Nikon vs. Cannon

[Slough]

Nah - you're under-achieving. If money is not an issue why not go for a P45 set-up? That would take the Valkyries for a ride, wouldn't it?   
[a href=\"index.php?act=findpost&pid=106329\"][{POST_SNAPBACK}][/a]

Nah - you're under-achieving. Book time on the Hubble Telescope.

[Mark D Segal]

Nah - you're under-achieving. Book time on the Hubble Telescope.
[a href=\"index.php?act=findpost&pid=107059\"][{POST_SNAPBACK}][/a]

That made my day!  

[John Camp]

Interesting post John, but I hardly think some-one paying around 8K for a 1DsMkiii would have the profile of a "camera pointer"  -  that kind of dough usually signifies a more serious-minded photographer!
[a href=\"index.php?act=findpost&pid=106966\"][{POST_SNAPBACK}][/a]

For a particular kind of pro -- a newspaper guy -- the expense isn't in the camera, it's in the photographer's salary. At a modest-sized metro paper, a photographer on an average salary will cost the paper ~$100,000 a year, when you include the paper's share of social security, medical insurance and pension; compared to that, an $8,000 camera that will last for several years is small potatos.

Like this: suppose you were able to shoot and download each series of shots on a continuing basis, via cell phone. That means that an editor could sit at a computer and watch the shots streaming in. Say it's a political rally of modest importance or a football game: they know they've got space for one shot on the inside metro for the political rally, and when a shot streams though that's "good enough," they cut the photographer off, and move him to the next assignment -- say, the closest house fire. Meanwhile, at the football game,  they know they're going to use three shots in color: they want one close-up of a quarterback being chased or caught, one pass reception, one touchdown. As they get each one, they could check it off and maneuver the photographer around the field to maximize his chances of getting the next one. I worked for newspapers for quite awhile, and believe me, this type of situation would be very attractive for editors who are pushed both for time and material. I bet you could squeeze a couple of extra assignments per day out of a photographer: his camera would become a kind of monitoring device that would not only tell you what your guy is doing, but would tell you when you could move him on. It would also remove most of his judgment, and there are a lot of editors who would like that, too. That's where the "camera-pointer" line comes from.

JC

[BJL]

I suspect there might be a large number of trade secrets and patents involved in the refinement of such a sensor which gives Canon an advantage.
[a href=\"index.php?act=findpost&pid=106887\"][{POST_SNAPBACK}][/a]

I can see no way that designing a larger sensor requires extra "trade secrets" other than the fabrication issues that I mentioned above, which a number of sensor makers clearly know how to handle. As far as designing as opposed to fabrication, larger photosites are clearly no problem, and once the individual photosite is designed, filing out a larger design by using adding more rows and columns of them seems rather trivial. I have to think that the issue is purely a business decision based on balancing costs (both development costs and unit manufacturing costs) against the expected sales potential.

Perhaps the reason is the very odd aspect ratio which would result, 1:1.29, which is slightly squarer than the 4/3rds aspect ratio. Perhaps this is seen as a marketing disadvantage.
[a href=\"index.php?act=findpost&pid=106887\"][{POST_SNAPBACK}][/a]

The aspect ratio does not have to be so odd: the range I suggested includes options for 5:4 and 4:3 shapes (24x30, 24x32), rather common choices in larger formats for both sensors and prints.

But maybe there is a strong conservative force at work in the 35mm based world, demanding images of the shape that is familiar from 35mm film, no matter how easy it is to crop. And even if this is achieved by limiting the camera to what is essentially a crop from what could be achieved with a "taller" frame: one of the same width but greater height. For example, Canon persists with a 3:2 frame format of 28.1x18.7mm in the EOS-1 D MkIII even though the fabrication size limits mentioned by Canon would almost certainly allow a taller format like 28.1x20mm and probably up to about 28.1x22mm. The total chip size limit mentioned by Canon is 33x26mm, so even if 28.1mm is the maximum active area width due to losing 3.9mm at the sides, the same 3.9mm lost vertically would allow up to 28.1x22.1mm. It would be too much of a coincidence for the maximum possible width and height for the active area come out to be in 3:2 proportion when the 33x26mm limit is quite a different shape (1.27:1).

Keeping the 1D series crop format at exactly is a mystery to me when Canon keeps making the point that this is about the largest sensor size currently possible without the cost penalty of multiple step fabrication.  Firstly, a taller frame could easily be used with a 3:2 crop mode and crop lines marked in the VF, so nothing is lost compared to a frame that is less high but no wider. Secondly, there are many professional photographic situations that benefit from less wide shapes like 4:3 and 3:2, and when cropping to those shapes, the extra height would give a somewhat larger format. And isn't bigger always better, at least if the extra cost is not too great and the lenses available all cover the larger frame?

