the back-focus distance.
-
More and more people are buying dSLRs like the Rebel and d70, for them they get quality better than film with smaller, lighter lenses for cheap. I have a feeling Canon sells a lot more Rebels than any of their pro/high end equipment.
For average joe, "cropped" sensors are nothing but advantage.
The squeeze is already happening. When the first of the 8MP P&S cameras appeared, such as the Sony F828 and Minolta A2, a lot of camera buyers who did not already own a stack of Canon or Nikon lenses but wanted the quality that a 6MP DSLR could deliver, opted to buy an 8MP P&S as a cheaper and more convenient alternative. A lot of those potential buyers were rather concerned about the noise and birefringence (purple edges). The resolution might have been equal to that of a 6MP DSLR, but noise was clearly greater and therefore no possibility of a high ISO option with such small sensor cameras.
There were good reasons for anyone concerned about image quality to reject these (relatively) noisy, small sensor cameras.
Those reasons are disappearing with the introduction of the 10MP Sony DSC-R1 which boasts a sensor size virtually as large as that of the 20D and a Carl Zeiss 15-75mm zoom with F2.8 maximum aperture. How does that compare with my Sigma 15-30/F3.5 zoom that costs more than half the price of the DSC-R1 including lens?
It wouldn't surprise me if this new Sony can produce sharper images than the 20D with nearest equivalent Canon zoom lens, and images with equally low noise at equivalent ISO settings.
-
My view has been for a long time that whatever the focal length, if a smaller image circle is the design criterion, then a higher resolution is possible.
I have no doubt that a 600/4 on a 1Ds2 would deliver higher resolution than a 400/2.8 on a Nikon D2X.
This has got to be a typo or something, if I have misquoted you I apologize.
-
After reading this thread, which is interesting, I find that although it has addressed both sides of the issue well, it has as BLJ said ignored market trends. I wonder if two cameras were put on the market and the only difference between them being FF vs. DX. Everything else, features and price were the same, which would sell better?
If the image quality up to 400 ISO was the same, I'd go for the DX over the FF everytime. And if I could find a point and shoot with comparable quality in image to this hypothetical camera, I'd probably buy that too. To me the 12-24 DX delivers almost everything I need at the wide end and the extra reach at the long end seems to work for me.
So, my answer to the "Ya, right" is "Ya, all right."
-
If you crop the picture from the 5D by 1.6 (like the 20D sensor) then you get exactly 8 megapixels. In other words you can have the best of both worlds.
-
The image is not cropped in the same way that you crop an image in an image editor by literally throwing away pixels.
Yes it is, if you are comparing two different formats with the same pixel density, such as the Canon D60 and 1Ds2.
Shoot the same scene with both cameras, from the same position using the same lens. The D60 image should be identical to the 1Ds2 image cropped in an image editor to the same size as the D60 image, ie. by literally throwing pixels away.
Note: I don't mean identical with regard to every characteristic, such as color hue and balance, and noise at high ISO etc., but identical with respect to resolution at base ISO.
Ray, I re-read your post and I now understand where you were coming from. We were just looking at things a little differently. You make some good points. :)
T
-
Of corse you are right Ray.
But whatever the pixel size is, we always have to calculate the pixel pitch, because that is the size of that tiny little square which we are going to see seamlessly besides the next little square. Therefore the 20D with a pixel pitch of 6.42 microns gives us a 28% better resolution than the 5D with a pixel pitch of 8.24 microns for any same sensor size.
-
Ray,
I am glad we seem to agree that this is just about a larger minority of high budget SLR users like you changing to 35mm DSLR format, rather than an absurd prediction that the smaller DSLR formats wil cease to dominate overall digital SLR sales.
A few facts though:
- the price difference between the 20D and 5D is about US$1800, well over the $1000 in your $500 vs $1500 comparison. Sensor price difference seems to be the main factor in this US$1800 difference.
- The format price difference goes up if you wish to get the much talked about better high speed performance in telephoto shots, because to use the same ISO with a larger format needs a longer focal length but the same minimum aperture ratio. Compare a 200/2.8 to a 300/2.8 for price (and weight), for example. What is more, the same minimum aperture is more often not even available for the longer focal length: where Nikon DX can uses 200/2, 300/2.8 or 400/2.8, the equal f-stop 35mm same FOV counterparts of 300/2, 450/2.8 and 600/2.8 do not exist. If instead you use the same focal length with the larger sensor and crop, all the image quality advantages of the larger sensor are cropped away.
