Color: lens or body ?

When shooting RAW, and processing in LR, what is more responsible for the colors - camera or body ?

I am not talking about extensive color play in post. Say you take some shots, bring them in LR, and only adjust exposure and WB. Would the way the colors have a certain "look" at this point be mainly due to the camera handling of raw data, or mainly due to the lens rendering of these colors ?

I understand that it's a function of both factors, but one is probably more important than another ?

I used to think that it's body. But now I am not sure.

Someone gave me an old Konica Minolta 7D with Sigma 12-24 (the original version) in Sony / Minolta A-mount. When shooting on a bright summer day, the resulting photos would have that unique golden "glow" to them. Having read much about "Minolta colors" I attributed it to the way KM processed / presented their RAW data.

However now that I got an adapter to use the same lens on my Sony Nex (that has a different color feel than KM) I see that I still get the same "golden" feel in photos. A bit on the cooler side but bumping the WB slider in LR just a tad to the right produces the same unmistakeable golden tone. (If I try this with shots done with any other lens, I just get a yellowish photo).

So, it is the lens ?

You are forgetting the third factor...the Raw converter, and within the complex chain of processing (demosaicing, color space conversions, etc.) first and foremost the camera profile. Theoretically the function of the profile is to describe how the camera reacts to color and to act as a basis upon which to translate the capture to accurate color. However, profiles are also frequently used as vehicles for intentional color alterations and inaccuracies (compare Camera Neutral and Camera Landscape). Moreover, the limitations imposed on Adobe in creating profiles can mean that similarly named LR profiles for different cameras may not be equally accurate or efficient.

We never see an unprocessed image, so processing has to be added as the third factor affecting color and it cannot be eliminated as consistent from body to body or lens to lens. There is software for analyzing unprocessed Raw data, but it is not a viable alternative for most of us. So separating the three factors is a bit like knowing what happens inside your refrigerator when the door is closed.

tzalman wrote in post #16455557
You are forgetting the third factor...the Raw converter, and within the complex chain of processing (demosaicing, color space conversions, etc.) first and foremost the camera profile. Theoretically the function of the profile is to describe how the camera reacts to color and to act as a basis upon which to translate the capture to accurate color. However, profiles are also frequently used as vehicles for intentional color alterations and inaccuracies (compare Camera Neutral and Camera Landscape). Moreover, the limitations imposed on Adobe in creating profiles can mean that similarly named LR profiles for different cameras may not be equally accurate or efficient.

We never see an unprocessed image, so processing has to be added as the third factor affecting color and it cannot be eliminated as consistent from body to body or lens to lens. There is software for analyzing unprocessed Raw data, but it is not a viable alternative for most of us. So separating the three factors is a bit like knowing what happens inside your refrigerator when the door is closed.

Thanks, this makes sense. However the reason I brought it up is that I shot this lens on an old Konica Minolta body, and loved the golden glow it produced. Put it on Sony (which tends to be on a colder side) and with little tweaking got the same golden glow. So I suspect the lens is also very important in what the final color looks like.

Do you have a sample of this 'golden glow'? I'd love to see it.

It really is a composite of

1. lens
   
2. camera body
   
3. postprocessing settings
   

• If I take a photo with my Tamron lens, it is warmer than my Canon lenses at the same WB values
  
• If I take a photo with my Canon 17-55, it seems just a tiny bit more neutral than my other Canon lenses at the same WB values
  
• If I take a RAW with my 5D and use the exact same Lightroom settings as my 40D, the 5D image is noticeably warmer than the 40D image; but I can neutralize both images to the same rendering -- simply not with identical post processing values!
  

You are basically asking if the image quality, in this case color, would be different in between bodies and lenses. The answer is no. As long as you are shooting with a decent body with a good quality lens, the color or image quality is the same. The ISO noise would be different under a higher ISO setting. Generally, a raw file from a 5d mark I is not much different than a raw from the 5d mark III. Assuming the ISO setting is average in the range of 100 to 800. EVERYTHING comes down to the post processing. The more you want your shots to standout, the more involved your post processing becomes.

This is all speculation... but

In my experience the largest difference comes from lenses.

Usually the difference manifests itself on blue/yellow or warm/cold scale. The way the colors are rendered and I think I can see it with different bodies and raw converters.

