[Bf-committers] qdune review

Joe Eagar joeedh at gmail.com
Wed Nov 7 15:28:32 CET 2007


I implemented a simple form of bucketing for DSM.  I've been meaning  to 
clean up that code and integrate it into the renderer, yet I cannot get 
SVN merge to work, so its kindof on hold.

Also, much as I hate to say it we probably should look at Pixie.  Maybe 
even examine its source code; perhaps we could steal stuff from there 
without trashing all of eeshlo's work.

Joe

Ton Roosendaal wrote:
> Hi,
>
> Brecht, Campbell and I sat down for an afternoon checking on the status  
> of the micropolygon (qdune) renderer. Although we agree it's extremely  
> useful and has superior features, the integration of this system (i.e.  
> to use all required Blender features) will take so much additional work  
> that we decided to put this on hold for a while.
>
> The immediate short term target for Brecht is to get the particle fur  
> system to render, using improved storage and rendering tricks, but  
> within the current rendering system. We also will look at using some  
> kind of bucketing (generation of geometry data per tile).
>
> This is a summary of the status of qdune integration now:
>
> - added exporting API, shared by Blender internal as well as qdune.  
> Exports mesh/curve/surf geometry.
>    (note, this is similar to the GSoC project, but different  
> implementation)
>    (note, it can be used for other exports too)
>
> - added qdune code for preparing rendering data to be handed over to  
> Blender internal shading
>    (texture coords, lamp coords, etc)
>
> - almost working: shadowbuffer rendering
>
> - to be done: irregular buffers, halos, envmap, renderlayers, passes,  
> lamphalos, threads
>    (main reason for postponing it)
>
> - to investigate: speed of the qdune system is below expectation  
> (slower than Blender internal in comparable shots). Maybe the biggest  
> gain is in more complex situations
>
> - to be done longer term: raytracing (AO, shadow, mirror), panorama  
> render, ...
>
> To be able to make a decision, we also listed the pros and cons for  
> adopting a micropolygon system in Blender:
>
> Pro:
> - Displacement mapping, motionblur
> - Unified render (transparent, solid)
> - Support for big scenes, because of efficient bucketing
> - Better anti aliasing (for shading too)
> - More efficient shading (cache misses)
>
> Con:
> - Raytrace is hard to integrate (AO, softshadow)
> - Still relatively slow compared to Blender
> - Lots of work still to get it on functional equal level...
>    (shadowbuffers, envmaps, halos, lamphalos, render layers, passes,  
> threading, ...)
>
>
> Brecht definitely is willing to work on further integration later.  
> But... would help if we could find a partner for him to check on the  
> integration work and maintenance.
>
> -Ton-
>
> ------------------------------------------------------------------------ 
> --
> Ton Roosendaal  Blender Foundation ton at blender.org  
> http://www.blender.org
>
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers
>
>   



More information about the Bf-committers mailing list