Presumably your prediction about the smaller DSLR formats (please, can we avoid the inaccurate and derogatory term "cropped format"?) needing "a full range of high quality EF-S lenses" only refers to Canon's EF-S mount DSLRs; the survivial of DX format, Four Thirds, etc. do not rely on EF-S lenses!
Even then, you have not offered the slightest evidence that the price penalty of the larger format will ever be low enough for the majority of SLR buyers: US$1800 is far, far too much: it is more than the gap between entry level and top of the line in 35mm film SLRs! Looking at film SLR prices, every few hundred dollars leads to a substantial reduction in sales volume.
Finally, the current range of lenses for EF-S mount DSLRs is probably adequate for the great majority of SLR users. What lenses do you think are missing for EF-S mount that are relevant to more than a small fraction of SLR users? Hint: mainstream SLR lens usage these days is dominated by "non-pro" zooms slower than f/2.8, along with primes at the wide and telephoto extremes of f/4 or slower, and maybe macro lenses. Everything else is low volume.
-
There should be no doubt that whatever the costs and difficulties that are currently associated with 35mm sensor production, they will prove to be a temporary state of affairs. That's the nature of technological progress.
Your assumption has at least one fundamental flaw.
The same advances that will make 35mm sensor production easier/cheaper will also benefit smaller sensors.
That is just as inescapable as the famed "technological progress".
You just can't get away from the following two problems without a fundamental change in how sensors are made:
- larger chips/sensors waste more of a silicon wafer
- larger chips/sensors have a higher probability of defects
Whether the price differential will become substantially lower is mostly guesswork.
-
The businessman in me says "Canon screwed the pooch" in their latest direction. Did they waste the R&D (time) to release a basic spec FF.
I can't see any "significant" cost reduction in the cost of FF chips from here. The HT industry doesn't work that way. Thats been stated many times before. Maybe higher density and more Mpxls, but not reduced cost.
But as competitors release cameras in the 11-13 mpxl range with far better bodies (lets face that the 5D has a low cost/spec prosumer body) initially at a 30% discount (pro spec'd body) and ultimately a 60% discount (prosumer body), the sales tide will turn against FF
Is 12 mpxl's (give or take) enough?? Most will say so.
Volume production of FF chips will help to reduce the cost/price of the 1DsII and likely make it more competitive with the D2x so I can forsee price reductions/rebates of the 1DsII or more likely a 1DsIII as this is more canon's style.
I fully expect Canon to release a pro DX chipped camera to compete on cost with Nikon.
I expect Nikon and Canon to explore ways (primarily in software) to make the image capture from the Dx chip more palitable . Noise reduction, lens aberation correction in software is coming (Nikon). Canon will fix the nasty edges with FF digital. An excuse for new camera models BTW.
In the end, the cost differential will make the difference between who ends with the lion share of the business. We've gotten close enough to the "end game" with regards to image quality, so price differential becomes a significant factor in the decision to purchase.
Bob
-
The term 'cropped format' is not derogatory but precisely meaningful and provides an indication as to why it will probably not survive. ... a 20D with an EF lens is indisputably a cropped format.
The cropping myth.
Essentially all lenses that give a field of view less than about "normal" crop the image formed by the lens, with the degree of crop depending mainly on how narow the desired angular FOV is, independent of format. Roughly the FOV gathered by most such lenses is about 40º to 50º, projecting an image circle far larger than needed for the format, and the hidden "in-the-lens" crop factor is the ratio between that image circle angular coverage and the smaller angular coverage of the sensor/film.
Thus for example, a 300mm lens used in 35mm format crops the image by the same proportion as a 200mm lens used with DX format, as each crops down to about 8º FOV. In fact, with true telephoto lens designs, a 300mm lens is internally something like a 200mm lens at the front, but with greater magnification in the diverging rear elements, like a built in teleconverter. (That "built-in teleconverter" is the definition of a true telephoto lens design.) And like any teleconverter, what it is effectively doing is cropping, since the extra enlargement make more of the image formed by the front part of the element fall outside the sensor.
The cropping fallacy comes from comparing what happens when the same focal length is used with different formats, instead of the relevant comparison which is between lenses that give the same FOV in the respective formats by using different focal lengths. Longer focal length telephotos have a greater degree of hidden "in lens" crop.
When lenses of the same FOV are compared, one sees that all the shorter lens does with a smaler sensor is compress the same light from the subject into a smaller part of the focal plane, so that the same light is gathered by the sensor, with no more or less cropping loss.
