Don't affect colors problem

[fAEkE]

So, we are doing some movies and need to render to .exr for post. When rendering stills we usually burn in the gamma correction, not using Don't affect colors. The problem now is that we don't get the same results with the two methods.

I have gamma/LUT on in the preferences with Input gamma 2.2 and Output 1.0. I tried using Linear multiply with gamma 2.2 first, and then campared it with a render where I use the same settings but switch on "Don't affect colors" and sRGB in the framebuffer. The result is not the same. What am I doing wrong here?

[fAEkE]

Anyone? Must be alot of you who has worked with both gamma 2.2 and sRGB

[bob-cat]

If you usually "burn" in the gamma, all you need to change is the linear color mapping gamma settings back to 1.0.
Use the sRGB button on the VFB to view the image correctly.

[fAEkE]

No that doesn't work. It produces the same result as using "Don't affect colors" except that the render result has much more noise.

I have tried all this at home and at work. sRGB does not produce the same render as gamma 2.2. Is it supposed to work that way?

[fAEkE]

Vlado, can you please explain this to us? This is a big issue for us and we need it solved as soon as possible.

[Morne]

baking the gamma with the save from max is not the same as baking the gamma with save from photoshop. The one mathematically works out things as it should, while the other works out mathematically different. Saving linear exr (with "don't effect colours" ticked) is the way to go. (ie not baking the gamma at render time save)
AE or Photoshop or Nuke or whatever will interperet your linear (no baked gamma) correctly and display it correctly. From here you can then save your final result. (ie work linear all the way untill you're ready to output the final product)

[fAEkE]

Yes we know.

If I render an image with gamma 2.2 and Don't affect colors off, and then take a print screen of the framebuffer, and then compare it to a print screen when rendered with 2.2, don't affect colors on and sRGB on, they don't match. Shouldn't they match?

[instinct]

I am not sure what you are comparing to be honest. But in general "rendered linear with gamma 2.2 burned in, sRGB OFF in the VFB" should match "rendered linear with don't affect colors ON, sRGB ON in the vfb"

If you are rendering non-linear this will never match a linear rendering as the renderer produces different output (light/shadow propagation, color bleeds etc)

Regards,
Thorsten

[fAEkE]

Im comparing the two methods you are describing and they do not match.

[lllab]

faEke, this problem is normally only if you have some color management problem on your os or software settings, and/or bad calibrated screens, to my knowledge. normally is very much the same. maybe check these things?
stefan

[dlparisi]

How different are the two images? Keep in mind that gamma 2.2 and sRGB are slightly different curves.

[glorybound]

My understanding is sRGB isn't true gamma 2.2. I usually leave everything at 1 (max and v-ray), and apply a curve like this:



which is free on the top right, and 0.60 on the bottom left.

[fAEkE]

@lllab
Hmm.. we have tried on several pcs at work and at home, the same on everyone, so I don't think that is the problem.

@dlparisi
I would say slightly different. The sRGB, when compared in after effects, has more contrast than the gamma 2.2 image. And as far as I know, it should be the other way around?

One more problem we have is that we are getting some really strange results with elements when using sRGB. Rawshadow among others is just... totally screwed up.

[bob-cat]

If you could post some screen grabs of your settings and corresponding renders.
ie. gamma settings in the preferences of max, color mapping settings and VFB.

[instinct]

Agreed.

@glorybound: You are right in sRGB and gamma 2.2 not beeing the same, a manually adjusted curve will definitely be even more off tho. The difference is only very close to 0. sRGB has a linear segment close to 0 (between 0 and 0.055). The rest is gamma 2.2 This is to avoid the curve to be constant 0 for values below 0.

Regards,
Thorsten