With 3.7, rendering is now lightning fast, provided one has enough cores.
However, parsing speed is still dead slow. For many scenes, this is nor a big
problem, but I do a lot of math based animations where the most important factor
influencing total calc time is now the parsing part.

If I understand things correctly, 3.7 still uses only a single core for parsing,
and the calculations are not compiled. (My phrasing may be wrong as I'm not to
familiar with the program architecture.)

As an example on how bad things are, I recently wanted to determine a list of
the three neighbour triangles of every triangle in a mesh2 object. I first coded
it in Povray for ease of access to the mesh2 files. My mesh2 object has 180,000
triangles, and when I started it I could estimate that Povray was going to take
about 20 hours (!) to do the job.

I then modified the mesh2 file so I could enter it in another application
(Ultrafractal, of all things!),wrote the same code as in Povray, and I got the
job done in 32 seconds!!

Are there any plans to do any work in this field?

Post a reply to this message

Am 24.11.2011 19:29, schrieb Jos leys:
> With 3.7, rendering is now lightning fast, provided one has enough cores.
> However, parsing speed is still dead slow. For many scenes, this is nor a big
> problem, but I do a lot of math based animations where the most important factor
> influencing total calc time is now the parsing part.
>...
>
> Are there any plans to do any work in this field?

Yes.

Post a reply to this message