Nikon D3 vs Canon 1D Mk III Showdown

12/31/2007 02:30:20 PM · #1

Very interesting. Seems to be a very "clinical approach" to the subject. It'd be interesting to see the same clinical approach taken by others to see if the results are the same or different:

Nikon D3 vs Canon 1D Mk III Showdown

12/31/2007 03:12:07 PM · #2

I question the choice of focusing with live view. I would think it would be a truer test to focus through the view finder at the optimum distance for the focal length and aperture. Then lock the mirror up. I never really trust the image on the LCD. And since the D200 files are MIA it was far from a controlled test not to mention the construction debris and ambient light. Just sayin'.

I can't afford either camera but would also love to own either so I really have no dog in the fight just don't think it is all that controlled or scientific. They also had no control group image...

Message edited by author 2007-12-31 15:13:38.

12/31/2007 03:20:45 PM · #3

I'm not sure why we came away with two different opinions on how "controlled" the test was...

What I saw in the discussion that gave me the impression that it was very controlled was that: 1) shot with the same exact lens (that's something very hard to control between cameras from different manufacturers), 2) placed on a tripod, at the exact same distance (no movement of tripod or subject), 3) using the same lighting, 4) with mirror lockup on both cameras.

The place where I would still have room for doubt is on the processing end ... I know he says that everything was untouched (no USM, no noise reduction, etc). But I wonder if there is still some variance in how files from the two cameras are processed even if everything is turned off?

Message edited by author 2007-12-31 15:21:37.

Canon EF 70-200mm f/4.0L USM

12/31/2007 03:22:01 PM · #4

Originally posted by dwterry:

But I wonder if there is still some variance in how files from the two cameras are processed even if everything is turned off?

That too.

cujee

Canon EOS-30D

12/31/2007 03:27:02 PM · #5

My friend and I had this debate. He said D3 is the best camera out there. Even better than the mark 1ds III.

12/31/2007 03:27:49 PM · #6

If he used the same software for conversion, then I do believe that's about as close as you'll get to apples to apples. Certainly the use of the TeleVue APO 'scope as the optic is beyond reproach. The test seems pretty well controlled. Would be nice if he'd posted the RAW files for download somewhere.
I'd have preferred if they had used software like DSLR Focus to refine the focus, but that probably would not work here, it's made for astro use and needs a point light soruce. The live view focusing, however, is probably superior to what can be achieved through the viewfinder. It eliminates slight differences in optical path length by relying on the main sensor to directly indicate focus. The TeleVue has a great focuser, so it should not be incredibly difficult to tweak in focus.

Canon EF 70-200mm f/4.0L USM

12/31/2007 03:30:44 PM · #7

Originally posted by cujee:

My friend and I had this debate. He said D3 is the best camera out there. Even better than the mark 1ds III.

I don't have any trouble believing that. For each 'purpose' there is a tool which suits that purpose better. And there are absolutely things at which the D3 is a better camera than the 1Ds Mark III.

I love my 1D Mark III (no ... it's not a 1Ds). But I also love my 5D. And for very different reasons. I use them both and wouldn't trade either for the purpose for which I use it. Now ... if I were offered a 1Ds to replace my 5D ... I'd jump in a heartbeat. :-)

cujee

Canon EOS-30D

12/31/2007 03:32:29 PM · #8

Originally posted by dwterry:

Originally posted by cujee:
My friend and I had this debate. He said D3 is the best camera out there. Even better than the mark 1ds III.

I don't have any trouble believing that. For each 'purpose' there is a tool which suits that purpose better. And there are absolutely things at which the D3 is a better camera than the 1Ds Mark III.

I love my 1D Mark III (no ... it's not a 1Ds). But I also love my 5D. And for very different reasons. I use them both and wouldn't trade either for the purpose for which I use it. Now ... if I were offered a 1Ds to replace my 5D ... I'd jump in a heartbeat. :-)

yeah. He's a nikon man and i'm a canon, so we have our different opinions.

fotomann_forever

12/31/2007 03:38:36 PM · #9

Wow! Just wow!

12/31/2007 03:51:44 PM · #10

Originally posted by kirbic:

The live view focusing, however, is probably superior to what can be achieved through the viewfinder. It eliminates slight differences in optical path length by relying on the main sensor to directly indicate focus. The TeleVue has a great focuser, so it should not be incredibly difficult to tweak in focus.

Really. OK but what about trusting what the LCD is telling you? Could it lie?

