Compositing passes in post

Something I have never understood how to do (though I know why one would do it):

Render an image in passes i.e. occlusion, diffuse, reflection, etc. (which I can do with nodes)

And then composite them in photoshop(or other app), so that the end result looks as if you just rendered it without passes.

Is there a certain order in which one must go about combining the layers? Are there certain blend modes to be used?

Any help would be great…

First come the individual layers rendered as separate images. I think this can be controlled from the outliner. There’s the world button and under that is a render button. I think you can select the passes you want.

The passes ought to be rendered with alpha channels so the final comp will have all of the layers on top of each other at full strength. From there its an issue of tweaking each layer.

So to answer your question (I’m thinking in text… and I thought it was bad when I thought out loud): I would start with the color layer at the very bottom followed by the darkest layers (I’m speaking of the contents of each layer) to the lightest. For example you would end with the specular layer.

I hope that helps…

Your q would be an interesting noodle for this page http://wiki.blender.org/index.php/Manual/Render_Passes

I know that the AO pass was tricky to figure out. Some are mixed, some are multiplied. Start simple, and work your way up to include more of the passes. I will monitor this and update wiki as we learn.

Congrats on 2K, btw.

Papa

just a thank you note to you - I see your name poping up here a lot with answering questions.

Also thank you for often helping me :wink:

Claas

You’re welcome, Claas…and everyone. I love the idea behind Blender and this is the way I support it. And relax.

Hi, I do this a lot.

Usually, I don’t try to combine the passes back into the composite, but rather use them to modify the final comp. I render only the passes that are relevant to the render (if there is no reflection, I don’t render that pass) and I always include the “full” render as well (and usually motion vectors and object ID).

I export as multilayer openEXR into either Aftereffects, shake, fusion, or photoshop. I then use passes like object ID to “select” items and apply adjustments (say, if you set a pass to subtract, you can use it to modify the intensity of things).

Sometime I do combine passes into a final comp, but it’s usually because the previous method did not work and so the goal is not to reproduce Blender’s output but to make a “better” output.

This is my standard comping work flow until I learn of a better way.


Kevin

Thanks all for the replies… food for thought…
I’m not bad with nodes, but I’ve never had to do the whole render pass thing before. I’ll keep working on this. Blender.org has some good info, but it’s been down a lot…

In paint shop pro 7 one can save the layers as seperate images, then add a new rastor layer and copy and paste a layer onto the previous one setting the blend mode of the layer over the base.

Compositor is still the best though, in some paint programs it’s all linear like layer 1>2>3>4>Composite

In Blender you can do

1>2>>>>v
4>5>>>>3>Composite