Le Forgeron wrote:

> Warp wrote:
> 
>> Dejan D. M. Milosavljevic <pov### [at] ddmrmcom> wrote:
>>
>>> Sphere and plane acceleration.
>>> Nothig special. Just warming for real one.
>>> URL: http://www.ddmrm.com/graph/pov/pov_index.html
>>>
>>
>>   It would be nice if there was a description of what you have done and
>> some measurements (good measurements, not something like 3s to 2s 
>> "speedup").
>>
>>
> 
> Looking at the zip file for 3.5 sphere (and the #ifdef in particular), 
> he removed the length's adjustement of the depth for ellipsoid (removing 
> two divisions), as well as the transformation also for ellipsoid from 
> the object space to the world space (removing 2 Matrice.Vector products)
> 
> I have a feeling that either something else is missing or that the 
> #ifdef part is bogus and asking for trouble with CSG.
> 

Also, looking at the zip for plane, he removes all transformations 
matrix by adjusting instead the plane normal and origin instead (which 
make sense for an infinite plane with only rotate/scale/translate).
I'm nevertheless wondering if this could have a side effect on 
non-uniform texture. May be the plane patch is ok.
(Unless there is a way for a plane to first get a transformation, in 
which case the removed code was not dead enough).

Post a reply to this message