Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!decvax!ucbvax!UTAH-CS.ARPA!peterson%ring
From: peterson%ring@UTAH-CS.ARPA.UUCP
Newsgroups: mod.computers.apollo
Subject: Re: Need help doing pixels reads/writes on a DN660s, 560s, 550s...
Message-ID: <8701200639.AA01012@tukuhnikivatz.ARPA>
Date: Tue, 20-Jan-87 01:39:33 EST
Article-I.D.: tukuhnik.8701200639.AA01012
Posted: Tue Jan 20 01:39:33 1987
Date-Received: Tue, 20-Jan-87 22:36:22 EST
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The ARPA Internet
Lines: 25
Approved: apollo@yale-comix.arpa

If the DN5/660 you're using has all of the available color memory (i.e.,
both cards) you should be able to use gpr_$read/write_pixels for eight bit
images in "direct" mode.  The key is to avoid doing any operations a pixel
at a time - doing so essentially wastes the display hardware for doing this.

For your application it sounds like you might be better off storing and
processing the image in off screen memory, and updating the screen by
calling gpr_$write_pixels once for the entire image.  This happens very
quickly.

As for displaying 24 bit images, it's probably not worth the hassle to
use this "feature" of the Apollo color hardware if you want anything
that's even remotly interactive.  Apollo's newer workstations (570, 580,
3000, etc) don't even support 24 bit color in any mode (boo hiss...).

For working with full color images, you can try dithering or quantizing
the image from 24 bits to eight.  We've written some dithering software
for the Utah Raster Toolkit that works quite well.  Somebody at apollo
wrote a color quantizer based on Paul Heckbert's SIGGRAPH 82 paper - they
use it for displaying their flashy ray-tracing pictures.  It produces very
nice results but doesn't run very fast...

Cheers,
John Peterson (peterson@cs.utah.edu)
-------