Color Bleeding

[Tamis]

Hey peepz i'm tryng to find a way to render collor bleeding in a seperate pass.

sins i'm using ambiant occlusion i figured i don't realy need a GI solution but i do want some collor bleeding.

Dose any one know a technique to do this?(if they even exist)

[symek]

How do you want to do colour bleeding without GI? Is far as I know you must use GI for such effect. Irradiance mode of Vex Global Illumination light shader will do it for you. Its colour bleeding effect is quite subtle comparing to others engines. There was a number of threads here about it. In one of recently added, Jason gave an advice to turn up Global tint parameter for better bleeding results.

As this is a light shader all you need to do for rendering GI in separate pass is to prepare your surface shaders to be just lambert colours possibly multiplied by textures (I'd would render textures in separate pass but this is your choice - there are numbers of ways to manage passes).

I'm not sure if this can be done but you could also prepare standard shader with a separate component (export parameter) like above and specify in Mantra ROP additional plate with light mask set to light with GI shader applied. The main plate should exclude GI light from rendering. Thus effectively your export parameter will be the only one affected by GI.

I wonder if this is possible though, never tried...

hope this helps,

sy.

Edited February 1, 2008 by SYmek

[Tamis]

Wel the reason i want collor bleeding without GI(global illumination) is because Ambient occlusion is sort of a cheat to fake GI but it dosn't include collor bleeding.

So i was thinking maby there is some other cheat for that, like rendering out a very subtle(do you spell it like that?) ambiant occlusion pass that takes in acount the surface collor of the position of the object it's casting ray's from and then pass it on to the object it's hitting.

Based on the distance the ray traveld it would calculate how mutch collor would bleed.

[hoknamahn]

Wel the reason i want collor bleeding without GI(global illumination) is because Ambient occlusion is sort of a cheat to fake GI but it dosn't include collor bleeding.

So i was thinking maby there is some other cheat for that, like rendering out a very subtle(do you spell it like that?) ambiant occlusion pass that takes in acount the surface collor of the position of the object it's casting ray's from and then pass it on to the object it's hitting.

Based on the distance the ray traveld it would calculate how mutch collor would bleed.

As SYmek said you can use Irradiance (VEX Global Illumination light shader) which is the same as AO with one exception - it returns the color of surface if intersection occurs. I'm not sure does it depend on distance, probably not (at least not old H6/8 occlusion() and irradiance()). But you can write your own irradiance() with intensity dependent on distance.

Edited February 1, 2008 by hoknamahn

[symek]

Well, this should be judged by Masters of course, but what you're trying to do IS the Global Illumination effect in essence - thus it implies all pros/cons of such.

If you want to shoot the ray to check colour of surrounding surfaces, measure distance to it etc, you're dealing with heavy raytracing - the same as you would use irradiance (). The only way out to avoid it I can imagine is to use photons... which is possibly great idea. Personally I love photons and I'm right now in hell for research of them

[Tamis]

arn't photons somewhat the same to ray's?

i alway thought the dirence whas that they got shot from the light and not from the geometry itself?

i might be sayng completly inacurate stuff, but i'm very new to the underlyng tech of this.

[symek]

arn't photons somewhat the same to ray's?

i alway thought the dirence whas that they got shot from the light and not from the geometry itself?

i might be sayng completly inacurate stuff, but i'm very new to the underlyng tech of this.

They are not. I mean they are enough different to not to think about them this way . Moreover I think it's valid to say that they were designed to omit ray tracing which was in those times even slower then today - as you can imagine. They are cheaper to calculate for sure.

So basically photons can help you if you don't want to use irradiance or can help in irradiance loop as well

Edited February 2, 2008 by SYmek

[hoknamahn]

They are not. I mean they are enough different to not to think about them this way . Moreover I think it's valid to say that they were designed to omit ray tracing which was in those times even slower then today - as you can imagine. They are cheaper to calculate for sure.

So basically photons can help you if you don't want to use irradiance or can help in irradiance loop as well

Probably even better to forget about photons too and to use more effective algorithms. Especially in cases where occlusion is enough.