A Biased Evaluation of The Differences...

[Ray]

As I said before in this thread, from the point of view of the photographer and not the engineer, I couldn't care less how something is done as long as it is done well.
[a href=\"index.php?act=findpost&pid=167557\"][{POST_SNAPBACK}][/a]

In that case, you would not object to this 100% crop of a 5D image taken by Ken Rockwell at ISO 3200 which appears to have a very similar degree of noise as the D3 image, after processing the 5D image with Noise Ninja.

I show here the before and after comparison. I also lightened the 5D image and increased saturation a little.

If one camera applies in-camera noise reduction and the other doesn't, then the noise advantage of the former must be seen more as one of convenience than of fundamental improvement.

Now I'm not implying that the D3 RAW image doesn't have some degree of fundamental improvement in relation to the 5D, just that such improvement is perhaps exaggerated due to in-camera noise reduction which the 5D lacks.

[attachment=4655:attachment]  [attachment=4656:attachment]

[Mort54]

Now I'm not implying that the D3 RAW image doesn't have some degree of fundamental improvement in relation to the 5D, just that such improvement is perhaps exaggerated due to in-camera noise reduction which the 5D lacks.

The assertion that the 5D does no NR on RAWs while the D3 does is questionable, at best. According to Nikon, the D3 performs NO noise reduction over the ISO range supported by the 5D.

Nikon's D3 manual states the following: With NR set to OFF, NR is only applied at sensitivities of HI 0.3 and higher (that's higher than ISO 6400). With NR set to ON, NR is only applied at sensitivities of ISO 2000 and higher. This is right out of the manual. That's on page 299, in case you want to read it for yourself.

[Ray]

It's always been my understanding that Canon does aggressive NR, so at this point, until someone provides some proof, this is all just opinion.
[a href=\"index.php?act=findpost&pid=167589\"][{POST_SNAPBACK}][/a]

Well, Canon doesn't appear to have applied aggressive noise reduction on the 5D at ISO 3200, otherwise it wouldn't be possible to reduce noise so significantly merely by passing the image through Noise Ninja, a program which incidentally I'm not familiar with. I downloaded the trial version just to see what effect it would have.

Perhaps Nikon are making a distinction between luminance noise reduction (which they call NR) and chroma noise reduction which they don't describe as NR.

[Mort54]

Perhaps Nikon are making a distinction between luminance noise reduction (which they call NR) and chroma noise reduction which they don't describe as NR.

Where do you get that little bit of folklore from? Who says Nikon doesn't call chroma noise reduction NR. Who says these things you keep asserting?

Maybe Nikon isn't performing any NR reduction, period, at ISOs under 2000. That's what they've put in writing, so why do you keep making claims to the contrary. I've provided written documentation to back up my claim. Now how about something from the Canon side to back up your assumptions.

I think what's going on here is that some people's cherished notions are being threatened.

Well, Canon doesn't appear to have applied aggressive noise reduction on the 5D at ISO 3200, otherwise it wouldn't be possible to reduce noise so significantly merely by passing the image through Noise Ninja

When I apply noise ninja or LR NR to my D3 shots, I see a reduction in noise. So I don't understand the logic of this statement.

Ray, I apologize for being such a pain on this. But it's frustrating when all these statements get made without any obvious basis other than opinion. This is how urban legends get started. Someone says something, then it gets repeated, and eventually facts no longer mean anything.

[Rob C]

But Mort, that´s the beauty of urband legend: the truth is always first casualty, just ask any politician.

Rob C

[BernardLanguillier]

Well, Canon doesn't appear to have applied aggressive noise reduction on the 5D at ISO 3200, otherwise it wouldn't be possible to reduce noise so significantly merely by passing the image through Noise Ninja, a program which incidentally I'm not familiar with. I downloaded the trial version just to see what effect it would have.

Perhaps Nikon are making a distinction between luminance noise reduction (which they call NR) and chroma noise reduction which they don't describe as NR.
[a href=\"index.php?act=findpost&pid=167597\"][{POST_SNAPBACK}][/a]

Ray,

1. Have you shot with a D3?
2. Do you feel your 5D is letting you down or doesn't provide good enough an image quality? My guess is that you will answer no. With all due respect, then why do you even bother embarking in such discussions?

Some say the D3 is 2 stops better,

Regards,
Bernard

[Ray]

Where do you get that little bit of folklore from? Who says Nikon doesn't call chroma noise reduction NR. Who says these things you keep.

I wrote maybe. I'm just trying to offer an explanation. I just happen to be someone who doesn't believe everything I'm told or hear. If a thousand experts say the the Nikon D3 has a 2 stop noise advantage and not one does a proper comparison demonstrating this fact, I'm likely to think that maybe there's a little exaggeration going on.