A final note: it is hard for me to see a rational basic for the facts that
- one area of high end photography (everything larger than 35mm format) uses almost exclusively on 4:3 and/or 5:4 formats, with even square sometimes seeming more popular than 3:2;
- another area of high end photography (35mm and the 1.3x crop, along with "APS-C" DSLR formats) uses 3:2 formats exclusively; and
- at smaller formats still (FourThirds and compact digicams) the dominant format shape choice swings back to 4:3 sensor shape, with the 3:2 format digicams of years past now gone from the market.

Some of this has to be a largely irrational attachment to the familiar.

[BJL]

For a particular kind of pro -- a newspaper guy -- the expense isn't in the camera, it's in the photographer's salary.
[a href=\"index.php?act=findpost&pid=107077\"][{POST_SNAPBACK}][/a]

Maybe, but camera costs are probably independent of salary costs, and so saving a few thousand on each camera is still significant to the accountants. I see this when I have to justify the cost of a new office computer to my employer, even though the total cost of my desired options is about one month of my salary every few years.

But I do believe that there are other reasons for even expensive cameras like the EOS-1 D MkIII, D2Xs and D2Hs using formats smaller than 35mm film: the combination of the resolution/telephoto lens size advantages of having pixels that are not too big with the operating speed (frame rate etc.) advantages of not having too many pixels to read out and process.

[Ray]

I can see no way that designing a larger sensor requires extra "trade secrets" other than the fabrication issues that I mentioned above, which a number of sensor makers clearly know how to handle. As far as designing as opposed to fabrication, larger photosites are clearly no problem, and once the individual photosite is designed, filing out a larger design by using adding more rows and columns of them seems rather trivial. I have to think that the issue is purely a business decision based on balancing costs (both development costs and unit manufacturing costs) against the expected sales potential.

I must admit I just can't see how you could be right here. A CMOS sensor consists of various components at each photosite to process the signal. The design of many of those components will be patented. It's even possible the technique that Canon has used to reduce the gap between each microlens, in the 400D and later in the 1D3, is a patented process which Nikon or Sony cannot use. That doesn't mean of course that they cannot achieve the same result through a different process, if that's possible. There's more than one way to skin a cat, as the saying goes.

Unless Nikon already have a high pixel count, prototype FF sensor they are  working on, it's not going to be good enough to simply use the same processes and designs, but with larger photosites. They'll have to bestow a usable ISO 6400 capability on a pixel no larger than a D2X pixel, just to catch up with Canon. I'm assuming here, of course, that the successors to the 5D and 1Ds3 will have increased pixel count and be at least as good as the 1D3 in terms of low-noise/high-ISO capability.

Some of this has to be a largely irrational attachment to the familiar.

Definitely. Irrational attachment to the familiar is a strong force which seems very prevalent in human affairs   .

[John Camp]

Unless Nikon already have a high pixel count, prototype FF sensor they are  working on, it's not going to be good enough to simply use the same processes and designs, but with larger photosites. [a href=\"index.php?act=findpost&pid=107107\"][{POST_SNAPBACK}][/a]

Thom Hogan is looking for a ~22mp FF in the next Nikon D3x.

JC

[Mark D Segal]

Thom Hogan is looking for a ~22mp FF in the next Nikon D3x.

JC
[a href=\"index.php?act=findpost&pid=107126\"][{POST_SNAPBACK}][/a]

He may be looking, but will he find it? Anyhow, if he has inside information and you've heard it, surely Canon has too..............this could be interesting!

[BJL]

A CMOS sensor consists of various components at each photosite to process the signal. The design of many of those components will be patented.
[a href=\"index.php?act=findpost&pid=107107\"][{POST_SNAPBACK}][/a]

Ray, as I indicated before
1) Sony already has the technology to make good DSLR sized photo-sites, in both CMOS and CCD. In fact Sony is making CMOS photosites slightly smaller than anything from Canon: 5.5 microns in the D2X and D2Xs vs 5.7 microns in the 400D, and 7.2 to 8.2 in Canon's FF models. I have not compared per pixel performance between the D2Xs and the 400D, have you?

