This paper could make real GI for Project Durian possible

It can give similar to better looking results than photon mapping, but renders a whole lot faster.

The concept and everything needed for implementation is in the paper, it supposedly can render fast enough to be used by animators.

Photon mapping is a tried-and-proven GI method, while this, I don’t know, looks like a newly introduced method that is not tested in real production.

Erm… um…

Its a Pixar paper…

I’m fairly certain we have all seen this already. Farsthary included…but it does seem like a decent work flow(renderer workflow)…

From the looks of thing, and i am no expert on 3d stuff so am probably dead wrong, isn’t this an extension of AAO, which I think was also from a Pixar paper and was pretty new when first implimented in Blender.

p.s Huh! I got to the second page it is based on AAO which makes me wonder if any of you skeptical dudes read the paper before commenting. It could be a useful and cool extension of AAO

I am afraid this will be another lightcuts project for Blender. Still not sure whether it will be integrated in Blender or not.

I’m pretty sure that this paper has come up before and was already implemented as part of the AAO in the code for peach …just “commented out” because it caused some issue or other…

Yeah, the images from surfs up, pirates of the carribean and the fact that its a core feature in prman 14 and 3delight probably mean its not ready for production.

In other news, apart from the smoothness and speed of rendering, a key part of point based occlusion is generating the point cache itself. You need a fast way of dicing your scene into points, ideally taking displacement into account, and points as small as possible. micropoints if you will. like micro…polygons… hmm…

thats why REYES renders are so well suited for point based rendering, they essentially hijack the micropoly core and use it to write out points. I understand micropolys are on the todo list for durian, as such i wouldn’t surprised to see point based rendering techniques get a boost as well.

@The Verb did you actually read the paper or you just trying to be as negative as possible

  1. it has been used in movies pirates of the carribean and surf up are mentioned on the first page considering when that these movies were in production 2005 -2006 I would think that list has grown.

2.its based on the same technology as AAO which is already in Blender

I think there is a higher chance of this coming to blender internal before photon mapping or any other G.I technique considering what MichaelW said about the code been there but commented out. Hell if I was a gambler i would bet on it happening

we’ll see when we see. Until then…I have work to do.

Nice paper, would have been a great GSoC project to extend blender AAO with this. Now I feel that it is better to wait for Farsthary to do his work. There will always be better stuff coming up and I think it is better to complete what is already started and then look for the next thing or you will constantly be throwing away half-completed project for a newer one.

This paper is not sitting commented out in the code, it describes a slower but more accurate method than AAO, that doesn’t suffer from the over occlusion problem.

Well, obviously he didn’t. I hope he doesn’t think that “Pirates of the Caribbean” and “Surf’s Up” are not real productions. And Pixar? Who the hell do they think they are? :spin:

Oh, and don’t mention Renderman or the dreaded “R” word (shhh… ‘reyes’) and Blender in the same sentence or you will get flamed.

Oh I know… I get the evil eye each and every time I post here I swear lol :stuck_out_tongue:

As for the paper and it’s use… well I guess I dont have to go there as it’s already been said. IF you want to really check it out without having to buy Pixars software… go here

Pixie already supports this, though you would have to turn over to the dark side of Blender to RenderMan lol… seriously it works.

animate 78, I respect and support you on your project. Stay encouraged, good luck.

I would love to try the renderman spec renderers, but they seem to require a technical director to use. Hopefully they will become more artist (and idiot :wink: ) friendly in the next few years.

That is the plan. Even though Pixar’s PRMan has been around for a long time, it was not until recently that using it did not require vast experience prior. Renderman for Maya was a great way of bringing more artists to it. I agree, RenderMan is not for everyone, it’s the tools that bridge that gap between artist and technical director.

What really sucks is that after Blender 2.50 arrives, Mosaic will have to be rewritten, so development on this area will only continue in the 2.49 realm until it gets fixed.

AAO and SSGI already form a pretty good solution in my opinion for animation. The photon mapping and Final Gather are a lot of fun to play with two, but still personally more used to SSGI.

(Edit - all for new ideas in the rendering department though… :slight_smile: )

I don’t think ssgi is enough for Durian,it’s better to have a robust solution.

I know that the trend in commercial 3d apps goes towards unbiased rendering and raytraced multi-bounce final-gathered kitchentops but for animation point-based techniques are the bees knees. Or the cat’s meow. They rule, basically. I will always cherish the memory of rendering my first 4K frame with subsurface-scattering, micropoly displacement and pointbased AO and glossy reflections using a renderman compliant engine. If Blender really gets a micropoly renderer I will tour Europe with a Blender Flag, celebrating for weeks.