For example, if I make a claim that stacking half a dozen 5D images in CS3E, with identical exposure, results in a 2 stop noise advantage, I'll demonstrate this by showing the stacked result in mean mode of a number of shots at ISO 3200 next to a single shot of the identical scene at ISO 800.

Likewise, if someone makes the claim that the D3 has approximately a 2 stop noise advantage at high ISO, it should be quite easy to display a comparison showing the 5D (or 1Ds3) image at ISO 1600 next to a D3 image of the identical scene shot at 1/4th the exposure.

If there's a suspicion that one of the images has had in-camera noise reduction applied, chroma or luminance, or just more in-camera noise reduction than the other, then its appropriate to run both images through a program like Noise Ninja to see if one cleans up better than the other. It would be reasonable to expect that the image that has had less in-camera noise reduction would clean up better.

There's an old Buddhist saying from the Kalama Sutta which goes something like this. "One should not automatically believe without question what one is taught, whether by tradition, hearsay, scripture, logic, inference, appearance, agreement with established opinion, the seeming competence of a teacher, or even one's own teacher."

I think what's going on here is that some people's cherished notions are being threatened.

Maybe, but the only cherished notion of mine that's being threatened is my sense of objectivity and impartiallity.

I'm also a bit suspicious of Ken Rockwell's test, but not because it shows the 5D image as being noisier before being passed through Noise Ninja. Both shots were given the same exposure. However, it's reported that the 5D understates its ISO settings so that ISO 3200 is closer to ISO 4000. If that's the case, why has Ken Rockwell presented the underexposed image (or lesser exposed image) as appearing brighter. I had to lighten the 5D image to get the tonality looking similar.

This is why it's important to get the RAW images behind such comparisons.

Ray, I apologize for being such a pain on this. But it's frustrating when all these statements get made without any obvious basis other than opinion.

No need to aplologise. I haven't made any statements without any obvious basis other than opinion. I went to the trouble of downloading Noise Ninja in order to demonstrate my point. Didn't you look at the images? There's chroma noise in the 5D image which can be cleaned up nicely without destroying resolution to any significant degree. The chroma noise in the D3 image has already been cleaned up nicely.

[Ray]

Ray,

1. Have you shot with a D3?
2. Do you feel your 5D is letting you down or doesn't provide good enough an image quality? My guess is that you will answer no. With all due respect, then why do you even bother embarking in such discussions?

Some say the D3 is 2 stops better,

Regards,
Bernard
[a href=\"index.php?act=findpost&pid=167684\"][{POST_SNAPBACK}][/a]

Why do I bother, Bernard? I'm a natural skeptic. There's something that doesn't seem quite right here. I know how easy it is to misrepresent the truth with images (either wittingly or unwittingly). When I first saw comparisons between the D300 and 40D at Imaging Resources showing the D300 image as having much less noise at ISO 3200, I was very excited. Wow! I want one of those.

Then it became apparent in other images (and Ken Rockwell demonstrates this very clearly) that the low noise of the D300 is at the expense of detail destroying in-camera noise reduction, so basically I lost interest in the D300.

The D3 is in a different ballpark. However, there appears to be some chroma noise reduction taking place in-camera. If this is the case, I think a fair comparison should show what's possible with the other image after out-camera chroma noise reduction has been applied.

[Sfleming]

When the D3X (or whatever it's called) comes out, it will go the other way, and sacrifice noise performance to get resolution.
[a href=\"index.php?act=findpost&pid=167221\"][{POST_SNAPBACK}][/a]

Wanna bet?

I bet that Nikon pulls another rabbit out of the hat.

(D3 shooter)
 

[John Sheehy]

Wanna bet?

I bet that Nikon pulls another rabbit out of the hat.

(D3 shooter)
 
[a href=\"index.php?act=findpost&pid=167707\"][{POST_SNAPBACK}][/a]

Well, I don't think the compromise really exists.  It is only a challenge; if you go from 12 to 18MP in the same size sensor, you can let the pixel read noise go up by a factor of (18/12)^0.5 = 1.22x and still have the same image noise, but with higher resolution.

The challenge is reading out the pixels cleanly at a target speed.  More parallelized readout may be necessary, and this can increase the cost.

The idea that capturing photons in more numerous, but smaller photosites increases photon shot noise is nonsense, at the equal-sized image level.  If you expect a 12MP crop from an 18MP sensor to capture as many photons as a 12MP sensor the same size as the 18, then you will have problems.