2) A Sony/Nikon 24x36mm format sensor would probably have larger photo-sites that the D2Xs, as that 5.5 pixel pitch would give about 30MP, beyond what is generally expected at the next generation of Canon FF. Indications are that 30MP would be too far for Canon to go with its next FF model, at it would likely over-reach the resolution limits of many Canon lenses. (Not to mention rather uncomfortable upper limits on the DOF possible at apertures big enough to limit diffraction adequately and thus get the full 30MP worth of detail.)

... it's not going to be good enough to simply use the same processes and designs, but with larger photosites. They'll have to bestow a usable ISO 6400 capability on a pixel no larger than a D2X pixel, just to catch up with Canon.
[a href=\"index.php?act=findpost&pid=107107\"][{POST_SNAPBACK}][/a]

Firstly, the point about needing "pixels no larger than a D2X pixel" is not true as explained above: pixel pitch of about 7.2 microns wold be enough to "catch up with Canon" as far as current offerings, and I very much doubt that the next generation of 24x36mm sensors will have pixel pitch as small as the 5.5 microns of the D2X.

Secondly, talk about needing good ISO 6400 performance is taking the high ISO obsession to an extreme. Larger formats in particular are far more about high resolution and good dynamic range at low to moderate ISO than handing extremes of sensor underexposure (low light, high shutter speed shooting). Even Canon persists in making its high ISO, high frame models in formats smaller than 35mm, like the new 1D MkIII. And even the 20D, 30D and D200 offer higher frame rates than any FF model.
If Nikon goes to a format larger than DX, I envision it being for the type of photography served by the 1Ds series. (How often do you feel the need for ISO 6400 at apertures f/2.8 or bigger? At smaller apertures (higher f-stops) than that, the big format, big pixel noise advantage is illusory).
 
Thirdly, referring back to the comparison of the D2X's 5.5 micron photo-site design from several years ago to Canon slightly larger and more recent 400D photo-sites, how much of a high ISO performance gap is there? (Actually, I interested to see how Panasonic goes in pushing the limits of noise levels, highlight headroom and dynamic range with its new 4/3" format nMOS sensors with 4.7 micron pixel pitch. The nMOS design supposedly allows larger electron wells than CMOS at the same pixel pitch, helping to allow smaller pixel spacing. For some time now, Canon CMOS has had the largest minimum pixel pitch, for whatever reason.

I'm assuming here, of course, that the successors to the 5D and 1Ds3 will have increased pixel count and be at least as good as the 1D3 in terms of low-noise/high-ISO capability.
[a href=\"index.php?act=findpost&pid=107107\"][{POST_SNAPBACK}][/a]

I do not see much basis for that.
The 1DMkIII pixel size is the same as in the 1DsMkII, so going beyond the current 16.5MP of the 1DsMkII would require smaller pixels than the 1DMkIII. In particular, to be consistent with your above talk about Nikon needing pixels no larger than 5.5 microns, you should surely be looking at that same pixel pitch from Canon, meaning photosites about half the area of those 1DMkIII.   Why should we expect that Canon will anytime soon be able to go so far below 1DMkIII pixel size while still matching the 1DMkIII in terms of low-noise/high-ISO capability?

[Ray]

The 1DMkIII pixel size is the same as in the 1DsMkII, so going beyond the current 16.5MP of the 1DsMkII would require smaller pixels than the 1DMkIII. In particular, to be consistent with your above talk about Nikon needing pixels no larger than 5.5 microns, you should surely be looking at that same pixel pitch from Canon, meaning photosites about half the area of those 1DMkIII.   Why should we expect that Canon will anytime soon be able to go so far below 1DMkIII pixel size while still matching the 1DMkIII in terms of low-noise/high-ISO capability?
[a href=\"index.php?act=findpost&pid=107297\"][{POST_SNAPBACK}][/a]

BJL,
We could go back and forth like this till the cows come home. Neither of us, presumably, has inside knowledge of current prototype developments in either camp. It's always possible that Nikon, with or without collaboration with Sony, which is after all a competitor, might trump Canon at some future date and come out with a FF 35mm camera which is considered better than any Canon model in all aspects that count. Who knows!

My point basically is, if you are trying to choose between the 2 systems and format is an issue for you (which I think it should be), then going with the company that already has a superb track record in that larger format (having decided the larger format has more appeal) makes more sense than opting for a company that has no track record in that larger format.

However, there's no guarantee that such a decision will prove to be right in, say 5 years' time, after having accummulated thousands of dollars worth of Canon lenses. But I know which decision I'd make   .

