Marmoset 2 - Washed out Maps
polycounter lvl 5
Hello guys
Today I experienced some problem with Marmoset 2. I'm not new to marmoset.
I have an albedo map that works perfectly to other applications. But when its loaded to marmoset it is washed out.
The file format that I used for this one is 16bit .tiff.
I also used other file format like .TGA and .JPG
I also tried using other pictures, thinking that photoshop has the issue.
I hope you can help me with this. Thanks
Today I experienced some problem with Marmoset 2. I'm not new to marmoset.
I have an albedo map that works perfectly to other applications. But when its loaded to marmoset it is washed out.
The file format that I used for this one is 16bit .tiff.
I also used other file format like .TGA and .JPG
I also tried using other pictures, thinking that photoshop has the issue.
I hope you can help me with this. Thanks
Replies
Since we are talking about the SRGB option. I have another question.
I turned on sRGB on the normal map and it ruined it. The seams appeared. I tried Flipping the XYZ, turned off scale & bias, turned on object space. its still off.
Can you guys explain that to me. What is the solution or should I jsut leave the sRGB off. Thanks
While we're on the topic, gloss/roughmaps are typically (but don't have to be) linear as well. Metalness maps almost always linear, and specular maps are often sRGB, but can be linear as well. Diffuse/abledo, roughness/gloss, spec, and metalness can all be either linear or sRGB. The important thing is that you match your color space settings in Toolbag and if your asset is meant for a game, the final target engine (unreal, unity, etc).
I have been reading up on working with linear v. sRGB, but don't quite understand it as well as I'd like to. I noticed that by default, Toolbag 2 has sRGB checked for albedo and specular, and unchecked for metalness and gloss maps. I want to make sure I'm creating my textures using the correct color space, and that I'm making the appropriate decision whether or not to check sRGB in Toolbag 2.
If I understand correctly, if I'm working in an sRGB Photoshop file, I should use the sRGB value from the calibration charts, and if I'm not working in sRGB, I should use the linear values from the calibration charts. And I just need to make sure to keep the sRGB checkbox in Toolbag 2 consistent with the color space in which I was working.
While reading up on color space and color profiles for PBR textures (not necessarily specific to Toolbag 2), I've seen something along the lines of, "albedo maps are usually painted by the artist so they're sRGB; gloss, metalness, and normal maps represent mathematical values, not colors, and are computed in linear space."
However, as far as I can tell, that's not entirely complete or accurate. While using specular workflow means the artist can set any hue for the specular map, the light/dark value of the specular map is still supposed to be based on mathematically calculated data. The same goes for the albedo map when using the metalness workflow. As far as I understand, Toolbag 2 (as well as Unity and UE4) all give options for telling the engine whether a texture is sRGB or linear.
So if there is a reason for using a certain color space for a certain texture type, is that reason still relevant with modern engines able to convert gamma-corrected images to linear for processing and then back to gamma-corrected for monitor output, and with artists able to determine the correct value to use for either color space?
Almighty_gir said: I understand that normal maps are calculated in linear space based on mathematical values, but are my above questions any less applicable to normal maps than other types of maps? I usually bake normals, then copy normal map into a master Photoshop file that includes a separate layer for each map type. I often composite several normal bakes or make handpainted corrections. These corrections all occur in gamma-corrected sRGB space in Photoshop. So when I save out my final normal map, it comes from the same sRGB file as all the other maps.
Pardon my multiple rambling questions - this stuff has been brewing in the back of my head for a while, and I didn't really understand the theory well enough to properly articulate my questions until now.
Thanks much, guys!
This is essentially all that matters, and it looks like you've got a good understanding here.
Yes, these are common and logical reason for picking color spaces, but not necessarily absolute rules. Maps that represent mathimatical values make sense to be in linear space otherwise the distribution of values is uneven(curved) rather than strait (linear) so a sRGB Gloss map would have better precision in the darker values than in the highlights. You may or may not want that, there isn't really a right or wrong here (other than for Normal maps but I'll get to that in a sec).
The reason that diffuse and specular maps are usually authored sRGB is that these maps more so than other maps are painted based on how an artist "sees" them, and our eyes can typically make out more detail in the shadows and midtones than we can in the highlights, which is why sRGB is a thing in the first place.
Yeah, this is quite astute and more or less correct. Every texture map including albedo maps represents a mathematical value. For a diffuse map, it's the % of light in the R G and B spectrum that the surface reflects back to the viewer as diffuse lighting. For a spec map, it's the same but for direct specular reflections rather diffuse reflections. In reality, nobody should be painting their maps simply by look alone, you should use some reference for correct values wherever you can and check in engine under various lighting conditions.
Using the correct gamma space between apps is the most relevant thing here. Again there typically are no steadfast rules like GLOSS MAPS MUST BE LINEAR SPACE. If someone tells you that, they either don't understand how color spaces work, or don't think you understand them.
There are some exceptions however. If you're packing 3 maps into the RGB channels, these maps must all have the same color space in most engines. So if it's gloss/rougness, metalness, and AO in RGB, you would most likely want all to be linear space, but they could easily be sRGB too if you so desired, it's simply important that they're the same color space.
All of this stuff basically does not apply to normal maps because no normal map baker or generator creates normal map content in sRGB space. If that was a thing, the theory above would apply equally to normal maps. Theres good reason for this too, moreso than other maps, normal maps require extremely precise math to do their job correctly, and it would be a bad idea to store that data with curved precision. Normal maps aren't really "human readable/editable" in the sense that other maps are, so the desire to work with them in sRGB isn't really there either.
No worries, these are great questions.
As I think about this stuff I may post follow-up questions, for now just wanted to say great that you point out that using R, G, and B channels separately for separate maps requires all those maps to use the same color space. That hadn't necessarily occurred to me but that totally makes sense, since a single image uses the same color space for R G and B channels.
https://polycount.com/discussion/217320/assigning-textures-to-materials-in-python#latest