Good news, not only can I return my borrowed 1D3 to the owner 1 day later, I get his 5D too.

Here is the high ISO test: (I performed a sharpness test as well, and they are identical in sharpness, can't really tell the difference as far as the center is concern)

70-200 2.8 non-IS on tripod, 1D3 shots first, then use 5D, zoom in till similar FOV

Original:
http://www.kyphoto.hk/my_photos/1D3_test_shot/1D3_vs_5D_noise/Original.jpg

http://www.kyphoto.hk/my_photos/1D3_test_shot/1D3_vs_5D_noise/1D3_ISO1600_RAW.jpg

http://www.kyphoto.hk/my_photos/1D3_test_shot/1D3_vs_5D_noise/5D_ISO1600_RAW.jpg

http://www.kyphoto.hk/my_photos/1D3_test_shot/1D3_vs_5D_noise/1D3_ISO3200_RAW.jpg

http://www.kyphoto.hk/my_photos/1D3_test_shot/1D3_vs_5D_noise/5D_ISO3200_RAW.jpg

http://www.kyphoto.hk/my_photos/1D3_test_shot/1D3_vs_5D_noise/1D3_ISO6400_RAW.jpg

http://www.kyphoto.hk/my_photos/1D3_test_shot/1D3_vs_5D_noise/5D_ISO6400_RAW.jpg





Edited by kylee on May 20, 2007 at 04:33 PM GMT

Edited by Jeff on May 29, 2007 at 07:04 PM GMT (Reason: un-embedded links to oversized images)

Good stuff Kylee! Thanks a lot for doing these tests.

To my eye, I'm not seeing much of a difference between the 5D and 1D3 at ISO 1600. At ISO 3200 the 1D3 seems to have less noise. However if we compare the 5D ISO 1600 to the 1D3 at ISO 3200, the 1D3 seems to have more noise, which means it doesn't really have a 1 stop noise advantage over the 5D (I'd say more like a 1/2 - 2/3 stop advantage). Anyone else notice this or do my eyes just suck?

I think you have to compare it at each ISO setting. I concur that at 1600, the difference between the two is not that great, but as you move up in speed, the difference becomes more pronounced.

For example ISO 100 on the 1DM3 is probably quite similar to most other current model Canon DSLR's (disregarding a slight dynamic range advantage at 14 bit.)

I believe that sports photographers are going to benefit the most from this camera...those that shoot indoors. They can get away with ISO 3200 or maybe even ISO 6400 for newspapers, etc.

Alan

My eyes see it pretty much the same as thedigitalbean. This is really impressive for the 1D3 as the 5D has great high ISO performance.

my initial impression is 1D3 is a very pro body 5D

I like the way the 5D's look a little better. Impressive! Looks like the 5D's still got it!

thedigitalbean wrote:
Good stuff Kylee! Thanks a lot for doing these tests.

To my eye, I'm not seeing much of a difference between the 5D and 1D3 at ISO 1600. At ISO 3200 the 1D3 seems to have less noise. However if we compare the 5D ISO 1600 to the 1D3 at ISO 3200, the 1D3 seems to have more noise, which means it doesn't really have a 1 stop noise advantage over the 5D (I'd say more like a 1/2 - 2/3 stop advantage). Anyone else notice this or do my eyes just suck?

Looks like you've hit it on the head. I played around with the samples, did some noise reduction and resizing, and it looks like the resolution advantage of the 5D makes it even closer... a slight edge to the 1D3 at higher ISOs, but virtually identical at 1600.

Thanks for this comparison. I'll be looking forward to more of these as early adopters start posting samples. FWIW I profiled these in Noise Ninja and it reported the following indexes:

5D-1600: Noise Index 33, (15 lum, 18 chroma)
1D3-1600: 32 (16,16)
5D-3200: 48 (24,24)
1D3-3200: 47 (22, 25)

Certainly good numbers for the 1D3 but not 1 or 2 stops as some have reported.

Thanks for the posting. I looked at the oriental characters on the battery at ISO 1600 and to my eyes there isn't all that much difference in resolution.

I would love to see 100% comparison crops of a detailed image at ISO 100. Bet the difference isn't as much as one would think...

After using the 1DIII, Arthur Morris said "The converted RAW image exhibited very little noise. Both Robert O’Toole and I agreed that the noise in this and other ISO 3200 images from the Mark III was comparable to ISO 400 in the 1Ds MII and the MII 1DN."

The man may be an optimist and a Canon fan, but he's no fool. He sells his knowledge and can't afford to be too glib.

dcmiller wrote:
After using the 1DIII, Arthur Morris said "The converted RAW image exhibited very little noise. Both Robert O’Toole and I agreed that the noise in this and other ISO 3200 images from the Mark III was comparable to ISO 400 in the 1Ds MII and the MII 1DN."


I would be very, very, surprised if it's true. 3 stops in one camera generation? I doubt. Besides, noise tests that have been showing up later do not indicate this kind of superiority at all.

I don't think it will be 3 stops either. But I'm expecting a solid 1-2 stops. At this point in a product life cycle, the "truth" vacillates between elation and despair. Maybe DPP matters.

A bit more about the context for Arthur Morris' statement:

"I think that I had set the Custom Function for High ISO Noise Reduction and the image was processed in DPP with Low Noise Reduction set for both color and luminance. The converted RAW image exhibited very little noise."