P. S. How can you say your wording "cropped formats" is not derogatory, and then claim that this putative cropping is such a problem that it "provides an indication as to why it [EF-S] will probably not survive."
-
Sorry! I'm also guilty of confusing the issue. In the example I mentioned there probably wouldn't be much in it.
The point I'm making, which BJL seems to indicate is fallacious, and he might well be right, is this.
It should be possible to design a higher resolving 400mm lens for a smaller format, such as the D2X, than the current 35mm 400mm lenses, because of the smaller 'working' image circle required. Futhermore, if the APS-C format is to continue to compete with the larger full frame format as pixel density continues to increase in both formats, the smaller format will have to rely upon superior lens performance.
BJL seems to be saying that lenses longer than about 60mm rely upon a built-in teleconverter to achieve that telephoto effect, and we all know that teleconverters decrease resolution.
Since I'm not a lens designer, I'm having trouble understanding this concept, particularly in relation to Photodo results which show that Canon's sharpest lens is the 200/1.8. This lens is sharper at f4 than any of Canon's 50mm lenses at any aperture.
Canon's 400/2.8 IS weighs as much as the 600/4 and is only marginally less expensive.
-
APS-C is not in an enviable position.
Consider the use of these smaller sensors in "beginner" dSLRs. More and more people are buying dSLRs like the Rebel and d70, for them they get quality better than film with smaller, lighter lenses for cheap. I have a feeling Canon sells a lot more Rebels than any of their pro/high end equipment.
For average joe, "cropped" sensors are nothing but advantage.
-
It wouldn't surprise me if this new Sony can produce sharper images than the 20D with nearest equivalent Canon zoom lens, and images with equally low noise at equivalent ISO settings.
And it wouldn't surprise me if Sony uses essentially the same sensor technology at the slightly larger DX format for the next Nikon DSLR, so returning to 12.5MP of the D2X, but without the fancy four channel read-out and such needed for its higher frame rate capabilities. That could produce the much rumored "D200" with price under US$2,000. A DX format sensor is less than 20% larger in area than the R1 sensor, so the sensor cost will not be so much greater.
If so, I wonder how that would compare to the 20D? To return to the original topic of this thread, it would probably allow a mixture of "finer grained" prints [meaning higher ppi] and increased telephoto reach with a lens of the same focal length through greater cropping latitude.
In fact, subtracting the cost of the lens in the R1, replacing the EVF system by a reflex mirror and pentamirror OVF, Nikon and Sony could probably put such a 12.5MP DX format sensor into a bare bones, low frame rate, entry level DSLR for about US$1,000, but I do not expect that, at least for a while.
-
BJL,
The next level of P&S development is the design of sensor and lens as an integrated package including DXO type corrections. Not that I have any insider knowledge of what's on the drawing board. I'm just using my nous.
The impression I'm getting from your erudite comments is that major sensor manufacturers churn out the chips and lens manufacturers do their best to fit the lens to the sensor.
-
BJL,
...
The impression I'm getting from your erudite comments is that major sensor manufacturers churn out the chips and lens manufacturers do their best to fit the lens to the sensor.
Ray,
first "arcane" and now "erudite"; I am flattered!
The trend of adapting lens designs to the needs of new sensor technology is fairly strong and persistent; it has been going on at least since Leica attacked the problem of designing new lenses for a new type of sensor, the tiny 24x36mm frames of film with new, higher resolution emulsions.
Canon does this not only with its new short back-focus EF-S designs but also with quiet redesigns of the rear elements of some EF lenses, to reduce flare problems due to the high reflectivity of most electronic sensors. In a smaller way, Canon's now completed series of F/4L zooms is a response to new sensor characteristics: several stops more usable ISO speed than film, allowing many photographers to trade one stop for high quality lenses that are lighter, less expensive, and offer wider zoom range than the old f/2.8L series.
Of course it goes a bit the other way too: the designs of sensors in most current DSLRs are clearly partly chosen (format size) for adequate compatability with existing lenses. Even if only as a transitional step while a new range of lenses is developed.
By the way, DXO style corrections are already offered as part of the Four Thirds system: lenses communicate the neccesary information to the camera, allowing some corrections to be done in camera, with others like distortion correction doable later on the computer. Nikon also offers some of this, for the special case of rectilinear correction to images from its 10.5mm DX fish-eye lens.