I can see why Nikon dumped Sony sensors!

olliess wrote:
Jack Hogan wrote:

if you take a D600 and a 6D in controlled conditions, they will produce remarkably similar colors?
If so, then wouldn't the DxO measurements suggest that both cameras are both inaccurate in remarkably similar ways?
Why are you assuming they are inaccurate? They are not, the SMI is a very rough index, a difference of a few points means very little. Check on the other hand the Color Depth and Portrait scores:

e0c25c1574a64ebcaf528f99d84ba5be.jpg

If you are worried about color accuracy get yourself an X-rite Color Checker Passport and the Adobe profile builder. I used it a few times but had such a hard time noticing the difference that I stopped using it - for my uses (advanced amateur) it is way more trouble than it's worth imho.

Jack
 
Last edited:
FTH wrote:
Cacophonic Visions wrote:

Color response/accuracy isn't as much the sensor as it is the CFA placed in front of it. As someone said earlier, the color profile can even be modified in post.

Saying "all" Sony sensors this generation have that color problem is a bit broad a statement, more accurate would be to limit it to the ones with Nikon CFAs on them, as the ones with Sony's CFAs don't have the inaccurate color problems you're blaming the sensors for. But hey, at least it's not as inaccurate as Canon's 6D's color response.

But go ahead and keep telling yourself it's the sensor's fault for the inaccurate colors if it makes you feel better.
 
Cacophonic Visions wrote:
FTH wrote:
Cacophonic Visions wrote:

Color response/accuracy isn't as much the sensor as it is the CFA placed in front of it. As someone said earlier, the color profile can even be modified in post.

Saying "all" Sony sensors this generation have that color problem is a bit broad a statement, more accurate would be to limit it to the ones with Nikon CFAs on them, as the ones with Sony's CFAs don't have the inaccurate color problems you're blaming the sensors for. But hey, at least it's not as inaccurate as Canon's 6D's color response.

But go ahead and keep telling yourself it's the sensor's fault for the inaccurate colors if it makes you feel better.
 
Cacophonic Visions wrote: It's just too bad Nikon hasn't figured out how to engineer a CFA and rendering to get the colors from their Sony-made sensors the way Sony does from their own.
These things are decided in focus groups, people. The hardware is not the main factor (at least for Bayer CFA DSLRs of the same generation). Perhaps Sony photogs chose Sony after looking at a few OOC images and liking them? Likewise photogs that own competing brands?

Then each went home, put their Raws through LR standard, got something completely different, all hell broke loose and they decided to start this thread :-)
 
Last edited:
I think that the D600 images have a more pleasant color to them...at least to my eyes on my computer screen.
 
Jack Hogan wrote:
FTH wrote: I am very happy that Nikon finally got the rid of Sony and moved to Toshiba - colors look fair more accurate - no almond paste effect, less magenta and warm tones over-saturation.
FTH, has it occurred to you that a good number of people here (including me) prefer the LR D600 profile to the D5200's in the example in the OP? And that if you were to ask participants in this thread my guess would be that your chances of people preferring one LR rendition versus the other would be no better than what you would get by tossing a coin? And how many would have preferred Nikon's rendition as shown by Mako?

And that perhaps the change you see from CCD days is due to the fact that when Nikon developed the standard and uniform-across-bodies Picture Controls, focus groups of people just lilke you chose the current look? And that you being one who instead prefers his skin pleasingly but unnaturally pink, would have been in the minority then and therefore disappointed now?

And that these color differences have very much to do with rendering recipes and very little to do with sensor performance? So who cares whether Nikon chooses fab A vs fab B as long as they produce excellent sensors? Which both of these definitely are, but the jury is still out on Toshiba banding?
did it occur you why medium format is still using CCD?
 
FTH wrote:

did it occur you why medium format is still using CCD?
Because they do not make CMOS chips that big and CMOS is cheap to produce but very expensive to design while CCDs are the opposite and the market for medium format is very small?
 
Last edited:
Jack Hogan wrote:
olliess wrote:
Jack Hogan wrote:

if you take a D600 and a 6D in controlled conditions, they will produce remarkably similar colors?
If so, then wouldn't the DxO measurements suggest that both cameras are both inaccurate in remarkably similar ways?
Why are you assuming they are inaccurate? They are not, the SMI is a very rough index, a difference of a few points means very little. Check on the other hand the Color Depth and Portrait scores:
They are not inaccurate even though an index of accuracy shows them to be relatively inaccurate (even if roughly)?

How would a favorable Color Depth score speak to this question?
If you are worried about color accuracy get yourself an X-rite Color Checker Passport and the Adobe profile builder. I used it a few times but had such a hard time noticing the difference that I stopped using it - for my uses (advanced amateur) it is way more trouble than it's worth imho.
I'm not deeply worried, but I am curious. How would you "fix" metameric errors originating in the sensor with a post-capture profile?
 
FTH wrote:
Jack Hogan wrote:
FTH wrote:

did it occur you why medium format is still using CCD?
Because they do not make CMOS chips that big and CMOS is cheap to produce but very expensive to design while CCDs are the opposite and the market for medium format is very small?
absolutely not. CMOS is actually cheaper to produce
Yes, but very expensive to design, as I mentioned above.
 