He also showed web-sized final images of the results and they look quite good. I can't comment on how he made the comparison, but my suspicion (based on some of the testing we've seen in this and the previous 1D3 vs. 1Ds2) is that he was comparing it to the 1Ds2 from memory rather than a critical side-by-side comparison.

Then again, who knows, maybe the 1D3 RAW file cleans up a lot nicer than the 1Ds2 (but I have my doubts it'll be 2 or more stops better).

I believe his main camera is the 1DsII. So he should have SOME idea. But even reducing his estimate by 50% gives 1.5 stops. Still very useful.
Anyway, since I expect to be writing a check for $4500 next week I want no bad news.......

dcmiller wrote:
I believe his main camera is the 1DsII. So he should have SOME idea. But even reducing his estimate by 50% gives 1.5 stops. Still very useful.
Anyway, since I expect to be writing a check for $4500 next week I want no bad news.......

I really hope he is right, and I know what you mean, I have one on pre-order too...

Thanks for posting these samples Kylee! Very well done test shots.

The difference is smaller than I expected from the other samples I've seen. I thought I saw a full stop difference between the 5D and 1D3, too, but this difference looks a lot smaller. There's a real and useful difference at ISO3200, but I don't know that I'd use ISO6400 from the 1D3 given how much more noisy it is than the 5D. You'd have to absolutely nail the exposure and probably shoot RAW to make good 8x10's out of the 1D3's ISO6400.

It's interesting that the 1D3 retains about the same amount of detail relative to the 5D (ie: VERY slightly less) in all the tests. The 5D has been pegged in the past to use more aggressive noise reduction than the 1D2 and 1Ds2, leading to a bit of softness at high ISO versus more noise and more detail.

Since the 1D3 follows the 5D's results, you could conclude that part of the noise improvement in the 1D3 is due to more aggressive noise reduction in camera.

Interesting stuff!

csd2020 wrote:
Thanks for this comparison. I'll be looking forward to more of these as early adopters start posting samples. FWIW I profiled these in Noise Ninja and it reported the following indexes:

5D-1600: Noise Index 33, (15 lum, 18 chroma)
1D3-1600: 32 (16,16)
5D-3200: 48 (24,24)
1D3-3200: 47 (22, 25)

Certainly good numbers for the 1D3 but not 1 or 2 stops as some have reported.

Thanks csd2020, how does these numbers in Noise Ninja translate into how many stops of difference?

Thanks.

Just call it 0.1 stops. Realistic given the ISO capabilities of the 5D. Please somebody, show me a file that shows a 1D3 file at ISO 1600 equal (detail and noise) to a 5D or 1D2N file at ISO 400.

Will Patterson wrote:
I like the way the 5D's look a little better. Impressive! Looks like the 5D's still got it!


yup-- but them again it's going to be near impossible for any camera to come along and smoke the 5d in terms of pure IQ. I'm sure at best is we'll see small minor improvements at great cost. With 5d prices where they are-- I think it's by far the best bargain out there for non-sports shooters.Then again, i'm sour cuz the store said I slid to #160 in the waiting list for the 1d3 ;(!

I had a deposit on it and that many people paid in full already to jump ahead of me.

I'll be lucky if that gets filled before xmas.

mark petri wrote:


Then again, i'm sour cuz the store said I slid to #160 in the waiting list for the 1d3 ;(!

I had a deposit on it and that many people paid in full already to jump ahead of me.

I'll be lucky if that gets filled before xmas.




which store was that?

csd2020 wrote:
Thanks for this comparison. I'll be looking forward to more of these as early adopters start posting samples. FWIW I profiled these in Noise Ninja and it reported the following indexes:

5D-1600: Noise Index 33, (15 lum, 18 chroma)
1D3-1600: 32 (16,16)
5D-3200: 48 (24,24)
1D3-3200: 47 (22, 25)

Certainly good numbers for the 1D3 but not 1 or 2 stops as some have reported.

Any numbers for the ISO 6400 shots?

My visual observation puts the difference at about 1/3 stops at 6400, perhaps less at 3200. I had expected a 1-stop difference at 3200 but these images don't show that.

As a 5D owner, that pleases me since I don't have quite the strong tug pulling me towards another body (granted, there are many other features that give the Mk III an advantage, but one does have to ask their self if those features are needed for their own style of shooting).

kylee wrote:
my initial impression is 1D3 is a very pro body 5D

This comment appears to be very close to dead on and indicates that Canon is and has been near the point that it will take a radical change in technology to produce a major improvement in quality. This makes things very interesting for their marketing department.

I believe that the laws of physics dictate that further advances in resolution vs. noise will be very, very difficult. The 1D3 enjoys a modest gain in pixel density over the 5D and produces but a small improvement in noise performance. The steps will be smaller and smaller.

The next rendition of the 5D might well share that pixel density, just as the 5D now shares the same density as the 1D2n. It's noise characteristics will probably show a similar small improvement as the Mk III does/did over the Mk II.

Tom_W wrote:

Any numbers for the ISO 6400 shots?

5D - Index 78 (38 lum, 40 chroma)
1D3 - Index 74 (36 lum, 38 chroma)

Sorry for not including this in earlier post.

So the bottom line seems to be: the Mark3 has really great high ISO performance, even a bit better than the 5D.

But considering this test the claims made elsewhere ("1-2 stops better than any other Canon camera") look way overhyped. It's very good, but it's not miles ahead.

brunobarolo