Amazing, really, that the person who did the test didn't take the time to critically focus one of the cameras. If one were really interested in an image quality comparison, one might think that they'd not allow one of the cameras to be backfocused on the wall (or whatever it is). Regardless of how it happened, it affected the results, and is obvious looking at the images.

Secondly, the Nikon almost surely has some amount of internal NR going on (whether the D3's NR setting is set to 'on' or 'off' is neither here nor there), and the MkIII's was clearly turned off. Not exactly a 1:1 comparison.

The D3 images look great, there's no doubt about it. Look forward to people getting their hands on production units to see what can really be done with it.

All this tells me is that they are both "great" tools. Nuff said!!!!

Tom

Yeah, the focus does look a bit off on the MK III, doesn't it. Manual focus would have been better here.

However, can't possibly understand the next comment about noise reduction. Might be a bit presumptions about the Canons being clearly turned off. And really don't know how the conclusion came about with surely Nikon having some internal NR going on.

I don't understand why anyine would care. Whatever processing is going on is somewhat irrelvant, I would have thought the results of that processing is what matters.

Either way, the Canon noise looks fine to me anyway. But if someone is going to present this sort of crtical side by side comparison, then the lens , techniques and methods used would be important.

And I don't hold to the "look at my tests, with no sharpening, or NR ,auto mode and no PP done" thing either. Set whatever you need up in camera to get the best from it.

Case in point, jpeg images from the D200 are soft on the standard settings. So change the settings to sharpen +1 or +2.

nikt wrote:
Yeah, the focus does look a bit off on the MK III, doesn't it. Manual focus would have been better here.

However, can't possibly understand the next comment about noise reduction. Might be a bit presumptions about the Canons being clearly turned off. And really don't know how the conclusion came about with surely Nikon having some internal NR going on.



I agree, the focus on the Canon image is off. The 24-105 isn't the sharpest lens ever, but it's a heck of a lot better than that sample indicates. Of course, this is a noise test, not a sharpness test.

Like you said, we can't really say who did what kind of noise reduction. But looking at the images I will say this:

* Both cameras are spectacular at ISO3200. I wouldn't hesitate to print a 16x20 from either camera. Take a step back and think about THAT accomplishment for a minute!

* At ISO6400 the Nikon seems to show a tad less noise. Neither camera looks like it's applying noise reduction like you might get from software like Lightroom or NoiseWare/NoiseNinja.

* At ISO12800 the Nikon shows MUCH less noise, although one might expect that given the "push processing" done to the Canon. The Nikon image certainly shows artifacts consistent with noise reduction. Look at the speckled false detail in the shadow of the toy and compare to the ISO800 picture. It's clear Nikon is doing more noise reduction than any Canon or Nikon camera we've seen in the past. There is definitely detail lost from this noise reduction... look at the "Egypt" text on the globe compared to the ISO800 sample, for example. But despite the loss of detail, image quality is still very usable. You should be able to make excellent out-of-camera 8x10's at ISO12800, and with a little attention from noise reduction software it should be usable even larger.

* ISO25600 ain't too pretty! It looks a bit like my wife's Fuji F30 at ISO3200, or my 5D pushed a couple stops from ISO3200. Even still, you could definitely get usable 4x6's from this setting, or even an 8x10 if your audience isn't too particular. I could definitely find use for this ISO25600 for direct-to-print 4x6's at certain nighttime events, where response time is more important than image quality.

All told I'd say I'm a bit surprised that the 1DMk3 is pretty comparable to the D3 at ISO6400, but the fact that the D3 "unlocks" such high ISOs is a significant feature, and one I'd love to have, even at the expense of noise... much like ISO800 is invaluable to me for shooting the occasional wildlife grab shot when shooting landscapes on a tripod.

Canon would do well to follow Sigma's suit and give 1DMk3 owners an ISO12800. Looks like they might have to recall the cameras, anyway, so give people a present when they get the camera back!

j.curtis wrote:
Max Penson wrote:
It is funny how people think that once you use RAW you'll always get better results than the camera JPEG. This is simply not true for today's modern cameras.

