Is real time raytracing possible?

Is real time raytracing possible?

Post by wfs.. » Thu, 16 Nov 1995 04:00:00



Have there ever been any attempts at real-time raytracing on any platform?
Would u mind telling me about them?
Thanks.
 
 
 

Is real time raytracing possible?

Post by N.. » Fri, 17 Nov 1995 04:00:00



>Have there ever been any attempts at real-time raytracing on any platform?
>Would u mind telling me about them?
>Thanks.

It is currently possible to Ray trace at an incredible rate,
improvemnets have increased expidentially over the past few years, but
if you look closely at what is in essence exploring every path of
light within an image essepically when you add complexity to the
equation it is possible to see that for a while it will not occur with
any fluidity.The answer in essence is NO! I recent years research has
taken place into trying to rework the theories, but these have
resulted in cheating methos and have deviated some what from the true
ray tracing. I hope this has helped

Bye

 
 
 

Is real time raytracing possible?

Post by Brian Tucke » Fri, 17 Nov 1995 04:00:00


I think I saw something about it when I was visiting Cambridge Uni.
I believe they were possibly looking at putting the intersection algorithms on
chips or something.

Please don't quote me on this, I just picked up a piece of paper in the
computing deparment, read the title and stuck it in my bag with all the other
bits of paper I happened to be collecting that day.

Brian Tucker

 
 
 

Is real time raytracing possible?

Post by Steve Hollas » Fri, 17 Nov 1995 04:00:00



< The answer in essence is NO! I recent years research has
< taken place into trying to rework the theories, but these have
< resulted in cheating methos and have deviated some what from the true
< ray tracing.

    A very odd answer.  True raytracing is a cheat.  Scanline methods are a
    cheat.  Radiosity is a cheat.  The standard illumination equation is a
    cheat, as are all other variants I've seen (Cooke-Torrance et. al.)

    Rigorous Computer Graphics Proof:  "Looks good to me."


 
 
 

Is real time raytracing possible?

Post by Roger Heathco » Sat, 18 Nov 1995 04:00:00





> >Have there ever been any attempts at real-time raytracing on any platform?
> >Would u mind telling me about them?
> >Thanks.

> It is currently possible to Ray trace at an incredible rate,
> improvemnets have increased expidentially over the past few years, but
> if you look closely at what is in essence exploring every path of
> light within an image essepically when you add complexity to the
> equation it is possible to see that for a while it will not occur with
> any fluidity.The answer in essence is NO! I recent years research has
> taken place into trying to rework the theories, but these have
> resulted in cheating methos and have deviated some what from the true
> ray tracing. I hope this has helped

> Bye

on a 10mips machine you could do a damn good trace in 24 hrs.
to do this in and hour you`d need a 240 mips machine,
for 1 minute you`d need a 240*60 mips machine, thats 14400 (ithink)
1 sec about 864000 mips
50 fps - 43200000 mips

43,200,000 mips =

43,200 billions instructions per second =

43, trillion instructions per second,

Fastest machine in the world ...... 1.8 trillion instructions per second.

However, You could probably do a usable trace on a 10 mips machine in a lot
less that 24 hrs ! and cut the frame rate.
That brings it within the realms of possibility. But I doubt the US military
will let anyone use their super computers to run a ray traced version of doom,
ETC....

Rog.

 
 
 

Is real time raytracing possible?

Post by John Panettie » Sat, 18 Nov 1995 04:00:00




>>Have there ever been any attempts at real-time raytracing on any platform?
>>Would u mind telling me about them?
>>Thanks.
>It is currently possible to Ray trace at an incredible rate,
>improvemnets have increased expidentially over the past few years, but
>if you look closely at what is in essence exploring every path of
>light within an image essepically when you add complexity to the
>equation it is possible to see that for a while it will not occur with
>any fluidity.The answer in essence is NO!

I for one don't understand why it can't be done.  Let's consider a depth
one ray trace for now, since as you say the number of operations
increases rapidly when reflections are accounted for.  Since bounding
volumes can be generated fairly rapidly on the fly, the order of
operations for tracing a single ray in a dynamic scene can be made close
to the log of the number of objects in a scene.  Generating several
thousand such rays slows things down tremendously, but it has occured to
me for quite some time that all a hardware manufacturer would have to
do is design a circuit that can (1) spit out u,v coordinates of a ray's
intersection with a plane and (2) decide whether or not a ray hits any
of a certain variety of bounding volumes (prism, sphere, etc.)  If the
design is kept relatively simple, several of these circuits can be put
on a single chip (redundant circuitry would help bring the price down)
and each circuit could raytrace a portion of the screen in parallel.
I don't know how many circuits would be needed before this method pays
off, but the equations are not that complex and I don't see why it should
be a problem.