12/31/2007 03:58:08 PM · #11

Originally posted by thegrandwazoo:

Originally posted by kirbic:
The live view focusing, however, is probably superior to what can be achieved through the viewfinder. It eliminates slight differences in optical path length by relying on the main sensor to directly indicate focus. The TeleVue has a great focuser, so it should not be incredibly difficult to tweak in focus.

Really. OK but what about trusting what the LCD is telling you? Could it lie?

Not really. The LCD may not get you 100.0% of the way to the best possible focus, but it won't tell you you're in focus when you're not.
The viewfinder is not 100% reliable, FWIW. In order for the viewfinder to be accurate, the optical path length is very critical. It must be precisely the same as to the sensor. It can be thrown off substantially just by using the wrong thickness of focus screen, or omitting shims. Using the main sensor eliminates path error, and using live view at maximum magnification should enable one to zero in on best focus pretty accurately. The only better way would be to use point light sources and an application like DSLR Focus that relies on the output image to tell you whether you've achieved perfect focus or not.

Tamron SP AF 28-75mm f/2.8 XR Di for Canon

12/31/2007 04:06:34 PM · #12

Originally posted by kirbic:

Originally posted by thegrandwazoo:

Originally posted by kirbic:
The live view focusing, however, is probably superior to what can be achieved through the viewfinder. It eliminates slight differences in optical path length by relying on the main sensor to directly indicate focus. The TeleVue has a great focuser, so it should not be incredibly difficult to tweak in focus.

Really. OK but what about trusting what the LCD is telling you? Could it lie?

Not really. The LCD may not get you 100.0% of the way to the best possible focus, but it won't tell you you're in focus when you're not.
The viewfinder is not 100% reliable, FWIW. In order for the viewfinder to be accurate, the optical path length is very critical. It must be precisely the same as to the sensor. It can be thrown off substantially just by using the wrong thickness of focus screen, or omitting shims. Using the main sensor eliminates path error, and using live view at maximum magnification should enable one to zero in on best focus pretty accurately. The only better way would be to use point light sources and an application like DSLR Focus that relies on the output image to tell you whether you've achieved perfect focus or not.

Gotcha.

Bobster

Canon EOS-20D

01/01/2008 11:53:00 AM · #13

Originally posted by thegrandwazoo:

well the higher res of the D3 would give better LCD results? ;)

wavelength

Nikon D300

Nikon AF Nikkor 50mm f/1.8D

01/01/2008 12:43:27 PM · #14

The only thing I question is whether the in-camera sharpening and contrast levels were really set to ) in the Canon. I know they usually come with a default "Canon look" setup, and Nikon defaults to on 0 sharp/contrast/saturation.

The poster only specified that no sharpening was applied in the RAW conversion, so I would say that of course the Canon was sharper. However, the Nikon just looks like it was out of focus from the start. Since this guy was hanging the optic as the lynch-pin, do we know what kind of converter he had to interface with the different mounts? Did that affect focusing? Why not try both with live-view and with normal shutter release to see if the focusing was simply off? Compensating for proper shutter speed would have been adequate to kill any camera shake speculation.

I just have trouble meshing this with all the pro reviewers saying just the opposite is all, seems that the testing methodology simple relied too much on the optic, which if not properly tuned for the cameara (back-focus anyone?), then would kill the whole thing.

Canon EF 24-105mm f/4.0L IS

01/01/2008 01:10:23 PM · #15

Originally posted by wavelength:

However, the Nikon just looks like it was out of focus from the start. Since this guy was hanging the optic as the lynch-pin, do we know what kind of converter he had to interface with the different mounts? Did that affect focusing? Why not try both with live-view and with normal shutter release to see if the focusing was simply off? Compensating for proper shutter speed would have been adequate to kill any camera shake speculation.

I just have trouble meshing this with all the pro reviewers saying just the opposite is all, seems that the testing methodology simple relied too much on the optic, which if not properly tuned for the cameara (back-focus anyone?), then would kill the whole thing.

It does, but it doesn't... it really looks like there's just a stronger AA filter in play. The D3 does seem to lose a *lot* of detail as ISO is increased, and it shouldn't, unless NR has been introduced in hardware, between sensor readout and writing the RAW file.
There's no way we'll ever really know whether focus was "nailed" with both cameras, we'll jut have to take it for what it's worth. It's not really that difficult to optimize focus under relatively controlled conditions like this.
The method of attachment to this type of optic is a T-adapter and T-ring. The back-focus is compensated using the telescope's focuser. Given that the Nikon and Canon only differ by about 2mm in back-focus distance, the focuser can very easily accommodate this.
The other way to approach this would have been to use a non-G Nikon lens, which would be just as viable on the Canon body, albeit without AF.
When all is said and done, this is only one test, and if it doesn't jibe with other test results, then we can dismiss it.