[John Sheehy]

Some say the D3 is 2 stops better,
[a href=\"index.php?act=findpost&pid=167684\"][{POST_SNAPBACK}][/a]

They might say that, but it might also be physically impossible.  They might be influenced by NR.

Personally, noise-related IQ to me is not whether you see noise clearly in an image; it's about whether or not you see detail through the noise, whether both are boldly natural, or both subdued by filtering.

[Streetshooter]

They might say that, but it might also be physically impossible.  They might be influenced by NR.

Personally, noise-related IQ to me is not whether you see noise clearly in an image; it's about whether or not you see detail through the noise, whether both are boldly natural, or both subdued by filtering.
[a href=\"index.php?act=findpost&pid=167848\"][{POST_SNAPBACK}][/a]

Hi John,

Can I ask what camera you use ?

Cheers   Pete

[Ray]

Some say the D3 is 2 stops better,
[a href=\"index.php?act=findpost&pid=167684\"][{POST_SNAPBACK}][/a]

Those are probably the guys who shoot in jpeg mode, Bernard   . I think the clue to the D3's low noise performance at high ISO is found in the first field report of that camera on LL by James Russell who was very impressed with the D3 but very oddly used it in only in jpeg mode.

I took my 5D to the Canon Service Centre in Bangkok for repair a couple of weeks ago. I collected it today. It so happens that the main Nikon agent in Bangkok is just a few minutes walk from the Canon Service Centre.

Apparently the D3 is so popular they didn't have one in stock, except one for demo purposes. I told them that was fine. I just wanted to compare high ISO performance with my 5D.

I arrived a bit late in the afternoon, about an hour before closing. The place was crowded and I didn't feel like setting up everything with tripod and being completely meticulous, so I just took a few dozen hand-held shots with each camera of the same dark corner in the showroom using the same focal length of 50mm but at a variety of f stops and shutter speeds ranging from f5.6 at 1/8th to f11 at 1/100th.

I autobracketed each 5D shot +/- 1/3rd stop and each D3 shot +/- 1/3rd & 2/3rds of a stop. The D3 can take a series of 5 autobracketed shots. The idea was to make sure I'd have a number of equal exposures from each camera, at ISO 3200 and above, that I could match and compare, which I'm in the process of doing.

The results are pretty much as I expected so far. With RAW images converted with the latest ACR, I see only a marginal improvement in noise in the D3 images, from ISO 3200 to 32,000. Passing both images through Noise Ninja at the same default settings narrows the gap further to pixel-peeping proportions.

Here's an example of the 5D at ISO 3200 (underexposed) compared with the D3 at ISO 12,800, both shots at 25th sec and f11. I show the ACR window for each shot then 100% crops that have been filtered using Noise Ninja.

[attachment=4709:attachment]  [attachment=4710:attachment]  [attachment=4711:attachment]

Here's another pair of 100% crops with a more agressive noise reduction applied equally to both images. I think the gap has now narrowed to virtual invisibility. I might buy that Noise Ninja program   .

[attachment=4712:attachment]

Here's the highest ISO I could match from the shots I took. F11, 1/50th, IS or VR used with both lenses. The 5D shot is 3 stops underexposed at ISO 3200 (really ISO 4000). That works out at ISO 32,000. The D3 was set at ISO 25,600 with a -0.67 EV adjustment in exposure bracketing. Both sensors have received the same amount of light at false ISO settings.

After passing both image through Noise Ninja with the same settings, this is the result.

[attachment=4713:attachment]

Unfortunately, I botched the shots slightly. I was too close with the 5D shot and have had to enlarge the D3 image on screen by 108% instead of 100% to get the image the same size. So, if you think both images look about equal, you can claim the D3 has the edge because it was taken from a very slightly greater distance.

[John Sheehy]

Hi John,

Can I ask what camera you use ?

[a href=\"index.php?act=findpost&pid=168001\"][{POST_SNAPBACK}][/a]

Yes, you can.  In fact, you just did.

Just keep in mind that before 1Dsmk3 RAWs became available to me, I said that the D3 was clearly the best low-light DSLR available.

If you're looking for a fanboy, you won't find one here.  They don't come much more  objective than me, but if you're a fanboy yourself, then objectivity may sometimes seem like "the other side".

[BernardLanguillier]

They might say that, but it might also be physically impossible.  They might be influenced by NR.

Personally, noise-related IQ to me is not whether you see noise clearly in an image; it's about whether or not you see detail through the noise, whether both are boldly natural, or both subdued by filtering.
[a href=\"index.php?act=findpost&pid=167848\"][{POST_SNAPBACK}][/a]

Those who are saying that the D3 is 2 stops better are most probably not just talking about noise, they are talking about overall image quality, meaning mix of detail retention, color saturation, ability to post-process the image, look of the noise,...