Jack Hogan wrote:
rkumar wrote: It is not without reason that Nikon's Capture NX2 software has a "D2X" color option.
Except that that option has nothing to do with hardware...
Nikon D2H 75

Nikon D50/D70s 84

Nikon D200 84

Nikon D2X/D2Xs 83

Nikon D80 79

Nikon D90 82

Nikon D300 84

Nikon D300s 77

Nikon D7000 78

Nikon D5100 78

Nikon D5200 79

Nikon D600 77
You forgot the best one of them all: D40 = 85. Of course we forget to mention that it also has one eighth the color resolution of the D600 :-)


Nothing to do with hardware, but Nikon seems to know as well that the later cameras (D300s, D7000 etc) don't get the color as well as D2X.



D40 = 6 mega pixels, D600 = 24 megapixels

4x megapixels, 2x linear resolution. I don't see one eighth. Anyway it is not a good comparison since one is DX and one is FX.
 
FTH wrote:
Jack Hogan wrote:
FTH wrote: I am very happy that Nikon finally got the rid of Sony and moved to Toshiba - colors look fair more accurate - no almond paste effect, less magenta and warm tones over-saturation.
FTH, has it occurred to you that a good number of people here (including me) prefer the LR D600 profile to the D5200's in the example in the OP? And that if you were to ask participants in this thread my guess would be that your chances of people preferring one LR rendition versus the other would be no better than what you would get by tossing a coin? And how many would have preferred Nikon's rendition as shown by Mako?

And that perhaps the change you see from CCD days is due to the fact that when Nikon developed the standard and uniform-across-bodies Picture Controls, focus groups of people just lilke you chose the current look? And that you being one who instead prefers his skin pleasingly but unnaturally pink, would have been in the minority then and therefore disappointed now?

And that these color differences have very much to do with rendering recipes and very little to do with sensor performance? So who cares whether Nikon chooses fab A vs fab B as long as they produce excellent sensors? Which both of these definitely are, but the jury is still out on Toshiba banding?
did it occur you why medium format is still using CCD?
CCD has always been generally superior given good light, but CMOS is the better choice where low light performance is required. That's the main difference in practice.

This thread though is about CMOS sensors.
 
olliess wrote:

They are not inaccurate even though an index of accuracy shows them to be relatively inaccurate (even if roughly)?
Turns out the index is not very precise nor meaningful by itself...
How would a favorable Color Depth score speak to this question?
... so it is better to complement it with other data which together help to bring it closer to what photographers mean when they talk about color performance.
I'm not deeply worried, but I am curious. How would you "fix" metameric errors originating in the sensor with a post-capture profile?
Are you going technical on me? :-) If the color is not there it is not there. On the other hand it may be there but the rendering is not doing it justice. Therefore if one really wants the most accurate colors that their camera can produce, they can take a picture of a standard target and have a custom profile built for it - in a process very similar to caliibrating/profiling a monitor.

But you knew all that ;-)

Jack
 
Jack Hogan wrote:
olliess wrote:

They are not inaccurate even though an index of accuracy shows them to be relatively inaccurate (even if roughly)?
Turns out the index is not very precise nor meaningful by itself...
[...]
Are you going technical on me? :-) If the color is not there it is not there.
Well I am honestly asking, if you take the above points together with the slow downward drift in recent Nikon cameras' measured SMI, are we likely losing some (more and more) meaningful color information, or is it purely an artifact of the way the index is measured?
On the other hand it may be there but the rendering is not doing it justice. Therefore if one really wants the most accurate colors that their camera can produce, they can take a picture of a standard target and have a custom profile built for it - in a process very similar to caliibrating/profiling a monitor.

But you knew all that ;-)
You're too kind. ;-)
 
rkumar wrote: I don't see one eighth. Anyway it is not a good comparison since one is DX and one is FX.
It is admittedly a bit of a stretch, hence the smiley, but I was referring to the 24.2 vs 21 bit color depth as shown in the table a few posts up.
 
olliess wrote:

Well I am honestly asking, if you take the above points together with the slow downward drift in recent Nikon cameras' measured SMI, are we likely losing some (more and more) meaningful color information, or is it purely an artifact of the way the index is measured?
I believe it has more to do with the fact that manufacturers are making the CFA lighter in an attempt to give us cleaner images: it is a bit of a tradeoff, but with a fast processor it does not necessarily mean worse color performance for our purposes. What's the SMI of our eyes? I definitely would not base any buying decisions on a slightly better SMI: it's just a rough indication.
 
Cacophonic Visions wrote: It's just too bad Nikon hasn't figured out how to engineer a CFA and rendering to get the colors from their Sony-made sensors the way Sony does from their own.
These things are decided in focus groups, people. The hardware is not the main factor (at least for Bayer CFA DSLRs of the same generation). Perhaps Sony photogs chose Sony after looking at a few OOC images and liking them? Likewise photogs that own competing brands?

Then each went home, put their Raws through LR standard, got something completely different, all hell broke loose and they decided to start this thread :-)
You are exactly right Jack!

People keep on comparing cameras based on lightroom and adobe raw. They should call it "how good or bad adobe renders a raw image". Adobe raw does not render Nikon images well at all.

If people want to compare raws, then use Capture nx2 (not to be confused with view nx2) and you'll see the real results on how Nikon cameras really compare.
 

Keyboard shortcuts

Back
Top