[Bf-committers] Problem of Rendering Speed

[Yaohua Xiong]

Hi joe, Matt

Thank you for your replys!

> Anyway, I'd also recommend asking these sorts of questions in the
>
> support forums at blenderartists.org, not on this list :)


Hi Matt, I do post this problem in blenderartist forum as you can see
in the initial mail post. The reason why I repost this question in
this developer list is that I expect different kind of answers. In the
blenderartist I expect an answer for an end user---how to solve this
problem at once. In this developer list I hope to get a more in depth
understanding of the problem like the reason behind, how that could be
improved through coding/recoding something because I'm trying to join
the blender developer community.
Currently I plan to do some minor improvements(or modification for I'm
not sure if I do "improve" :) ) on the particle system, I'v read all
the doc that I could found on the wiki, but I think asking what's
wrong behind the problem is a good way learning from other developers
experience and get me familiar with the blender architecture quickly.
So I think this the only place I can post :)

>
> No, 2.5 doesn't have the new raytrace accelerator yet, and lots of
>
> somewhat detailed objects spread over a large scene chokes the current
>
> octree horribly.

but this still not explain the problem. With the first
approach(object+ipo) there are 2011 cars object spread over, with the
second approach there are 2011 particles each rendered as a car object
spread over, so the render time should be the same. but the actual
render time for the second approach is 300 times longer.
I have a explanation for this, but I'm not sure if it's right. Because
there are 600 frames data for the particles, the same particle in
different frame are represented with different objects, so in the
second approach there is not 2011 objects but 2011*600 objects need to
filling into the octree. If this is true it must be an deficiency of
the render system/particle system/both, still I'm not sure.