Using / understanding Diffuse pass ?

i’m curious about a thing: why blender’s diffuse pass is multiplied by color pass while shadows and AO passes aren’t ?
Don’t you prefer to look at “raw” light passes with no texture/object color info?

imagine this case: you want to tint all shadows blue and highlights orange, and generally adjust your lighting using nodes, so:
-enable all passes

  • start from color and composite light on it
  • use nodes to tint, contrast,etc… the shadow layer
  • multiply the shadow pass on the image
    at this point only projected shadows are tinted ! self shadowing is unaffected ,
    Because that’s the diffuse pass ( = shaders)

…only, it’s different because it’s multiplied by color …
I solved it by doing " diffuse -divided- color pass " with a mix node.

Isn’t this unmultiplied version much easier to use? For example can be combined with shadows or ao easily , and then multiplied with color.

I’m used to this workflow from vray, that can always output two versions of every light pass multiplied or not (called ‘raw’). i like it, but maybe is there a reason why blender works differently ?

A related question , that i haven’t solved (nor i found infos in the wiki manual)
Diffuse pass is also ‘unclamped’. I mean: if you have strong lights and you enabled exponential or range >1 (in the world options tab) these corrections aren’t applied to the diffuse pass. resulting in a burnt out diffuse pass. (= hdr image with values above 1 )

There are many ways to clamp it, with curves, tonemap node, map value , etc… is there a ‘proper’ way or a way to match exactly the range/exp. settings of world tab ?

I hope these considerations are useful or interesting , and thanks for any info on the unsolved questions.

i realize the whole point of the previous post might be unclear: you might be wondering why completely rebuilding the render from passes doing color+light+ao+spec ?
isn’t it simpler to take the output image and alter it using the channels you want , the way you want ?

Absolutely yes, i do the 2nd almost always. But is very important to know how the render engine assembles the passes , at least in theory because It helps a lot to avoid getting lost with nodes.

There isn’t a guide on that ? or a scheme?

( the post above is about the specific role of diffuse pass )

Of course you’re right. Know how to use the different passes is relevant to compositing.
There are some articles on blenderunderground
For the AO, I guess the mix node you have to use, depends on the AO mode: add, sub, both …
The diffuse pass is the amount of light reflected by the material, it has no shadow.
For the range-exposition I don’t know, I don’t use them.

that article on blender underground looks very useful !

You’re also right about diffuse pass not containing shadows … i was applying a concept from handrawing to 3d but it’s wrong, also self-shadow is a bad translation of a slightly different italian term, nevermind.

Anyway what i was thinking is that diffuse and shadow passes often need to be used togheter (can be called “direct light pass” )
In the example of tinting shadows blue : imagine a house on a field , you want to tint the shadow on the ground (shadow pass) but also the wall facing opposite to the sun (for that you need the diffuse pass )
For me, is often simpler to use the combined “direct light” . no object color infos but everything about light color, intensity, shadows .

Yeah, I understand what you mean. But if you want to give blue tint to the shadows and in the same time to the diffuse pass, it’d be easier to take the combined pass and add a RGB Curves node. You’ll then tint in blue the dark by adjusting the curves.