Re: FLIR effects using Performer

New Message Reply Date view Thread view Subject view Author view

Alex Kekesi (akekesi++at++nvl.army.mil)
Thu, 27 Jul 95 12:31 EDT


>
> I read in the Performer archived message list several messages from
> people who were looking for ways to simulate a forward looking
> infrared detection set. The basic approach seems to be to modify the
> color table to get shades of grey or green. As anyone gone farther than
> that in simulating the FLIR effect? Like is there a way in performer to
> simulate the effect of persistance?
>
> Nicolas Gauvin
> nic++at++cae.ca
>

Nicholas,

There is still is some question here at the lab as to your term "persistance"
when used with FLIRs. Nonetheless, let me step up to the plate. As you already
know, the basic method of manipulating the color table on a per polygon basis
really does not give the proper effect for FLIR simulations. This is especially
true on databases with low polygon counts. Something more is needed. We have
found that a proper balance of software and hardware seems to come very close
to acheiving a realistic sensor sim.

When first attacking this problem (ie, real-time blur, bilinear scan, noise,
etc.) we attempted to do it all in Performer. The real-time noise was no
problem. However, blurring using convolve or the accumulation buffer (as
mentioned by Rob Jenkins in his previous message) were too slow. This is where
a hardware solution was needed. For many of the sensor effects we decided to
use Datacube's real-time image processor. This seems to do the trick for real-
time blur, as well as bilinear scan. The Datacube allows for an 8x8 kernel to
be applied over the image to the get the proper "blur" effect as seen in 1st
and 2nd gen sensors. The bilinear scan is done similarly.

Now, if you are talking about the trailing effect (or "plumbing") that occurs
when an object moves across a sensors FOV this is a more complicated problem.
One possible solution is to use some sort of trailing model. We're currently
looking into this with Coryphaeus Software. They have some really nice visible
trailing models that we are thinking about adapting for IR simuations. However,
I'm sure someone out there may have some good ideas as to how IR tracks can be
left in a real-time environment directly through Performer. (A little bit of
brain food for the Performer crowd. :^) )

As our simulation grows I'm sure more and more effects will be added. If
you're interested, I could put you in touch with some Datacube representatives
to talk about their real-time image processors. Further, if you wish to do
more reading on the subject, we have published a couple papers pertaining to
real-time FLIR simulations for Distributed Interactive Simuations (DIS). The
papers are:

"PAINT THE NIGHT: High fidelity IR scene, atmosphere, and sensor simulation
        enhancements for the DIS environment" - presented at the 12th DIS
        conference and in the publication.

"DIS IR simulation models for fidelity, signature texture, and atmosphere-
        sensor effects" - presented at SPIE '95.

You may contact me for copies of these papers.

Lastly, if you wish to see our simulation we will be showing a slightly
modified version of it at SIGGRAPH '95 in the Interactive Communities area.
(look for OSIRIS - Optimized Synthetic Infra-Red Interactive Simulation).

Hope this at least points you in the right direction!

Alex Kekesi

============================================================================
Alex Kekesi phone: (703)704-1770
E-OIR Measurements, Inc.
Night Vision & Electronic Sensor Directorate
Fort Belvoir, Virginia akekesi++at++nvl.army.mil
============================================================================


New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:51:42 PDT

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.