Do you have a lot to learn.

First of all, jpeg is compresses the image. How can one say there is no loss in image quality? You can't.

Secondly this is getting way off topic.

Thing is, you aren't looking at a RAW file here. You're looking at a JPEG conversion of a RAW file. So by your own assertion, you've just lost image quality as you would have if you had shot RAW. Also, with a high enough JPEG quality setting, the compression is VERY near lossless. You won't be able to see a difference without zooming in to 2:1 or more, and even then the difference from a lossless format is so minute that you would not be able to notice it without a gigantic print viewed inches away.

Bottom line, high quality JPEGs have so little lose in image quality beyond lossless files that they might as well be lossless for our purposes.

Hi Andre, I am also a new Nikon shooter albeit film w/ a couple Nikkor lenses but I learning/loving the Nikon SLR features. I am not partial to either system but my future purchases will depend on how Canon treats its customers with faulty MKIIIs. Mine appears to be a pretty good copy but I bought mine late.

The controls of the MKIII are better than my old 1D MKII. Just my opinion and many of my friends.

I have yet to see a new full frame camera lose out to new cropped dSLR in terms of noise. Yes the Nikon D3 will have less noise than MKIII. I'm fine with that.

Yes, the 1Ds MKIII will produce images that will crush the 1D MKIII or D3 esp. in prints. Ok, crush is a strong word. D3 and 1Ds MKIII are FF, 14-bit, CMOS sensors. I have yet to see a 1D MKIII or D3 image at ISO 200 beat the other in any category. Now add the extra pixels on top of that for the 1Ds MKIII. Which image will hold more shadow detail? Now print a poster. 1Ds MKIII will have better prints. We can argue about colors but that's subjective. That's not to say Nikon won't come out with a 25 MP camera in 2008 but the 1Ds MKIII will be the camera to beat for image quality and it should be for $8000. D3 will be the camera to beat in most other categories.

Ok, I'm trying not to get sucked in to the Canon vs Nikon debates. Both offer great stuff for a hobbyist like myself. I should be looking at your portfolios rather than comparisons.

Have a great weekend everyone, I'm going to try to take pics rather than fight over cameras that aren't even available. Afterall, we're talking about the top 3-4 dSLR cameras in the world. 99.9999% of the world don't even know what these cameras look like.

Edited by White, Jack on Oct 05, 2007 at 07:11 AM GMT

Edited by White, Jack on Oct 05, 2007 at 07:12 AM GMT

nikt wrote:
However, can't possibly understand the next comment about noise reduction. Might be a bit presumptions about the Canons being clearly turned off. And really don't know how the conclusion came about with surely Nikon having some internal NR going on.

Nik, I guess my point was that Canon and Nikon have chosen two fundamentally different strategies with dealing with chrominance noise; Nikon has chosen to effectively remove it in-camera, and Canon chooses to leave it in, to be necessarily removed later. I personally can't stand the Canon CMOS chroma noise, it lends nothing of value to the image, is easily removed either in-camera or in post, and I can't understand why Canon doesn't just remove it in-camera. Anyone who has any familiarity whatsoever with Canon's files would likely agree that there is no chroma noise reduction in those images. I make no statement about luminance NR because I don't use it, but I'd doubt there was any of that used, either.

As far as my 'assumption' that Nikon employs in-camera chroma NR, I think it's evident looking at the files, being that they use Bayer array sensors. Just as John frequently offers the unconfirmed 'fact' that Digic III employs some amount of internal noise reduction (which I don't doubt), I think it would be folly to assume that Nikon doesn't do the same, especially given the characteristic lack of chroma noise from their Bayer sensors. Luminance noise is likely a different issue entirely, but again, I don't nullify it in post, as I think it smooths detail.

Respectfully,

Flying Dog

look how many views this thread has compared to posts.
lurking much?

