QUOTE(dave_f @ 01/05/09, 06:23 PM) [snapback]298193[/snapback]
I'm sure I'm not the only person to have thought this!
No your not, the problem is that vector images dont contain tiled lookup so the render becomes extremely tedious to make. In essence the 3d application would need to convert the entire thing into triangle repreaentations of teh vectors.
This has been done, but not on maya. Iv seen 2 implementations of this both for renderman ONLY. Even they are a bit slow, but renderans actually more suited to thsi than most renderers.
You can also just import the curves in and tessellate them. You dont even need to bend this tesselated stack yu can hallways use ray tracing for look up. I can make a example once i get home.
QUOTE
It seems like it SHOULD be a pretty simple bit of programming...
Yes but also VERY VERY slow. You have to understand that the vector image is rasterized by a scanline render. And having random lookup scan line render is often a n*n problem. You could accelerate this with a lookup structure tough. In esence its as hard as doing a vector rendering engine on steroids. Worse some fo teh effects in svg would be impossible to do since tehres no pixel dimesnions and the convolves dont work on unrasterized data.
Because the svg engines in use still today havent become flawless its actually a job for tthousand persons to do it right.
This is a comparable situation to why teh vector renderer is slow.
theres a 3rd alternative, and tahts tu use subimages and rasterize one more optimally each frame.
fourth,
Dont worry about the precicion, as for example if you compare the aa crispness of after effects with a much lower resolution render in mentalray you will find mental ray does a better job than adobe succeeds in doing by analytic render. Your actually deceiving yourself here a bit.