Angus Dorbie (dorbie++at++bitch.reading.sgi.com)
Wed, 10 Apr 1996 01:44:24 +0100
I forgot to answer this bit, the summary is that you can split the download &
use multiple textures and there are big performance gains from this on IR, it
also ties in very nicely with the hemisphere requirement for at least 3 OTW
channels. For more speculative rambling detail read on.
Images must be a exponent of 2 in an edge so the closest texture size would by
1024x1024 on a single download, although you could load 1024x1280 into a
subregion of a 1kx2K image on the right platform.
One thing it makes sense to do is download the image sections and draw portions
of the mesh which you have the image data for, while always having an
outstanding download requirement unrelated to any mesh image you may currently
be using.
This is easy with the dome example because the OTW scene is rendered in three
parts so you'd draw the first OTW & download it, then draw the second &
download then the third & download then draw the dome mesh using the portions
of the image data 1 then 2 then 3, hopefully most of the third image will be
downloaded by the time you use it in the mesh. So how does this affect your
texture size? Well you could easily download the images into multiple textures
which may cause edge problems which would mean you'd have to split the mesh or
use the equivalent of a TX_SELECT texture clamp. Or you could load the image in
sections into a larger image. If youre worried about blowing all that texture
memory you could draw OTW 1 & download then draw OTW 2 & download then dram
mesh 1/3 1 then download OTW 2 over the used texture, etc. This is only
possible because you have multiple OTW channels & you may loose some
performance on the block after the mesh download.
Rgds,
Angus.
-- Angus Dorbie, The Reality Centre, Silicon Graphics Ltd, UK dorbie++at++reading.sgi.com
This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:52:41 PDT