Prof_Fate

01/01/2008 06:35:18 PM · #16

Originally posted by kirbic:

The RAW file is by definition untouched by even the in-camera de-mosaic algorithm, so no in-camera settings (other than ISO) would affect the outcome. Using the same RAW converter takes away another variable.

...

The D3 does seem to lose a *lot* of detail as ISO is increased, and it shouldn't, unless NR has been introduced in hardware, between sensor readout and writing the RAW file.

I've read that some camera now does apply noise reduction to the RAW files in camera, and there is nothing one can do about it. I can't recall what camera, but since most of what I've read lately has been about Nikon...I suspect that's who's doing it.

There's no way we'll ever really know whether focus was "nailed" with both cameras, we'll jut have to take it for what it's worth. It's not really that difficult to optimize focus under relatively controlled conditions like this.
The method of attachment to this type of optic is a T-adapter and T-ring. The back-focus is compensated using the telescope's focuser. Given that the Nikon and Canon only differ by about 2mm in back-focus distance, the focuser can very easily accommodate this.
The other way to approach this would have been to use a non-G Nikon lens, which would be just as viable on the Canon body, albeit without AF.
When all is said and done, this is only one test, and if it doesn't jibe with other test results, then we can dismiss it. [/quote]

DrAchoo

Panasonic Lumix DMC-LX5

01/01/2008 08:45:06 PM · #17

Man the ubergeeks on DPReview are really flopping their dicks around in wild displays of alpha maleness over this one...

Nikon AF Nikkor 50mm f/1.4D

01/01/2008 08:51:18 PM · #18

Originally posted by DrAchoo:

Man the ubergeeks on DPReview are really flopping their dicks around in wild displays of alpha maleness over this one...

Discussion
Re-Test

WCpilot

Nikon D700

01/01/2008 08:55:49 PM · #19

Originally posted by DrAchoo:

Man the ubergeeks on DPReview are really flopping their dicks around in wild displays of alpha maleness over this one...

This comment has easily made my day, if not my week. Thank you.

01/01/2008 08:59:22 PM · #20

Originally posted by thegrandwazoo:

Originally posted by DrAchoo:
Man the ubergeeks on DPReview are really flopping their dicks around in wild displays of alpha maleness over this one...

Discussion
Re-Test

Only one problem ... he has totally taken the Canon out of his comparison (he only shows what the D3 does in this scenario), so we have no idea what he is shooting or how it compares. For all we know, the text was a font 30 points bigger and 10 feet closer. ("for all we know") There's nothing to compare. At least the first guy compared the two, kept as much "constant" as possible, and documented everything he did.

Still waiting for a better retest...

01/01/2008 09:00:24 PM · #21

Originally posted by dwterry:

Originally posted by thegrandwazoo:

Originally posted by DrAchoo:
Man the ubergeeks on DPReview are really flopping their dicks around in wild displays of alpha maleness over this one...

Discussion
Re-Test

Only one problem ... he has totally taken the Canon out of his comparison (he only shows what the D3 does in this scenario), so we have no idea what he is shooting or how it compares. For all we know, the text was a font 30 points bigger and 10 feet closer. ("for all we know") There's nothing to compare. At least the first guy compared the two, kept as much "constant" as possible, and documented everything he did.

Still waiting for a better retest...

True. But the discussion about the crop factor and the raw converter used is a good one.

01/01/2008 09:08:02 PM · #22

Originally posted by thegrandwazoo:

True. But the discussion about the crop factor and the raw converter used is a good one.

I whole heartedly agree on that front.

01/01/2008 09:18:04 PM · #23

Originally posted by thegrandwazoo:

True. But the discussion about the crop factor and the raw converter used is a good one.

I guess the guy who did the original test had a few choices... he could have:

1.) kept the distance the same and let us judge the difference in pixel pitch
2.) Changed the distance such that the pixel pitches on the subject would be equal
3.) Changed the distance such that the framing would be the same

Choice 2 would give the best indication of "acuity" differences between the two cameras. He chose (1) because it's by far the simplest to do. Choice (3) would be a logical choice for judging total rendered detail in a scene, but a poor one for judging acuity on the same details.
As it stands, the difference in pixel pitch between the two cameras is something like 18%, which is reflected in the relative size of the text in the 100% crops. The difference in detail rendered is far greater than can be accounted for by this difference. Only time will tell whether others confirm or refute these results.