[Mark D Segal]

In very practical terms, for all those interested in FF format, or larger, the next big decision point about system preference will come with the first model to supercede the 1DsMkII. If it's Canon, the main decision will be the user-value of the extra pixels relative to the cost; if it's any one but Canon - that raises major issues about investments in lenses and other peripherals. It's hard to predict technology, being on the receiving end; but the track record suggests it will be Canon. Assuming the format remains the same size, the pixels will need to be smaller. That in turn raises questions about "pixel quality" (i.e. hardware and firmware)which won't really be answerable till the erstwhile camera is tested - again, because the vast majority on the receiving end simply don't know what is up their sleeves at the moment. There may be breakthroughs in materials technology, sensor design, manufacturing technology and signal processing that we'll learn about only when they appear. As usual - it's a conundrum for people entering the market or contemplating an upgrade - how long do you wait relative to the unknowns of what's coming and when!

[djgarcia]

In the meantime I save my money AND buy printer stuff .

[BernardLanguillier]

It's hard to predict technology, being on the receiving end; but the track record suggests it will be Canon. Assuming the format remains the same size, the pixels will need to be smaller. That in turn raises questions about "pixel quality" (i.e. hardware and firmware)which won't really be answerable till the erstwhile camera is tested - again, because the vast majority on the receiving end simply don't know what is up their sleeves at the moment.
[a href=\"index.php?act=findpost&pid=107379\"][{POST_SNAPBACK}][/a]

Yep, this is obviously speculation at this stage, but we can already anticipate a few more things:

1. There will be no agreement on the measurment of noise for the next high end DSLRs. Some will like the Canon noise better, some will prefer the Nikon noise better. Overall, most people will not bother checking and will just assume that the Canon is less noisy because it is a Canon,

2. There will be no agreement on the measurment of DR for the next high end DSLRs. Some will prefer the smoother curve of the Canon with more noisy shadows, while others will prefer the more contrasty Nikon with cleaner shadows,

3. The comparisons, if any, will typically be done using a neutral RAW converter like Lightroom or ACR. It appears at the present time that those converters are benchmarked using Canon cameras and then extended to cover Nikon as well, the end result being that few Nikon shooters are happy about the default conversions delivered by ACR/Lightroom, while Canon users are overall very happy. There will therefore be no fair comparision of the RAW image quality of these 2 bodies,

4. There will also be no agreement on how to measure the physical resistance of the bodies (resistance to light rain for instance...),

-> We will probably see statements about the Canon superiority that will not be backed up by actual detailed comparisons, because of these issues, and also on the basis that pixel peeping is something of the past.

All in all, Nikon has been slowly taking back the upper hand in overall DSLR sales (at least in Japan and the EU), and I foresee a strong resistance from Canon in the high end, both in actual product and also in the marketing dept.

The only real question to my eyes being: is the Nikon going to be superior enough that some key Canon shooters have no choice but to acknowledge it?

Cheers,
Bernard

[Mark D Segal]

Hi Bernard,

Re your point 3 - is this because Canon has been less secretive about the recipe than Nikon - remember all that stuff about the proprietary algorithms when the D2X came out?

On the whole what you describe looks like a menu for some exciting debates to come - I wouldn't for a moment assume that pixel-peeping is "something of the past"; at best it is "dormant" - just waiting for the next good opportunity to demonstrate it is live and well. Anyhow, fine, from time to time the industry needs a good pixel-peep doesn't it - keeps 'em on their toes, which is just what we consumers need most - competition.

Cheers,

Mark

[BernardLanguillier]

Hi Bernard,

Re your point 3 - is this because Canon has been less secretive about the recipe than Nikon - remember all that stuff about the proprietary algorithms when the D2X came out?

Mark
[a href=\"index.php?act=findpost&pid=107561\"][{POST_SNAPBACK}][/a]

Hello Mark,

Most probably so. It is pretty ironic that in some circles, the willingness of Nikon to keep control on the RAW conversion will probably do them a diservice since their camera will only be evaluated with a third party RAW converter.

In all fairness though, those who bother to check the abilities of a top level DSLR should do it using the best availalbe RAW converter, and not a standard one. It will probably end up being different tools for both bodies.

Doing so requires to:

1. Acknowledge that ACR/Lightroom might not be the best converter for all applications,
2. Bother identifying the best tool for each DSLR, and then bother learning how to use it.

In a way, comparing the D3x and the 1dsIII using lightroom is like doing a comparison of image quality using Sigma lenses on the ground that you want to compare sensors...

Completely stupid since nobody will use these bodies with Sigma lenses.

