Dave Akers (dla)
Thu, 21 Oct 1999 16:44:01 -0700
First, I should mention that COLOR_INDEX mode _is_ supported in
Volumizer 1.1. As you point out, the documentation is certainly lacking in
letting you know about how to use it, however.. Sorry about that!
Volumizer's choice of RGBA or COLOR_INDEX as an external data format is
actually determined by the format of the data on disk. If the data is
stored in color index form on disk, then you should specify COLOR_INDEX as
the diskDataFormat in the call to voAppearanceActions::getBestParameters().
This will automatically select COLOR_INDEX for the external format. If
your data is actually stored in RGBA format, then you're stuck using RGBA
for your external format. By the way, COLOR_INDEX mode is not handled as a
post-interpolation in hardware, but instead as a pre-interpolation step..
(So your internal texture format will be RGBA8_EXT if you're using
COLOR_INDEX external format.)
Another option is to read in luminance data from the disk, and apply an
RGBA lookup table. This will automatically take advantage of hardware post
interpolation if it's available on your graphics board.
Finally, you can also get color by supplying per-vertex colors within
the geometry specification of your model.
Does this help clarify things a bit? Let us know if you have any other
questions.
-Dave
On Thu, 21 Oct 1999, Manfred Weiler wrote:
> Hi,
>
> on an Indigo with Maximum Impact Graphics I noticed that Volumizer has a
> suboptimal strategy in visualizing a colored volume:
>
> Volumizer proposes to use a RGBA texture an RGBA pre interpolation lookuptable.
> But when using a color index texture with a RGBA lookup table one can not only
> save 75% of texture memory but also use a hardware supported post interpolation
> lookup table.
> Why doesn't Volumizer use this strategy?
>
> BTW. After having a look in the documentation I noticed that
> Volumizer doesn't even support the color index format.
> Is there another way one can achieve colored visualization without
> using that big RGBA-Textures.
>
>
> Manfred.
>
>
>
This archive was generated by hypermail 2.0b2 on Thu Oct 21 1999 - 18:10:03 PDT