                                        John Panettiere

 
 
 

Is real time raytracing possible?

Post by Avi Pilos » Tue, 21 Nov 1995 04:00:00


: Have there ever been any attempts at real-time raytracing on any platform?
: Would u mind telling me about them?
: Thanks.

Only direct in harware, AFAIK. Stuff like a procesor for every pixel etc...

###################################################################


                 -       http://www.cse.unsw.edu.au/~avip/

Quote:> Button: "If we are what we eat, I could be you by morning."

 
 
 

Is real time raytracing possible?

Post by Peter Gru » Tue, 21 Nov 1995 04:00:00


 Ho> A very odd answer.  True raytracing is a cheat.  Scanline methods

If they were to film "The Graduate" today, the uncle would advise the
young man "Semantics my boy. Semantics."

Full fledged raytracing is not a cheat on full fledged raytracing though.

- gruhn

... Boku wa hon desu.
___ Blue Wave/QWK v2.12

-----------------------------------------------------------------------------

This message was processed by NetXpress from Merlin Systems Inc.
-----------------------------------------------------------------------------

 
 
 

Is real time raytracing possible?

Post by Alex Mo » Wed, 22 Nov 1995 04:00:00


Actually, John Carmack (3d engine programmer for id) created a realtime
raytracer for a very simple 2d bsp engine (like doom, but faster-
rendering levels) with only the walls (or maybe some other simple
geometry) being in the world. He could; however, only get it going at 10
fps.

Alex Mohr

 
 
 

Is real time raytracing possible?

Post by Steve Hollas » Wed, 22 Nov 1995 04:00:00



> A very odd answer.  True raytracing is a cheat.  Scanline methods ...


< If they were to film "The Graduate" today, the uncle would advise the
< young man "Semantics my boy. Semantics."
<
< Full fledged raytracing is not a cheat on full fledged raytracing though.

    Ignoring the circular logic =^), a big problem I have with this thread
is that nobody has bothered to define real-time or "full fledged
raytracing".  Using the assertions of some in this thread, I could well
argue that real-time scanline rendering is impossible on today's
architecture:

        4k x 4k resolution
        75Hz
        1 million triangles:
            48-bit color
            16-bit multipass transparency
            phong shading
            texmapped
        environment-mapped
        shadows
        50 light sources

    What exactly is "full fledged raytracing"?  Temp*anti-aliasing?
Caustics via backtracing?  Area light sources via jittered sampling?

    Could I not assert that the following would be an example of real-time
raytracing:

        64 x 64 resolution
        9 Hz
        4 spawn max
        4 spheres
        1 directional light source
        1 ambient light source

    What do folks mean by full-fledged?  (what do folks even mean by
full-fledged scanline rendering?)  What do people mean by realtime?


 
 
 

Is real time raytracing possible?

Post by Jonathan Coh » Wed, 22 Nov 1995 04:00:00



>Have there ever been any attempts at real-time raytracing on any platform?
>Would u mind telling me about them?
>Thanks.

The Army Research Laboratory has a 96 processor SGI Challenge array
which it uses for real-time raytracing of CAD models.  I don't know much
more about it, though.

Jon

 
 
 

1. Real Time Plotting Possible?

Hi,

I would like to use GNUPLOT to read in some data that is being collected
to display data in real time. This would require GNUPLOT to continually
read and re-read the data file to add the new information.

I will explain my current setup. I have a job running that polls some
devices and writes the data that it collects to an ascii file. Then I use
xnmgraph (a tool that is part of HP OpenView Network Node Manager) to read
the data and display it. It does a very nice job of displaying the data
in plotted fashion. However, I need to migrate the job off of my OpenView
manager and therefore xnmgraph will not be available to plot the data.
Does GNUPLOT do this sort of thing? Or is anyone aware of some other tool
that I can use?

thanks,
Mike Runge

2. MPEG-playerr for UNIX-workstations via FTP !

3. QRay - an interactive 3D modelling system in Java with built-in real-time raytracing engine

4. 2 33066

5. Is there a real time algorithm for raytracing ?

6. Unpredictable Canvas->Polygon() results.

7. Real Time Raytracing demo!!!

8. freetubes.com update, Easter, baby stuff, hardware, seamless tiles and more...

9. !!!!!!!Real Time Raytracing with a CRAY and direct Hook up

10. ?: real(time) fast raytrace picture/anim

11. Real Time Convolution - timing question

12. Real Time Convolution on SGs - timing question

13. Am I getting real bold and italics?