high precision models

[Seth G. <sga...@li...>]

Taniwha has suggested that GL_GetAliasFrameVerts could be used by the
software renderer to get interpolated animation ("lerp"d or "tween"d so
the animation looks smooth instead of choppy) on .mdls

It could also simplify implementation of high precision model rendering,
since this function also converts the byte coordinates to floats for GL
rendering.  I like this idea, but it could prove tricky:

The software renderer DOES convert coordinates from bytes (all frames
stored in memory this way) to floats (temporarily for rendering and
clipping), but then it turns them into ints for rendering.

So far, so good, but the software renderer splits into two paths when
converting it's vertices depending on whether the model is clipped. (look
for "trivial_accept" in sw_ralias.c and sw32_ralias.c)

In the case where clipping is not needed, there is an assembler version of
the function that converts the bytes to floats to ints (in sw_ralias.c,
but not sw32_ralias.c)

I think we'll have to drop that assembler function (or at least make it
conditional) to get either interpolation or high precision models.  How
big of a difference does this assembler function make?

Comments?  Suggestions?
  __    __    _                                    _    __    __
_/  \__/  \__/  Seth Galbraith  "The Serpent Lord"  \__/  \__/  \_
 \__/  \__/  \_ sga...@kr...  #2244199 on ICQ  _/  \__/  \__/
_/  \__/  \__/    http://www.planetquake.com/gg     \__/  \__/  \_