Angus Dorbie (dorbie++at++bitch.reading.sgi.com)
Fri, 23 Aug 1996 17:50:47 +0100
I thought the question was can the iR perform bump mapping in real time and
the answer is emphatically *YES*. If you want to ask a different question
then fine maybe you get a different answer. Nobody was being suckered and
the explanation of the technique was clear.
There is another view here, the RE2 & iR are systems which can perform
all of these tricks because of the flexibility of the GL. Youre not limited
to what the demos show.
Are you saying that someone from SGI showed a customer the spotlight demo
and told them they could do it for free in his flight sim so they bought a
whole bunch?
This beggars belief, purchasing decisions aren't made on this basis.
Requirements are published, competative bids are made and an evaluation
determines where the contracts are placed. I can accept that mistakes could
have been made in the process, but blaming that spotlight demo completely
lacks credibility.
It sounds like whoever bid this should have been aware of the capabilities
of the RE2 when they made their bid. Those are the skills intergrators are
supposed to have, that's on of the reasons end users use them. I remember
being involved in similar processes while working for intergrators and
extreme care was always taken when evaluation the additional load imposed
by effects like lobe lighting.
IMHO I think SGI demos for iR aren't suckering anyone because nobody has yet
seen a demo that exploits the machine to the full (but I haven't seen the
latest CA&D stuff).
Rgds,
Angus.
On Aug 23, 9:37am, Steve Baker wrote:
> Subject: Re: bump mapping using Infinite Reality
>
> Angus said:
>
> > Brian Cabral rhowed me a real-time bump mapping implementation
> > a year ago running on RE2, this was also explained on a recent
> > SGI graphics course. He's now published a paper on a more
> > general shading approach which I haven't seen this).
>
> OK, I think I follow this explanation - but it's a typical
> 'rigged demo' approach. You end up having people believe that
> the machine can do something that is never going to work in
> most 'real' applications.
>
> The approach suggested appears to need three rendering passes
> (two for the texture and one for the final object?)
>
> My favorite use for bump mapping is super-realistic oceans
> scenes - with sunlight glittering off the water - and fan-shaped
> sun reflections as the sun sets over the horizon. VERY COOL!
> Proper bump mapping can do this with a single polygon and a
> teeny-tiny bump map with some interfering sinusoids.
>
> The problem in using the approach that Angus describes (if I
> understand it - which I may not) is that my bump map is maybe
> a 128x128 map, but it is repeated hundreds of times in each
> direction in order to texture the water from horizon to horizon.
>
> Hence when I peel the skin off my ocean - I need a 12800x12800
> texture map - rendered twice and then pasted back onto my ocean.
> Hardly a realtime solution.
>
> Some of the really cool SGI demos really piss me off because they
> lead people to believe that you have a usable technique when in
> fact, you don't.
>
> A classic (and very costly) example of this was a project that
> my company has been working on for several years. It was a fairly
> conventional RE2 flight simulator. Before we sold the system to the
> customer, we did many demos of the system flying around some
> nice looking terrain, during the day - and at night. But we didn't
> show a night time landing. During night time landings, the pilot
> turns on his landing lights - which just like car headlights,
> illuminate a little puddle of light ahead of the aircraft.
>
> So, can the RE2 do proper landing lights? Well, there is
> a really cool demo on the RE2 of a room being illuminated with
> a really beautiful spotlight. YOu can even change the spotlight to
> look as though its shining through a photographic texture.
>
> Well, that certainly looked (to our customer) like the machine
> had the ability to do that - so he purchased a large number
> of these machines.
>
> Of course, if you look at the spotlight demo, it's doing lots of
> special-case stuff, multi-pass rendering, etc, etc. SOme of the
> techniques it uses are incompatable with other features of the
> machine - some of the coplanarity kludges don't work with it, etc.
>
> There was no way we could deliver a reasonable frame rate on a
> complex database with a general purpose scene using that technique
> without nearly doubling the amount of hardware required.
>
> So - can the iR do bump mapping? The answer is emphatically *NO*
> it can't - not at the same time as coming even close to
> the pixel and polygon rates that the spec sheet say.
>
> So unless you are building a rigged demo to try to sucker
> some unfortunate customer into buying something that won't do
> what he wants - don't bother with bump mapping.
>
>
> Steve Baker 817-323-1361 (Vox-Lab)
> Hughes Training Inc. 817-695-8776 (Vox-Office/vMail)
> 2200 Arlington Downs Road 817-695-4028 (Fax)
> Arlington, Texas. TX 76005-6171 steve++at++mred.bgm.link.com (eMail)
>
> =======================================================================
> List Archives, FAQ, FTP: http://www.sgi.com/Technology/Performer/
> Submissions: info-performer++at++sgi.com
> Admin. requests: info-performer-request++at++sgi.com
>-- End of excerpt from Steve Baker
=======================================================================
List Archives, FAQ, FTP: http://www.sgi.com/Technology/Performer/
Submissions: info-performer++at++sgi.com
Admin. requests: info-performer-request++at++sgi.com
This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:53:24 PDT