5D3, 6D and D600 true RAW (dcraw) comparison
/forum/topic/1171136/0

1
       2       3       4       5       end

cgarcia
Registered: Dec 04, 2012
Total Posts: 22
Country: Spain

I have done a proper true RAW comparison (using dcraw only for demosaicing) between 5D3, 6D and D600 using the RAW files from www.focus-numerique.com. These are 100% crops from each camera, comparing aproximately the same subject (thus, the D600 image has the greater area and the smaller comes from 6D). Note that dcraw still doesn't natively supports the 6D, thus these results must be taken with caution (anyway, at least they are useful compared with any jpeg from camera or most jpeg conversions, usually garbage in order to make any conclusion).

Each picture has a label bellow indicating the camera name and ISO level (100-102400, except for the D600 which maxes out at 25600). Left is 5D3, center 6D and right D600. Warning, huge file (32 Mbytes) at 98% jpeg quality:

6D vs 5D3 vs D600

It seems that 6D has appreciable less noise than 5D3, more importantly at low ISO settings, far beyond the improvement expected for the slightly higher pixel size. It closes some of the gap in the shadows with the Sony sensor (D600 has still a bit less noise, though). At high ISO settings the 6D has a bit less noise than D600 (but note that D600 has nearly 20% more pixels, thus at image level they both could be very similar). I suspect that 6D will improve the dynamic range at low ISO settings. Canon is hearing, but needs more time... :-)

I have used the dcraw tool, "upgraded" for supporting the 6D. Using default values, of course no noise reduction, and the white balance calculated by averaging the full image (instead of by the camera). Dcraw conversion (at least without tweakings) maybe is not the best, and artifacts are visible, but indeed has no noise reduction at all.

I'm not sure if the colors are accurate for the 6D, since I manually upgraded dcraw for the 6D, only to take into account the masked pixels at the image border. This could also impact in the noise (I'm not sure) thus these results should still be taken with caution, until dcraw is properly upgraded by its author. Meanwhile the current version can be easily patched by searching the line containing "5920" (the 5D3 raw image width) and adding a new block (e.g. just bellow it) like this:

} else if (is_canon && raw_width == 5568) {
height = 3670;
width = 5496;
top_margin = 38;
left_margin = 72;

(5568 is the 6D raw image width, but the max. usable area is 5496x3670).

You can download dcraw source code from Dave Coffin's page (www.cybercom.net/~dcoffin/dcraw/)



mmurph
Registered: Apr 18, 2004
Total Posts: 2654
Country: United States

Thank you for sharing! I need to take this to my desktop to review on my larger, calibrated monitors. On my laptop.

Thx! And welcome to FM!

Michael



thw2
Registered: Dec 27, 2004
Total Posts: 2791
Country: N/A

Well done, has been waiting for this for a while!



Michaelparris
Registered: Sep 15, 2008
Total Posts: 2148
Country: United States

6d for the WIN...imho



cgarcia
Registered: Dec 04, 2012
Total Posts: 22
Country: Spain


I have improved a bit the dcraw fix. Now it uses the adobe coefficients to correct the colors, taken from 5D3. With this update, the 6D colors almost exactly match the 5D3, which I think it also means that both sensors share a lot of things:

5D3 vs 6D vs 600D (6D colors corrected)

However, this doesn't seems to have changed the 6D noise advantage. By the way, the D600 doesn't match the colors, but... it is in fact not supported by dcraw!. I didn't realize this previously.

Here are the two full pictures at ISO 100 converted by dcraw (6D color corrected) allowing to compare the noise in any place:

5D3 ISO 100

6D ISO 100

Source for the RAWs:

http://www.focus-numerique.com/test-1566/reflex-canon-6d-bruit-electronique-12.html



spdntrxi
Registered: Oct 06, 2006
Total Posts: 405
Country: United States

6D looks good



jorkata
Registered: Sep 02, 2009
Total Posts: 646
Country: United States

'Cleanest' Canon sensor ever (sans the 1DX, I guess).



Pixel Perfect
Registered: Aug 16, 2004
Total Posts: 19548
Country: Australia

D600 and 6D are basically a tie IMO, and 5D III is maybe 0.25-0.5 stop behind after ISO 1600.



Michaelparris
Registered: Sep 15, 2008
Total Posts: 2148
Country: United States

cgarcia wrote:

I have improved a bit the dcraw fix. Now it uses the adobe coefficients to correct the colors, taken from 5D3. With this update, the 6D colors almost exactly match the 5D3, which I think it also means that both sensors share a lot of things:

5D3 vs 6D vs 600D (6D colors corrected)

However, this doesn't seems to have changed the 6D noise advantage. By the way, the D600 doesn't match the colors, but... it is in fact not supported by dcraw!. I didn't realize this previously.

