Flying Dog wrote:
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.
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.
Oct 05, 2007 at 10:29 PM
Andre Labonte Offline • • • • • • Upload & Sell: Off
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.
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.
@ 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
Considering nobody has gotten their hands on a 1DsMk3 yet, let alone shot one side-by-side with a D3, I guess we'll just have to wait and see on that one, eh? I'm certainly not going to declare a winner before both cameras are released!
Get the focus right and retest, still think the Nikon will be better, but on one of the images you can see the Canon is focused way back on the pillow not on the action figure. I think the Canon deserves a fair chance with proper focus.
CKrueger wrote:
Considering nobody has gotten their hands on a 1DsMk3 yet, let alone shot one side-by-side with a D3, I guess we'll just have to wait and see on that one, eh? I'm certainly not going to declare a winner before both cameras are released!
Chris,
On that we can agree ... I would like to see side-by-side comparisons in a valid test by independend, professional 3rd parties. And only then will we have a full picture of IQ (pun intended ).
But please, if you are going to make statements, make sure they don't defy the laws of physics and are substantiated. Saying the the 1Ds MK III will have better IQ than the D3 because 1Ds MK III has more MP is ludicrous as there are many factors to IQ aside from MP. Also, due to the laws of physics, image sensors with smaller photo detectors (i.e. more MP for a given sensor size) have always had more noise issues despite the smaller grain. Detector induced noise due to small photo detectors dominates over grain size and my little math exercise above explains part of the reason why. This may change when going to the 4MP or smaller sensors however.
Andre Labonte wrote:
But please, if you are going to make statements, make sure they don't defy the laws of physics and are substantiated. Saying the the 1Ds MK III will have better IQ than the D3 because 1Ds MK III has more MP is ludicrous as there are many factors to IQ aside from MP.
I never said the 1DsMk3 will have better IQ than the D3. Are you perhaps thinking of another poster?
However, I can't contrive a situation where the D3 will have more RESOLUTION than the 1DsMk3. At the absolute worst (ie: a Coke bottle lens) they should be the same.
To put some proof behind my words, I have a Canon 350D and 5D. The 350D has a pixel pitch of 6.4. The 5D is 8.2. Even with my least-sharp lens (a Sigma 70-300APO shot wide open at 300mm) my 350D resolves more detail than my 5D.
Now extend that experiment to the D3 (pixel pitch 8.4) and 1DsMk3 (pixel pitch 6.4)... it's virtually the same situation. Clearly you would need an absolutely AWFUL lens to see equal resolving power between the these two cameras. I doubt even a cheap 28-XXmm kit lens would be soft enough. And I doubt even more that anyone using either of these bodies would use anything but the best lenses with them.
Andre Labonte wrote:
Also, due to the laws of physics, image sensors with smaller photo detectors (i.e. more MP for a given sensor size) have always had more noise issues despite the smaller grain.
That is completely untrue. Compare the Canon 1D to the Canon 1DMk3. The 1D has much larger photosites, but has more noise. There are plenty of other examples of the same. Noise is not determined by photosite size alone. There are hardware/software processing and components differences at play as well.
Intrinsic noise IS related to sensor size (larger photosites => better) and the initial gain circuits. However, with the appropriate signal conditioning and software-based noise-reduction techniques, the noise can be controlled and images made to look nicer!
Oh you're most certainly right, Avi! Smaller photosites tend to be more noisy. But that's only a tendency, and a small part of the total picture. (No pun intended!)
Avi B wrote:
Intrinsic noise IS related to sensor size (larger photosites => better) and the initial gain circuits. However, with the appropriate signal conditioning and software-based noise-reduction techniques, the noise can be controlled and images made to look nicer!
Please...........go look at any of the medium format backs, you will not find one outperforming a 35mm sized sensor wrt high iso imaging
J
Oct 09, 2007 at 04:59 PM
Andre Labonte Offline • • • • • • Upload & Sell: Off
CKrueger wrote:
That is completely untrue. Compare the Canon 1D to the Canon 1DMk3. The 1D has much larger photosites, but has more noise. There are plenty of other examples of the same. Noise is not determined by photosite size alone. There are hardware/software processing and components differences at play as well.
Hey, you finally got me on something! You are correct, my statement should have said, "all else being equal, ...." However, aside from noise reducing software, photo detector size and leakage current are the two largest physical factors in image noise; trust me as I worked on the Foveon chip and had to deal with some of the leakage issues through improved isolation. Given that the D3 and 1Ds MK III are of a contemporary timeframe in the evolution of the technology, it is likely that most all else is the same and that the photo-detector size will dominate.
Oct 09, 2007 at 05:13 PM
Andre Labonte Offline • • • • • • Upload & Sell: Off
CKrueger wrote:
I never said the 1DsMk3 will have better IQ than the D3. Are you perhaps thinking of another poster?
You did not say so directly, White, Jack did and I responded that more MP does not = better IQ to which you then gave an erroneous explenation of how the smaller grain of a 22MP image could lead to less noise than a 12MP image on an 8x10 print. The basic jist was that you were supporting Jack's statement with an erroneous one. (see page 5 of this thread)
But whatever, it is clear from your other posts that we both agree that MP and/or resolution is one aspect of IQ and that there are many factors involved, including mitigation of high-ISO noise.
At the end of the day both systems will be excellent performers and I would consider myself lucky to have either (and quite happy too! )
