> Date: Wed, 6 Dec 2000 14:58:25 -0800
> From: Joe Kiniry <[EMAIL PROTECTED]>
> Organization: DALi, Inc.
>
> a. Compressed geometry does not support texture information (This is the
> killer, from our perspective - what non-simplisitic or industrial rendering
> is interesting without textures?)
The current version of the compressed geometry format (Level I) does indeed
have this limitation, and yes, when it was designed (c. 1995) the expected
applications were of the industrial CAD type (back then these were the only
apps with datasets large enough to be interested in compression, and rendering
textures in real time was nearly impossible on most hardware).
Level II of the compressed geometry format supports up to 4 2D or 2 4D texture
coordinates per vertex and will be available with Java 3D 1.3 and Solaris
OpenGL 1.2.3. We also expect to have Level II hardware decompression support
available on Sun graphics accelerators at that time and a proposed extension to
OpenGL which will allow other hardware vendors to implement it.
> b. The compressed geometry APIs all have parameters of GeometryArray type
> but the specification and our experimentation seems to support the idea
> that not all GeometryArray objects are legitimate. In particular, it seems
> that QuadArrays are not valid. (Another big problem given that most of our
> entities are a mix of quads and triangles. While we can convert from quads
> to triangles, we'd rather not for all the obvious reasons.)
I think this may be a regression in Java 3D 1.2.1; unfortunately due to my
commitment at this point to OpenGL development I have not been able to
investigate this further. I haven't seen this problem in J3D 1.2. It is
definitely a priority for fixing by J3D 1.2.1 beta2 and by FCS at the very
latest.
> If the compressed geometry subsystem supported texture information and
> arbitrary geometry arrays then we would just store CompressedGeometry
> objects as part of our canning process and we would expect to see load
> times of just ~1-2 seconds per entity.
I think this is a reasonable expectation for J3D 1.3.
-- Mark Hood
===========================================================================
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".