Cheers,
Bernard

[NikosR]

Ray,

What exactly does your test try to prove? To me it shows how effective NN is at removing noise without saying much about the inherent noise and detail in the raw files (as rendered by your selection of RAW converter and RAW converter parameters).

Assuming ACR can squeeze the most from the Nikon files (something that many would dispute), at least I would have expected you offered us crops of the images pre-NR or an NR processed 5D image vs. a non processed D3 image.

PS. I would also like to know (since it does seem to have an effect on shadow noise) if the D3 images you use are 12 or 14bit.

[Ray]

Those who are saying that the D3 is 2 stops better are most probably not just talking about noise, they are talking about overall image quality, meaning mix of detail retention, color saturation, ability to post-process the image, look of the noise,...

Cheers,
Bernard
[a href=\"index.php?act=findpost&pid=168091\"][{POST_SNAPBACK}][/a]

In other words, there's a big sales push to keep Nikon in the game and maintain the pressure on Canon.

Over all image quality cannot be measured in f stops, Bernard. Post processing capability is all there in ACR.... detail enhancement, vibrancy, color saturation, temperature and tint, sharpening etc.

It's clear to me that the D3 applies a lot of color noise reduction in-camera whereas the 5D doesn't. With no further noise reduction of the converted RAWs in post processing, the D3 image at high ISO is clearly better. There are definite advantages here for the busy professional or journalist who wants to cut down time spent in front of the computer, but for those who just want to be able to shoot without flash in low light conditions and who are prepared to spend a bit of time processing each RAW image, the D3 offers little in the way of reduced noise compared with the 5D.

Oops! My hotel internet connection keeps timimg out.

I get the impression that at really high ISOs, such as 25,600 or 32,000, the differences between the 5D and D3 shots are most noticeable. I doubt it would be possible to clean up the 5D image to the same degree as the D3 image without compromising resolution. Nevertheless, I get the impression in general that the 5D RAW images straight out of the camera have marginally greater detail shining through the noise.

Here's a comparison showing the obvious chroma noise in the 5D image, followed by a comparison with the chroma noise removed using Noise Ninja. There's negligible loss of detail in the 5D image as a result of the color-only filtering.

[attachment=4722:attachment]  [attachment=4721:attachment]

I should mention the above shots are at f11 and 1/50th which should be sufficient for sharp hand-held results with IS (or VR) at 50mm. No sharpening has been applied at any stage of the processing.

[Mort54]

In other words, there's a big sales push to keep Nikon in the game and maintain the pressure on Canon.

That's quite a statement. Nikon doesn't need to be "kept in the game" (actually, I find this comment quite insulting, not to mention ludicrous). They were doing quite well before the D3 was announced (actually gaining market share on Canon). I expect things will only improve further now that the D3 and D300 have appeared on the scene.

Whatever the reason for the D3's noise advantages over the competition, enough experienced pros (many of them Canon pros) have remarked on the advantages that only someone in denial could still suggest otherwise. And however the D3 achieves this (alleged NR or whatever), it does it without sacrificing acuity and saturation and dynamic range, so what difference does it make how they get there - the advantages are real and useable. Saying the 5D could do the same if only it had in-camera NR applied to RAW's (a statement you continue to make without a shred of proof), the point is that it isn't achieving the same low noise levels. Shoulda, Coulda, But Didna.

Am I being a fan boy in this discussion? You bet :-) But it takes a fan boy to recognize a fan boy, and Ray, your protestations to the contrary, your claim of unbiased objectivity is quite amusing, to say the least.

[Ray]

Am I being a fan boy in this discussion? You bet :-) But it takes a fan boy to recognize a fan boy, and Ray, your protestations to the contrary, your claim of unbiased objectivity is quite amusing, to say the least.
[a href=\"index.php?act=findpost&pid=168102\"][{POST_SNAPBACK}][/a]

Well, I'm glad you find me entertaining. There's a strong tradition on this site of promoting the benefits of RAW capture. All of us with a bit of experience are aware that RAW capture gives us the potential to get the best results.

I'm not a fan boy in any respect. If you read my posts, I've praised the capabilities of the D3 in this thread. If you shoot jpegs, the benefits are obvious and clear. But I don't shoot jpegs and I do have the time to process my RAW images in the best way I can to get the best result.

Whatever the reason for the D3's noise advantages over the competition, enough experienced pros (many of them Canon pros) have remarked on the advantages that only someone in denial could still suggest otherwise.

