In article <3l3igvkndgn43fn0s873ac5n040bnh3a6h@4ax.com>,
 ABX <abx### [at] abxartpl> wrote:

> Have you tried to make the same blob2 via SDL as isosurface and compared
> results/speed? Is that possible at all?

It is possible, my original tests used isosurfaces and user-defined 
functions. I haven't bothered to compare the speed before now because 
the blob2 is so obviously faster. There are several optimizations that 
are not possible with isosurfaces. For example, the algorithm collects 
the components that are hit by the ray and usess that list during the 
intersection search, while an isosurface must add every component 
together for every evaluation.

Here's some timings:
tetrablob, blob: 3 seconds (4 with sturm)
tetrablob, blob2: 4 seconds
tetrablob, isosurface: 27 seconds

(Note: the isosurface seems to get some odd surface and edge blemishes 
that I'm not seeing with blob2, I might be doing something with my patch 
that could also be useful for isosurfaces.)

These tests aren't very good for comparing blob and blob2, but more 
complex blobs become impractical to implement as isosurfaces. A blob 
with 500 sphere components in a spherical cluster renders with the blob 
primitive in 7 seconds, with the blob2 primitive in 16 seconds. It's a 
little early to make this comparison though...I haven't implemented all 
the possible optimizations for the blob2 shape.

-- 
Christopher James Huff <cja### [at] earthlinknet>
http://home.earthlink.net/~cjameshuff/
POV-Ray TAG: chr### [at] tagpovrayorg
http://tag.povray.org/

Post a reply to this message