Artur,
The stripifier algorithm we use tries to improve strip lengths by adding
in "zero-area" triangles, where three consecutive vertices in the strip
create a triangle that has two vertices that are the same. For the most
part, this greatly improves strip lengths over our previous algorithm.
However, the stripifier results greatly depend on the data it is given,
so the 3DS loader could be giving the stripifier difficult data.
andrea
John Wright wrote:
>
> Artur,
>
> Our 3DS loader does indeed double the number of triangles for two sided
> materials. It's a simple solution to getting the lighting to work
> correctly. If the material wasn't lit then cull_none would be a lot
> easier (we tried that first).
>
> As to why the stripifier is making the situation even worse, I can only
> guess that because of the way the triangles are duplicated for two sided
> materials this may be presenting a worst case situation to the
> stripifier.
>
> - John Wright
> Starfire Research
>
> Artur Biesiadowski wrote:
> >
> > There is a strange thing going on with number of triangles. I have a 3ds
> > model. When viewed inside openfx it has 6236 triangles. When loaded
> > inside j3dfly it has 12472 triangles. 3ds loader seems to double number
> > of triangles - probably because of two sided material. So question
> > number one - What is a two sided material and why it has to be
> > duplicated, instead of just set to CULL_NONE ?
> >
> > But what really bugs me is what happens if I try to stripify the
> > geometry. After stripification number of triangles increasted to 20813.
> > Why ?
> >
> > And back to talk about GeometryInfo stripifier. After stripification
> > average strip length is 4.118052. I hardly see it as a good result. If
> > it is true that stripifier increased number of triangles to ease it's
> > work, we would have about 2 triangles per strip in original model... I
> > hope that just stat display is wrong - in other case we will have almost
> > 2 times triangle count increase with amount of strips being equal to
> > half of number of original triangles (after doubling by loader). Two
> > triangle strips hardly look like a performance gain for me.
> >
> > Artur
> >
> > ===========================================================================
> > To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
> > of the message "signoff JAVA3D-INTEREST". For general help, send email to
> > [EMAIL PROTECTED] and include in the body of the message "help".
>
> ===========================================================================
> To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
> of the message "signoff JAVA3D-INTEREST". For general help, send email to
> [EMAIL PROTECTED] and include in the body of the message "help".
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff JAVA3D-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
