note: this picture is bad, because I use “color only” effect when I make that “radiosity” effect in Photoshop
This is similar to the AO algo.
For myself, effect like radiosity need a better algo than that. it’s too simplistic. How would you cast light in a cube from side to side? Now what you expose will only do stuff similar to AO…
That’s how i see it.
That’s actually much faster / lower quality than AO. What endi is proposing is more of a post processing filter using the zBuffer for color bleeding. It’s a scanline technique while AO uses raytracing.
One way to test the method would be to do it as a sequence plugin, since those have access to the finished image buffer data (including zBuf).
I would be nice to get AO type effects with scanline renders.
oh, please, this is a fake thing, but maybe useable
and about sequence plugin: I hate it, like Zblur… I want it in the renderer
very clever idea and I agree that sequencer plugins are just not that practical.
go for it
Agreed, that’s why I said “for testing”. It’s much easier and faster code wise to test something like that as a sequence plugin that to dive in the render pipeline.
plugins python/sequencer are grand for development.
But wouldn’t the code be inherenty structurally different if it’s in the render pipeline?
I think it depends on whose willing to code it.
In this case, not really since it’s mostly a post render filter anyway (as I see it).
maybe somebody wants to make…
blender’s love for faking (and making it faster!) cannot be betrayed: this seems a good idea and could be useful especially for adding lightning depth in animation, where we need short calculation time and do not bother of phisically accurated results.
so I vote for a sequncer plugin test.
sadly i’m not a coder…
I just had a thought… isn’t this just a small rewrite of the current toon-edge system???
I mean toon edges now are made relative to the zbuffer. only this is more “distributed” and may take the color of the pixel into account.
I’m having a look at the code now… but I’m really not experienced enough I think.
What we’d really want to do is… get the closeness to the edge… or even just the difference in zbuffer to the pixel(s) next to it… and relative to that add “light”. I don’t think there’s actually a need to make it more complicated if we want simple scene lighting based on the zbuffer… to start with.
I’ve seen fake AO shaders that work similar to this, but mostly for shading crevises and the like based on how close they are to other faces in the scene. The closer to another face, the darker they get shaded, so you get those dark corners where not much light strikes, just like in real life. Anyway, I think you may be on to something. I don’t think that fact that it is a ‘fake’ way of doing it should stop anyone. After all, a lot of what is so great about CG is being able to break the rules so as to get a very stylised look to your image. Depending on the artists vision, a completely accurate aproach may not even be desired.
Exactly we’re talking pixels on a screen here… there’s no such thing as “real” there’s just real pixels.
having a look at the code… I can change the color of the edges when using the unified renderer… but haven’t figured out why the regular render doesn’t apply them… very odd.
I think ton may have written this part of the blender source himself… hmmm to bother him or not
What I’d like is a way for lights that cast no shadows to illuminate occluded faces less strongly or not at all based on simple z-depth calculations. 5 or 6 area lights with no raytracing or shadow casting but just some simple occlusion test (normal direction?) to determine how much to light up any given face. It would give nice soft shadows and render very quickly. Just now I’m experimenting with that but I keep having to put negative lights everywhere to darken occluded faces. Maybe if you could manually paint occlusion - after all, I know what bits are supposed to be dark. But that’s a different matter from the colour bleeding effect.
I’ve read of people doing AO with shadow maps but I don’t know if it would use huge amounts of memory.
I keep having to put negative lights everywhere to darken occluded faces
tinted negatives are funky stuff.
Tinted? He didn’t mention anything about tinted negatives did he? But ya negatives are pretty hard to control, at lest in my experience. When I was a newb I used them to get rid of specs lol. %|
BIG Funky bump up!
Care for anyone to try and build this as a seq tooll ?Trying to bake on texture AO to a mesh that is 16000 faves ain’t fun and for the most part has still yet to work for me to save time rendering…
One hopefull user waiting in the rain for this tool
I don’t completely understand the concept, but wouldn’t using reflection lights and negative lamps give better results?