Baking an albedo map - color management?

[michael_grosberg]

I'm trying to bake some textures I created in photoshop from one channel to another using Arnold and the Arnold_albedo map (since diffuse and arnold_diffuse seem to bake the scene lighting into the map no matter what I do).

The problem is, I'm getting unexpected results. The result is either too light id gamma is on, or too dark if it's off, and playing around with the gamma value doesn't help either (I have it set to the default 2.2). 

If I try the same thign with scanline and a diffuse map, the result is identical to the map I'm using.

This seems to be some sort of color management issue. Could be I'm missing something about color management in photoshop or Arnold or 3ds max itself? what do I do to set this right?

[madsd]

If you want 1:1 pixel in pixel out, use the Emission channel. set it to 1.0 and render an emission AOV.
It will print the pixels as they came in.

[madsd]

If you want to bake out this kind of emission.
Use the custom bake map.
You can insert an input, it will render it 1:1 like emission, just in the baker.

You can see the lightning and reflections are removed in this map.

[michael_grosberg]

Sorry but that didn't work either. The result is still too dark compared to the source texture. 

[madsd]

Disable the Environment exposure override.
If you need to grade, use the Arnold Imager nodes.

Else it could be you have a custom gamma setting, if so, reset back to default settings.

[michael_grosberg]

"Disable the Environment exposure override."

Can you explain what that is? I have no exposure control in my scene and also no camera - I'm working on a model for a real-time game engine. basically I was trying to bake several existing assets using a number of PBR materials (exported form substance) into a single texture but I can't get a simple 1-to-1 transfer of the source materials into the finalized output. 

I might just end up writing a script to do this using scanline.

[madsd]

Any chanc you can upload a bit scene data?

Its difficult to debug your case, with scene material, like a simplified random object with some quick export that does the same if your original is under nda.

A plane with something random on would be perfect

[michael_grosberg]

I tried creating a test file but can't quite get to the same situation. I know it's something to do with gamma because if I turn gamma off, I get the too-dark output, and it I turn it on in the new file, the output is fine. But I tried turning it on and off before and the result was still not accurate. I think there is some case in which gamma or color correction information is attached to a map when it's created, depending on the gamma value at the time of the map's creation, and this information is not user-accessible but still affects the output.

Now, I found a similar issue in another case. I got it down to a minimal file. in this case, I have two identical materials, created separately, once with gamma turned on and once with gamma turned off. each has a roughness map, and they both point to the exact same file. but if you click "view file" in each of the maps, you can see the map has a different brightness in each case, and the result in the viewport / rendering is different.

[michael_grosberg · Accepted solution]

okay, I found it. finally! the gamma settings ARE available to the user when you click on the image path. for some reason one image had its gamma set to override with a value of 1.0. |I'm guessing this is what caused the brightness variations elswhere, although I still don't know what would cause this override to exist in the first place as I wasn't even aware of it until today so I couldn't have set it myself and then forget about it.

Anyway, case closed. 

[lcCEYT2 · Accepted solution]

When baking an albedo map with Arnold, the material IOR value changes the baked color. To bake to albedo without color changes you need to set the material IOR to 1 before baking.

From AOV Arnold documentation:

"The fresnel in the diffuse_albedo is a result of the specular_IOR and how it affects the albedo of the diffuse to make it energy efficient. So with no specular, there would be no fresnel on the diffuse_albedo, as all the energy would be in the diffuse."