White, Jack wrote:

Yes, the 1Ds MKIII will produce images that will crush the 1D MKIII or D3 esp. in prints. Ok, crush is a strong word. D3 and 1Ds MKIII are FF, 14-bit, CMOS sensors. I have yet to see a 1D MKIII or D3 image at ISO 200 beat the other in any category. Now add the extra pixels on top of that for the 1Ds MKIII. Which image will hold more shadow detail? Now print a poster. 1Ds MKIII will have better prints. We can argue about colors but that's subjective. That's not to say Nikon won't come out with a 25 MP camera in 2008 but the 1Ds MKIII will be the camera to beat for image quality and it should be for $8000. D3 will be the camera to beat in most other categories.


Hi Jack,

Shadow detail is not a function of resolution but more of dynamic range. It is quite possible that the D3 will beat the 1Ds MKIII and vice versa. This critical aspect of image quality is yet to be determined. However, if one understands some of the physics behind image sensors, a larger sensor with less noise will most likely beat a smaller sensor with more noise (given the same processing and bit #), thus the D3 is more likely to have performance in this area.

Noise is a critical factor in image quality and clearly the D3 is KING!


We can argue about colors but that's subjective.

True ONLY if the photography does not wish to portray an image with accurate colors. And, how easy is it to adjust the settings so as to minimize post processing. From the huge number of images I have seen, Nikon is in the lead here.

Now print a poster. 1Ds MKIII will have better prints.

You have no basis for this statement. Yes the 1Ds MKIII will have more resolution, but resolution is only one factor that goes into image quality. Also, resolution does not become a factor in image quality until the print size becomes large enough to make pixelation visible. For the difference between 12 and 22MP to be visible we are talking about wall size or larger prints. Yea, laughable at best. Other factors like color, noise and dynamic range become important LONG before resolution does, at least for cameras with 12MP or more.

Again, Resolution DOES NOT EQUAL Image Quality. And resolution seems to be the only thing the 1Ds MKIII has over the D3

In the end, either system is an excellent system, but right now, for the average professional photographer who is not making billboards, the D3 appears to have the 1Ds MKIII beat and at $3K less.

Then again, like you said, this is not an issue for 99.9% of the people out there ... it sure is not for me since I'm going with a D300 (what my wallet and concience can handle ).

Best Regards,
Andre


Edited by Andre Labonte on Oct 05, 2007 at 11:00 AM GMT (Reason: gramatical error fix)

Ah, the post bug has struck! Where did my post go?

Andre

I think in truth for this format (35mm and APS-C) we are "there" wrt diminishing returns when it comes to all the factors in play and image quality in general esp wrt resolution. my D2x at 12.5MP gives me a tiff of about 70MB. I do not have the numbers for the new canon but at 4288 x 2848 (image size at 300 ppi of the D2x) to get to 2x the resolution you are surely talking More than 24MP , so the 12 vs 24 issue is moot and if that's the case give me 12


J

jmcfadden wrote:
Andre

I think in truth for this format (35mm and APS-C) we are "there" wrt diminishing returns when it comes to all the factors in play and image quality in general esp wrt resolution. my D2x at 12.5MP gives me a tiff of about 70MB. I do not have the numbers for the new canon but at 4288 x 2848 (image size at 300 ppi of the D2x) to get to 2x the resolution you are surely talking More than 24MP , so the 12 vs 24 issue is moot and if that's the case give me 12


J


Hi J,

You are right on all counts. We may see some more tweaking as time goes on on noise. Resolution would have to go to 48MP to double what we get with 12MP. UG, my hard drives groans at that thought. The one area I think we will see some huge improvements in as software and processing speed improves is with dynamic range. HDR on the fly is the next generation ... that would make you happy with all those incredible train shots!