I think it comes from multicoating and how light gets reflected inside the lens. When you shoot a blue/red scene, in a low contrast lens the red light bouncing around would make the sky less blue. The blue light from the sky would make the flower less red.

But some lenses bounce blue more than red, so reds are muted and vice versa.

It's all very subtle, but zeiss lenses produce great blues... more so than other lenses.

DocFrankenstein wrote in post #16456828
This is all speculation... but

In my experience the largest difference comes from lenses.

I shot these a very long time ago under artificial light, to avoid the slight change in daylight from passing clouds. A;ll have identical postprocessing values in LR:

First the Tamron 28-75 and then the Canon 17-55

Wilt wrote in post #16457270
I shot these a very long time ago under artificial light, to avoid the slight change in daylight from passing clouds. A;ll have identical postprocessing values in LR:

First the Tamron 28-75 and then the Canon 17-55

What's your conclusion from the results?

I have a video of a conversation between Jeff Schewe and Eric Chan, head of the LR/ACR design team (an appendix to the Luminous Landscape's tutorial on LR4) in which they discuss LR's DNG Camera Profiles. I haven't reviewed the video now, but to the best of my memory it went something like this:
Chan: All we can do is to bring production model cameras to our lab and profile them under controlled lighting to make the dual-illuminant profiles.
Schewe: So you get a bunch of units of a new model from the manufacturer or another source?
Chan: (Little laugh) It is often just one unit.
Schewe: So the profile could fall anywhere on the bell curve?
Chan: (Somewhat sheepishly) Yes.

I have no idea how wide that curve is, how much variation there is between units of any particular model. Of course there is some, there are tolerances for any physical device. That's why those who are meticulous about color make their own profiles. Thus the possibility at least exists that Adobe profiles are inconsistent in their representing the "average" camera. I wonder if Canon profiles are any more reliable; they are still generic, of course, but they certainly should be able to test a bigger sample. (My first digital cameras were Minoltas and I remember that about a dozen years ago a fellow who was designing a Raw converter for them managed, after a long struggle, to get an SDK and pull apart Minolta Raws. He was amazed to discover that they contained an embedded profile that he believed was unit specific. IOW, Minolta was profiling each camera as it came off the production line. I have never heard anybody suggest the same for Canon.) Wilt's comparison of a 40D and a 5D in LR is interesting. I think I will try a similar comparison of a 40D and a 5D2, using both LR and DPP. Watch this space.

Here's my test:
Lens = Canon 100 mm f2.8L IS Macro
Exposure = 1/40 @ f/4 @ ISO 400, metering the white and therefore 2 stops under.
In both converters WB was set by probing the white area in the center of the card.
In DPP 3.13.0.1 the 5D2 shot got +2 Exposure and the 40D got +1.75, while in LR 5.3RC the 5D2 was +2.25 and the 40D was +2.15. This equalized the white area in the center at 236-238. (The corners of the 5D2 are darker because it is FF and therefore has more light fall-off in the corners.) All other converter settings were either zeroed or turned off. Picture Style and LR profile were Camera Standard. Images were exported as tiffs in Adobe RGB (because LR soft proofing reported some of the color patches as out-of-gamut for sRGB.) The screen capture was assigned my monitor space and then converted to sRGB.

DocFrankenstein wrote in post #16457544
What's your conclusion from the results?

1. There is a fundamental difference between lenses.
   
2. There is a fundamental difference between sensors.
   
3. There is a fundamental difference between manufactures.

Elie,
Might I suggest shooting a section of some paper currency with complex coloring? I have used the ColorChecker for many tests including a number of tests between 40D and 5D in the past, but was nevertheless especially surprised later in the rendition of currency with the same LR settings.
I later determined that, when using a grey card to balance the shots, the differences between 40D and 5D became much smaller ones. One can certainly see model-to-model differences if one downloads RAW files from DP Review with tests that use identical targets.

First Nikon D700 and D5100...

Generally what is most responsible for color when developing a RAW file is the converter program and the profile that is used during said developing!

Wilt wrote in post #16457883
Elie,
Notice particularly the easily perceived differences seen in the Kodak color card above in each shot, vs. the Colorchecker below.

However I think the exposure is making more of a difference in the color than the color is being displayed differently. Exposure affects how colors are displayed!