Indeed there are advantage and I've remarked on them too. I'm not in denial. I'm simply presenting my findings. What else can I do? I'm giving you the images and I'm showing you the conversion method. It sounds more like you are the one in denial.

Here's another pair of images at f8 and 1/50th. The 5D was set on ISO 3200 and the D3 on ISO 6400. I don't see much difference between these images, except the 5D shot has more detail. The noise in the D3 image is finer but the image is also softer. Neither image has had noise reduction applied out of camera.

[attachment=4724:attachment]

Unfortunately, I'm slightly out with my FoVs and the D3 image has been enlarged 110% to make it the same size as the 5D image at 100%, so at least part of that greater accutance in the 5D image could be attributed to that.

Do you not believe your eyes, Mort54? Would you rather believe the opinions of experts unsupported by RAW comparisons?

Saying the 5D could do the same if only it had in-camera NR applied to RAW's (a statement you continue to make without a shred of proof), the point is that it isn't achieving the same low noise levels. Shoulda, Coulda, But Didna.

I never said that. Why are you misquoting me? I've said that it appears the D3 is applying more in-camera noise reduction than the 5D, especially with regard to chroma noise which can, however, be removed in software without degrading resolution.

I've also said that having such in-camera noise reduction can be a time-saver. I've never expressed any wish that the 5D have such apparent in-camera noise reduction but I have said, and have also demonstrated, that such objectionable chroma noise in the 5D image can be easily removed with Noise Ninja.

What's your problem? I'm merely trying to find out the real extent of any D3 noise superiority. There is some, but it's not nearly as great at the RAW level as it's been made out to be. 2 stops in jpeg mode, maybe, but no-where near 2 stops using RAW images.

That's quite a statement. Nikon doesn't need to be "kept in the game" (actually, I find this comment quite insulting, not to mention ludicrous). They were doing quite well before the D3 was announced (actually gaining market share on Canon). I expect things will only improve further now that the D3 and D300 have appeared on the scene.

Nonsense! All camera companies need to be kept in the game. If they are not kept in the game, they're out of the game or taken over by other companies like Minolta was. All these companies have marketing departments and do research on the customers needs and wants.

It's my experience that most DSLR owners I meet on my travels shoot in jpeg mode. I bet most journalists do too. I think Nikon are appealing to that customer base with the D3. It seems there are only a few fanatics on LL who consistently shoot in RAW mode and are prepared to spend time processing and manipulating their favourite images. Good results out of the box seem the preferred option for most.

[Ray]

Ray,

What exactly does your test try to prove? To me it shows how effective NN is at removing noise without saying much about the inherent noise and detail in the raw files (as rendered by your selection of RAW converter and RAW converter parameters).

Assuming ACR can squeeze the most from the Nikon files (something that many would dispute), at least I would have expected you offered us crops of the images pre-NR or an NR processed 5D image vs. a non processed D3 image.

PS. I would also like to know (since it does seem to have an effect on shadow noise) if the D3 images you use are 12 or 14bit.
[a href=\"index.php?act=findpost&pid=168098\"][{POST_SNAPBACK}][/a]

My tests are trying to determine if there is any fundamental noise benefit in the D3, at high ISO, which cannot be achieved by the 5D by other methods. The indications so far are, that any noise benefit is slight.

This is of some interest to me. I was excited about the low noise at high ISO in the 20D, for example, coming from D60. There, the improvement really was 2 stops, ie. a 20D ISO 1600 shot was on a par with a D60 ISO 400 shot.

Having now compared a few 5D and D3 RAW images, it is my view that the noise advantage of the D3 is nowhere near 2 stops. I've got more images to compare, that differ by a 1/3rd of a stop, 1/2 a stop and 2/3rds of a stop etc., so at this stage I'd hesitate to put a figure on the D3 noise advantage. It's probably about 1/2 to 2/3rds of a stop, but it's beginning to look as though I was not sufficiently careful with my shooting to be able to place a precise figure on it. I now believe that my 24-105 lens at 50mm is really something like 52mm (even though it's described as 50mm in Exif) and the Nikkor 24-70 lens at 50mm might really be something like 48mm.

It would be better if I could hire the camera for a day. I might just have to do that if this disbelief in my results persists.  

Assuming ACR can squeeze the most from the Nikon files (something that many would dispute), at least I would have expected you offered us crops of the images pre-NR or an NR processed 5D image vs. a non processed D3 image.

I have done that at least a couple of times in this thread. The 5D is very responsive to color noise reduction in Noise Ninja. The D3 image is not, presumably because most of the chroma noise has already been reduced in-camera.