But hey, I'm happy with my lowly D70s except for the AF; thus the desire for the D300. That's not to say i would not take a D3 (or even a 1Ds MK III) if someone gave me one

Cheers,
Andre

Andre Labonte wrote:
You have no basis for this statement. Yes the 1Ds MKIII will have more resolution, but resolution is only one factor that goes into image quality. Also, resolution does not become a factor in image quality until the print size becomes large enough to make pixelation visible. For the difference between 12 and 22MP to be visible we are talking about wall size or larger prints. Yea, laughable at best. Other factors like color, noise and dynamic range become important LONG before resolution does, at least for cameras with 12MP or more.

Not entirely true. Resolution has a direct impact on apparent noise in a print. When shooting high ISO, this effect is visible even in small prints.

Take a D2X and take two pictures. First is an ISO3200 picture at 1.5x crop and 50mm. Second is an ISO3200 picture at 2x crop and 37.5mm. These are the "same picture", in terms of framing. The difference is the cropped picture will have lower resolution (lets imagine we equalized DOF for the sake of argument).

If you print an 8x10 of both of these pictures, the 1.5x picture will have more pixels per inch, and will thus have "smaller" grain. Color blotches will be less visible, and grain speckles will appear smaller.

Indeed, the 1.5x picture will give a sharper print, too. Lesser enlargement and higher overall resolution will make the difference even at a moderate 8x10 size, especially with the detail-destroying high ISO of this example picture.

An interesting test will be when the 1DsMk3 is measured against the D3... which camera will show less in a print? The higher resolution and (presumably) higher noise 1DsMk3? Or the lower resolution and (presumably) lower noise D3? Given the large difference in resolution, I wouldn't care to wager. One thing is for certain: the 1DsMk3 will show more detail. You'd have to contrive an extreme test to make a 1DsMk3 show no more resolution than the D3.

I really don't think this is a pissing match between brand loyalists. It's just the facts of life for photographers.

I'll use a Has if I need a billboard size print. Otherwise, Canon and Nikon make great cameras that meet the demands of most photographers.

Good to see Canon's and Nikon's high-end pro cameras being discussed as true competitors. This is good for everybody. I don't think the 1d3 would have hit the market with AF problems had Nikon been competitive with pro sports shooters. I'm certain they would have paid a little more attention to their QC.

Interestingly, I can't wait for my D3, but I know many Canon shooters who are concerned about the 1ds3, given the 1d3 AF issue. Mistakes in QC can be harmful to a company's rep.

Chris,

@ 12MP an 8x10 has ~ 355 dpi
@ 22MP an 8x10 has ~ 468 dpi

That a difference of 15 microns! The human eye cannot resolve that. The improvment in noise due to the larger pixels will be much more significant that the effect of the dpi in an 8x10. Nice try!

Best Regards,
Andre

I wish this TEST was done using 1D3 Canon 24-70 vs Nikon D3 with 24 -70 Nikkor.

I wish this TEST was done using 1D3 Canon 24-70 vs Nikon D3 with 24 -70 Nikkor.

Flying Dog wrote:
nikt wrote:
However, can't possibly understand the next comment about noise reduction. Might be a bit presumptions about the Canons being clearly turned off. And really don't know how the conclusion came about with surely Nikon having some internal NR going on.

Nik, I guess my point was that Canon and Nikon have chosen two fundamentally different strategies with dealing with chrominance noise; Nikon has chosen to effectively remove it in-camera, and Canon chooses to leave it in, to be necessarily removed later. I personally can't stand the Canon CMOS chroma noise, it lends nothing of value to the image, is easily removed either in-camera or in post, and I can't understand why Canon doesn't just remove it in-camera. Anyone who has any familiarity whatsoever with Canon's files would likely agree that there is no chroma noise reduction in those images. I make no statement about luminance NR because I don't use it, but I'd doubt there was any of that used, either.

As far as my 'assumption' that Nikon employs in-camera chroma NR, I think it's evident looking at the files, being that they use Bayer array sensors. Just as John frequently offers the unconfirmed 'fact' that Digic III employs some amount of internal noise reduction (which I don't doubt), I think it would be folly to assume that Nikon doesn't do the same, especially given the characteristic lack of chroma noise from their Bayer sensors. Luminance noise is likely a different issue entirely, but again, I don't nullify it in post, as I think it smooths detail.

