[comp.periphs] Fancy Frame Buffers ?

dan@rna.UUCP (03/24/87)

	We have several needs for a graphics/image system that can display
moving objects and images in real-time. Any suggestions ?

	This is the way we see it:

	One architecture is the vector/polygon/object type, e.g. the Megatek,
E&S or the IRIS. They would allow rapid generation and INDEPENDENT movement of
multiple objects in real-time. The programming would be relatively easy, too.
A disadvantage (I think) is that images and other graphics requiring many
color/gray levels and pixel densities, especially those not well described as
objects, would be difficult. Question: are there any machines that do
anti-aliasing on the fly ?

	Another option is a frame buffer with multiple, perhaps per plane,
pan and scroll registers. I don't really know of such a beast although I
recall seeing an ad for one once.

	I should note that we currently have an Ikonas and have been pretty
successful with it but would like something easier to program and more capable.
Besides, its not clear how long Ikonas/Adage will be around. Any rumors ?
However, related to the second idea above, is it possible to put two (or more)
frame buffer controllers on an Ikonas and combine (with sync) the video outputs,
thus getting two independent FB's with two sets of pan and scroll registers ?

	Also, it would be nice to be capable to pre-compute multiple, perhaps
one bit deep, images (say a few dozen images) and be able to flash them up
within a frame time.

	Thanks.

					Cheers,
					Dan Ts'o
					Dept. Neurobiology
					Rockefeller Univ.
					1230 York Ave.
					NY, NY 10021
					212-570-7671
					...cmcl2!rna!dan
					rna!dan@cmcl2.arpa

crum@utah-cs.UUCP (03/24/87)

In article <615@rna.UUCP> dan@rna.UUCP (Dan Ts'o) writes:
>objects, would be difficult. Question: are there any machines that do
>anti-aliasing on the fly ?

The Evans and SutherLand "Shadowfax" project has developed a chipset that
includes anti-aliased line drawing capability.  The new E&S PS 390
"workstation" will have this capability (in addition to the interactive 3D
transformation and depth-queueing that E&S PS systems are known for).  The
PS 390 will be a raster system, whereas other E&S PS systems use calligraphic
technology.

I saw the Shadowfax chipset in action last Wednesday after an ACM talk by one
of the project engineers, Scott R. Nelson.  The testbed that I saw consisted
of 56 1024x768 bit planes and the custom chips on cards plugged into a DEC
MicroVAX.  The test software demonstrated perspective views of color
wireframe images being manipulated interactively.  The test software allowed
one side of the image to be rendered with anti-aliased lines while the other
was left without anti-aliasing.  Actually I saw two testbed systems, one
controlled by a MicroVAX running Ultrix, the other controlled by a MicroVAX
running VMS.

I was told that the Shadowfax chip that draws into raster memory can support
a resolution of 2048x2048.  The chipset uses 32-bit fixed point representation
to maintain the x, y, z, and w coordinates of line endpoints through most of
its hardware pipeline, just as the PS 350 does.  The current Shadowfax chipset
does not have the capability of anti-aliasing polygon edges, nor can it
manipulate curves in hardware without using some polyline approximation of
those curves.

I don't really like to discuss marketing considerations like cost, but I think
that it was mentioned that the meeting that a base PS 390 system should cost
somewhere around $40K.

I'm not sure if Scott reads this newsgroup, but I know that E&S is on the net.
Perhaps someone there could correct anything misleading that I have said here,
and give us some some PS 390 performance figures (which may not really mean
much).

Do any of you know if other manufacturers are working on CAD systems that will
do fast anti-aliasing?  I suppose that the technology used in broadcast video
engineering could also be of interest here (Bausch?).

Bravo, E&S...  You are leading CAD in some very interesting directions.

Gary L. Crum
University of Utah Computer Science Department
arpa:  crum@cs.utah.edu
uucp:  ...{seismo,ut-sally,devax,ihnp4 (among others)}!utah-cs!crum