Here are the two full pictures at ISO 100 converted by dcraw (6D color corrected) allowing to compare the noise in any place:

5D3 ISO 100

6D ISO 100

Source for the RAWs:

http://www.focus-numerique.com/test-1566/reflex-canon-6d-bruit-electronique-12.html

The circuit board and the white fabric and the book in some areas seem to have greater detail (not by much) on the 5D MKIII......Just an observation.



MayaTlab
Registered: Jul 28, 2012
Total Posts: 38
Country: France

I believe that for this comparison to be even fairer, it might be necessary to saturate more the red channel in the D600 files.



D. Diggler
Registered: Dec 27, 2011
Total Posts: 4744
Country: United States

Michaelparris wrote:
6d for the WIN


YES!



D. Diggler
Registered: Dec 27, 2011
Total Posts: 4744
Country: United States

jorkata wrote:

'Cleanest' Canon sensor ever (sans the 1DX, I guess).


I wonder if CGarcia could put the 1DX up there against the 6D.



ukkisavosta
Registered: Nov 28, 2010
Total Posts: 417
Country: Finland

Thank you for this information. It is very much appreciated!

I was waiting for something like this, as I recall eosfun pointing out many times that the 6D sensor will not disappoint. He has a tendency to be right in such matters.

Jaakko



ardvorak
Registered: Jan 05, 2005
Total Posts: 463
Country: United States

Great stuff. Looks extremely impressive; thanks for taking the time to do this!

Now, Ralph, go buy one!



Marco
Registered: Jan 21, 2002
Total Posts: 1415
Country: Italy

Thanks for the files!

As I already noted in similar comparisons the 6D is ever so slightly cleaner than mk3 at lower ISOs, but the latter has more microcontrast (slightly crisper rendition of text and you can actually see the texture - pores? - on the background while on the 6D it seems somewhat smoothed out).
Still about the same performance for a given output in my book... nice!!



StillFingerz
Registered: Jul 29, 2010
Total Posts: 2866
Country: United States

Perhaps this crap tech gimmick known as the 6D is actually usable, might sell, be a lower cost vehicle for the masses to move to FF, who'd a thunk it given Canon's horrible gears history, always stickin it to their loyal yet blind user base...shocking, simply flocking unbelievable

I'm in agreement, the 5D3 on some of those images looks a tad crisper, but if the 6D takes to sharpening like the 5D3 does...looks like we have another viable imaging tool...still gonna wait a bit for full reviews tho!



vchowdhary
Registered: Jul 12, 2009
Total Posts: 1190
Country: N/A

StillFingerz wrote:
Perhaps this crap tech gimmick known as the 6D is actually usable, might sell, be a lower cost vehicle for the masses to move to FF, who'd a thunk it given Canon's horrible gears history, always stickin it to their loyal yet blind user base...shocking, simply flocking unbelievable


I thought the 5d Mark 2 , d600 already gave the masses the exact same thing at a better price point. :-)



baltmin
Registered: May 26, 2008
Total Posts: 4
Country: Greece

I think that in this test the difference between 6D and 5D3 is even more noticeable than in my comparison test using capture one pro 7.0.1.: http://www.canonrumors.com/forum/index.php?topic=11279.30

In this test with dcraw the difference seems to be huge reaching almost 1 stop. I hope dcraw makes the job better and this is the real difference between the two!



StillFingerz
Registered: Jul 29, 2010
Total Posts: 2866
Country: United States

vchowdhary wrote:
StillFingerz wrote:
Perhaps this crap tech gimmick known as the 6D is actually usable, might sell, be a lower cost vehicle for the masses to move to FF, who'd a thunk it given Canon's horrible gears history, always stickin it to their loyal yet blind user base...shocking, simply flocking unbelievable


I thought the 5d Mark 2 , d600 already gave the masses the exact same thing at a better price point. :-)


Actually now, 4+yrs later, the 5D2's price is so low it's a killer deal...this wasn't the case at introduction.
As for the Nikon alternative, D600, it seems quite the knock out with a great price also, lets hope they can cure the dust/grease issue with the sensor, I'm guessing it is just a below norm QC problem that will get fixed; a dusty/greasy sensor is no fun tho...




StillFingerz
Registered: Jul 29, 2010
Total Posts: 2866
Country: United States

baltmin wrote:
I think that in this test the difference between 6D and 5D3 is even more noticeable than in my comparison test using capture one pro 7.0.1.: http://www.canonrumors.com/forum/index.php?topic=11279.30

In this test with dcraw the difference seems to be huge reaching almost 1 stop. I hope dcraw makes the job better and this is the real difference between the two!


It's still to early, full specs on the 6D's sensor haven't been revealed yet; that I'm aware of, and I'm wondering how much of the IQ and possible "better DR" is sensor design vs CPU/Firmware (Digic 5+) related? We'll find out at some point if we're seeing the 1st benefits/efforts of Canon's rumored, smaller fab process...



1
       2       3       4       5       end