Respectfully,

Flying Dog

I don't really have a problem with any of that actually. I guess my point is, in camera NR , post NR, no NR... as long as it looks good. From neither camera did I see smothing of noise of loss of detail, just slightly different . I too suspect Nikon, (Panasonic, Sony, Olympus, Canon ) perform noise manipulation even if turned off.

My Canons (and the 24-105L sample I have used) are a lot sharper. DPP offers both chroma and luminance NR, but it must enabled. So, my sense is the Mark III is not getting a fair shake here.

That said, the D3 results look great.

garyvot wrote:
My Canons (and the 24-105L sample I have used) are a lot sharper. DPP offers both chroma and luminance NR, but it must enabled. So, my sense is the Mark III is not getting a fair shake here.

That said, the D3 results look great.

I think everyone agrees with you on these counts. The 1D MK III images don't even look like they were perfectly focused.

Shouldn't this comparison be done with the 1DsMkIII and the D3? The price difference is pretty steep, but they're both FF... apples to apples.

I also agree that this comparison should have been done with the 24-70L or the 28-70L on the Canon. We're talking about comparing the performance of an f4 lens with an f2.8 lens on the Nikon body.

As it is, it's a pretty flawed comparison. Not sure it tells us much. However, the Canon held up much better than I expected against a Nikon body that's probably still far from being available.

Where the Nikon really fails in the comparison? You can buy the Canon body now. Here's how the comparison goes....

Picture of my daughter's indoor gymnastics meet from this week at high ISO... Canon: hmm, pretty good. Nikon: No image available.

Well, the D3 is aimed squarely at the 1Dmk3 market (Sports/PJ). That's why you'd compare the two.

Agree on the OOF/unsharp lens bit for the 1d3, but still the colours looked nicer on Nikon at higher ISO. This is why I bought a Nikon in the first place, coz the colours look a lot nicer to me and not plasticky. (Above is purely subjective and only my humble opinion).

I can't afford either camera as it is, but it's good to see Nikon coming out so strong vs its chief competitor.

Wish you would have used 1D3 with Canon 24-70mm and. Nikon 24-70mm. Thinking about getting the D3, but I like to see a TURE test with closer comparison.

The idea of the test appears to have been showing off the noise/colour characteristics of the two cameras. Having said that, a consumer lens on the 1d3 doesn't help things...

Anyway, if you just look at the test for what it is (unscientific comparison of high ISO noise between the two cams) then you get a ballpark idea of where the D3 stands relative to the 1d3.... I'm sure if you all cool your jets and wait a good month, we can get Phil's comparisons on dpreview which tend to be well-regarded.

Wait a minute, the 24-105 is not a consumer lens as it is an L. There are many descriptions of what an L lens is and Canon isn't quite clear about it themselves in terms of designating it as a pro/consumer lens but they say this:

"these lenses use special optical technologies [such as] Ultra-low Dispersion UD glass, Super Low Dispersion glass, Fluorite elements, and Aspherical elements to truly push the optical envelope."

and you find this statment:

"L" is for "Luxury".

In the Lens Work III book by Canon.

That said, I have used two different copies of this lens and find it, although a convenient zoom range, not particularly stellar in terms of image performance. In fact I would say much less than. It reminds me a lot of the Tamron 24-135 Anniversary which was a fine lens in it's own right but just not quite stellar. It's good, just not great.

It's not a great comparison but it gives you an idea. In my opinion I would be happy to use either.

Cubfan,

Hard to compare a D3 and 1ds3, since the 1ds3 is not available. Canon has kept it under wraps.

BTW, the D3 is in production in Japan (I believe they started Sept 1), so, the release date is probably still late Nov/early Dec. I would gladly wait for the D3 rather than buy 1d3 due to AF issues. But, then I'm a Nikon guy.

Sorry to see the Cubs go in three. I was hoping this would be their year.