|
|
I think to minimize computation, tiles would be preferable.
I think of radiosity and photons which are not local at all in
nature: they use large parts of the scene.
For example, having a bunch of render-slaves computers, one
could be master and dispatch tasks as the CPUs become
available: once a CPU is done with is tile, it requests another
region of the picture (without flushing its own data, that'd be
a shame to start from scratch with photons and radiosity.) The
master collects tiles and builds the final picture.
Radiosity is build in a hierarchical way, maybe there would be
some way of exploiting this to speed up things ?
I would much rather see that master/slave thing implemented
in POV than using a script-based dispatcher.
Best,
S.
"Rafal 'Raf256' Maj" <spa### [at] raf256com> wrote in message
news:Xns9504C0FBDA478raf256com@203.29.75.35...
> How about changing main render loop (that goes through each pixel of each
> line AFAIK) into some kind of server-client model.
>
> It would give greate possibiliteis of comfortable distributed render as
> well other possibilities.
>
> If You are changing i.e. finish of some isosurface You could rerender part
> of image. But better solution - to rerender only pixels that "touch" this
> isosurface.
>
> User will select exact are to re-render (i.e. in some GUI or other addon)
> and this add-on will call PovRay engine.
>
> Engine will parse scene, load all textures (and store them in memory etc).
> Then it will recive i.e. commands like
> RENDER LINE 76 PIXELS 64-158
> RENDER LINE 77 PIXELS 60-140
> via some named pipe or other programs communication protocol.
>
> And - importand - if we want to render another part of image (in same
> scene) immediately after, we just send info to Pov-Server to render some
> more pixels, without waiting for parse/load data.
>
> After all we send some
> SCENE DONE
> commend to dispose of loaded scene, and then maybe SCENE PARSE c.pov
> to load another version of scene.
>
>
>
> Second subject - distributed render. We just run several pov-servers, give
> them some way of coomunicating with master client (via LAN, via internet,
> or even via email for very long renders), master program sends scene files
> to each render-server, and assign them parts of image to render. Each
> server reports main client when he done it's part so that client-master
can
> assign new parts to new servers.
>
> Just a qucik thought, but any chance it is worth it?
>
>
> --
> http://www.raf256.com/3d/
> Rafal Maj 'Raf256', home page - http://www.raf256.com/me/
> Computer Graphics
Post a reply to this message
|
|