Hi all! What you think about this method?
[ATTACH]78962[/ATTACH]
[ATTACH]79052[/ATTACH]
Is this using radiosity rendering in any way? I’d be very curious how you did that and I can’t really tell from the image.
I am sorry for my bad English…is not here Radiosity method…
to all of objects new material is applied with high reflections in the second layer of render…
That’s cute, using the mirror as a guide! but…the cornell box is supposed to show radiosity, how the green cube’s color will radiate onto the red wall. I didn’t see that, but possibly because the mirror was too dark…and you used a multiply. If the red wall is true red, then there is no green in it, so 0 times anything and no green will be added. Try add instead.
Well, it’s work wery good… Only one bounce to compute realistic complex GI 
I did it with Blender internal 2.49!
[ATTACH]99494[/ATTACH]
[ATTACH]99495[/ATTACH]
Preview working fast on low samples in Indirect material…
[ATTACH]99499[/ATTACH]
two bounces
Your technique excites the imagination, and gets me going! I applaud your brilliance!
Thank you:) All that is required for this, it flatten hemisphere glossy reflections to form a disc on the plane … A single reflection for Indirect lighting and color bleeding + fallof reflections to sky color simulate full Global Illumination … (Depth = 0)
man , this is really oool !
we got GI in 2.49 !!!
You should commit this trick ;D
cool , ill try to understand it
if you optimize it as a software shader, it should work very fast … mostly speed suffers from a large number of samples, to get rid of noise … unfortunately I’m not a programmer … might somehow be accelerated and optimized through the PyNodes … I think, the dispersion of light and caustics as possible to get this method …
My method … Colored AO test on Radioroom
[ATTACH]101576[/ATTACH]
Simple raytraced GI and Reflective caustics possible in Blender Internal… May be refractive to…with Mirror Depth>=2 in Reflector material and WardIso specular shader (its produce caustics)…Working in 2.4х 2.5х.
Example:
[ATTACH]114568[/ATTACH]
[ATTACH]114569[/ATTACH]
Sorry for bad English…
This is really clever! I always love seeing unexpected and novel uses of the compositor. Thanks for sharing your method. I go dissect it now.
BI Refractive example
[ATTACH]116016[/ATTACH]
[ATTACH]116017[/ATTACH]
Now if someone converted these methods to C, combined them, and commited it to the render25 branch, then maybe there could be still be a reason to use BI for realism despite the shading refactor not being done yet.
Meanwhile is there a setup yet that takes all these methods and combines them? This looks even better than the render25 branch GI (caustics and no black dots)
There is one method in different ways.
1.Render Pass with glossy material for indirect lighting + compositing (more fast, indirect only from Diffuse Color, or only from UV layer Image textures )
http://3dcenter.ru/forum/index.php?act=attach&type=post&id=153004
Material
2. Node materials connected to group with glossy material for indirect lighting.( slow, more realistic. bumps, alpha…)
Based on the collection of light rendering equation in the hemisphere. To this end, we used one high-gloss material with subtract Z normal to diffuse like AO … Can only be one rebound, as Blender does not support recursion gloss. But this is enough to get a complete Indirect from AO (low samles and hide it from Render Layers for speed) + Direct. GI and simple caustics possible now with this method in BI 2.4х and high. Сonversion to C (global or/and local in material) make it more fast and useful…
This is really interesting! What about render times? If we get close to LuxRender I guess the nodes work is not worth the hassle…
Also: does this technique work with 2.5?
Edit: on a second thought I remember that Lux don’t support fur, smoke, etc., so if this technique could be refined would definitely be preferable.
nice realy good idea! but isn’t computing glossy reflections for all objects in the scene goddamn slow ??
I think ive read somewhere in this forum about someone who managed to fake glossy reflections with Environment maps. that could give amazing fast GI without the need of raytracing at all.
Also: does this technique work with 2.5?
Yes… in 2.5 work more faster…
works with all that is traced ray…
SSS, Fur, Emit (produce realistic texture lights)…etc…no need convert objects like Nurbs, Splines, Texts, Metaballs, Dupliverts to Mesh :)…
This GI method full based on Materials
What about render times?
2-4 times slowly than raytraced AO in equivalent samples
Many intereflections in Moror or RayTransp, supersampling (full sampling), raytraced shadows - slow
Glossy =1 in raytraced materilas = Full overasmpling - slowly… Use Glossy<1 and samles - faster
Buffer Shadows - fast
Soft raytraced Shadows and Area Lights - slow dramatically
Exclude AO pass from Render Layers, and 1х1 sample for Adaptive QMC AO with Treshold 1 - fast
etc…
I think ive read somewhere in this forum about someone who managed to fake glossy reflections with Environment maps
Please post link here. It’s wery interesting…
P.S. with method 2 (node materials) possible Backing GI to texture (Full Render)…
You talk about the limitation with glossy reflections, there’s an ancient technique using nodes that mixes several up to several small noise materials together and creates the impression of a blurry reflection or frosted glass which works around the limitation (ironically possible before blurry reflections were even in Blender)
Frosted glass
Have you been thinking of ways of how multiple normal nodes could fit into this? What about distortion nodes like scale and translate to move around light information?
Thanks, Ace Dragon. Think will be easy to connect, but it is very slow speed,may be… I’ll think about it …
Method on Material Nodes Example
[ATTACH]116304[/ATTACH]
[ATTACH]116303[/ATTACH]
On/Off AO, increase/decrease Glossy, Samples, Threshold, Max Dist in material “Reflector”. Connect yours materials to “GI” node as in example…Look at the result and speed …