Cheers,
Bernard

[Mark D Segal]

Completely stupid, but it will happen anyhow on the grounds that the conversion parameter should be a constant between the two comparators. In principle though you are right - each technology should be evaluated using the tools that the professionals would use to derive the best results it can yield - after spending the 1000s they will cost to buy - ouch - my money for the mega-megapixel DSLR is now being "identified" bit by bit.

[BJL]

Neither of us, presumably, has inside knowledge of current prototype developments in either camp
[a href=\"index.php?act=findpost&pid=107377\"][{POST_SNAPBACK}][/a]

Agreed: so I will disregard your previous attempts to argue how difficult it would be for Nikon to come up with a competitive larger than DX format option.

My point basically is, if you are trying to choose between the 2 systems and format is an issue for you (which I think it should be), then going with the company that already has a superb track record in that larger format
[a href=\"index.php?act=findpost&pid=107377\"][{POST_SNAPBACK}][/a]

That is completely different from your previous point, which was essentially a claim that Nikon needed to have pixel pitch no greater than 5.5 microns (28.6MP in 24x36mm format) with noise as good as Canon is promising for its still unreleased 1DMkIII sensor with 7.2 micron pixel pitch (16.5MP in 24x36mm format). Could you at least admit that those were totally biased terms of comparison, before shifting to a completely different line of argument?

In fact, I am trying a new policy: when a person's arguments are refuted, and the response is to  change to completely different arguments in support of exactly the same conclusion, my interpretation is that the person is working backwards in defense of an entrenched belief, casting around for arguments to support it (i.e. to "prop it up"). That is more like religious or political debate that rational, open-minded scientific discussion, so I intend to drop out of of any discussion which goes in that direction.

[Ray]

Agreed: so I will disregard your previous attempts to argue how difficult it would be for Nikon to come up with a competitive larger than DX format option.
That is completely different from your previous point, which was essentially a claim that Nikon needed to have pixel pitch no greater than 5.5 microns (28.6MP in 24x36mm format) with noise as good as Canon is promising for its still unreleased 1DMkIII sensor with 7.2 micron pixel pitch (16.5MP in 24x36mm format). Could you at least admit that those were totally biased terms of comparison, before shifting to a completely different line of argument?
[a href=\"index.php?act=findpost&pid=107750\"][{POST_SNAPBACK}][/a]

You are nitpicking. BJL. Since we have agreed that neither of us have any inside information of what's currently on the drawing board, and even though that point was not made at the beginning of the discussion, it must have been quite apparent to both of us. Any precise statements about future pixel pitch of FF sensors should have been understood in this context, as guesswork but based upon a reasonable prediction of trends.

It is not unreasonable to expect that the 1Ds3 will have a 22mp sensor, followed by a 5D successor with a similar (or even greater) increase in pixel count, causing Nikon to attempt to capture lost ground by introducing its first FF sensor with 28mp, or thereabouts. I say this is a reasonable predicition because Canon already have a DSLR with the pixel pitch of a 26mp FF sensor (the 400D) and probably are already working on a 1Ds4 and a 3rd generation 5D with pixels counts approaching 30mp.

These companies have some sort of road map, don't they? If you have been in the business of producing FF sensors for several years, as Canon have, then you've probably got a fair idea of what products you are going to roll out for the public in, say 3 years' time.

It might be reasonable to presume that Nikon has been doing at least some work on its first FF sensor, but it's not going to pass muster if it's a 16 or 22mp camera in 3 years' time when Canon is offering 30mp DSLRs.

If Nikon is working on a FF sensor that has a pixel pitch significantly bigger than 5.5 microns, then they'd better release it soon or the game is lost   .

[Mark D Segal]

I seem to recall a statement from Nikon that they had no intention of developing a 24*36 format. Of course they can change their minds, but that would be the clearest evidence to date of their intent. As for Canon, we don't have any idea yet - as far as I know - whether the next 1Ds will be 22, 26 or something else. Anyhow, idle speculation is fun.  

[BJL]

I seem to recall a statement from Nikon that they had no intention of developing a 24*36 format.[a href=\"index.php?act=findpost&pid=107843\"][{POST_SNAPBACK}][/a]

Me too, but arguably it was hedged with words like "no current plans". Nikon does seem to have kept up with some 24x36mm sensor R&D, and stories of prototypes being field tested have popped up from somewhat credible sources, like our host Michael Reichmann some years ago and Thom Hogan more recently. Of course not all prototypes lead to products.

I am inclined to bet against 24x36mm from Nikon, but not at extreme odds.