From: Steve Fuhrman (sfuhrman++at++covad.net)
Date: 03/07/2002 11:51:41
Hi,
I apologize for mistyping the URL:
IP address is 64.38.118.126
The lesson learned is to rush to eat lunch and then ask people for help.
Not the other way around.
Sorry for the mistake.
Thanks to Svend Tang Petersen for pointing this out.
Steve Fuhrman wrote:
> Hi,
>
> I'm trying to finish up a train simulator and ran into an issue with
> headlights that has me baffled.
>
> The headlight works as expected for about 56 out of 60 miles of a
> database
> which was auto-generated from "fake" data and then tweaked with Creator.
> Nothing is unique about the small section in the middle of the database
> that exhibits the problem.
>
> What is so confusing is that from one direction (trains go in forward
> or reverse on the tracks) the headlight works fine, but at the same
> location from the other direction the light looks "burned out".
>
> Here's an excerpt from a webpage I put up so you could view the problem
> as I would rather show 4 pictures than ramble on for 4000 more words ;-)
>
> See http://maxoutinc.bwcsoft.com and click on "Headlight Issues" to
> see the related pictures. Warning: The pictures are full-screen so
> those without
> fast bandwidth connections should either be patient or avoid entirely.
>
> Background information
>
> * No flat shaded tri-strips (using PFDBLDR_MESH_LOCAL_LIGHTING)
> * pfuTraversal functions already written to look for and fix
> "unusual" geometry or geostates (i.e. missing material, bad/missing
> normals, emissive components, etc.)
> * Headlight based on libpf/C++/projective.C example
> * Database built with Creator and then converted to pfb
> * flt files exhibit exact same behavior as pfb files
> * Database broken into "tiles" and headlight issues cross two tile
> boundaries.
>
> Hw/Sw Information
>
> * Performer 2.5
> * nVidia 2313 driver
> * GeForce 3
> * RedHat Linux 7.2
> * 1.4 Ghz Athlon
> * 1 Gb RAM
>
> Clues, Experiments, Questions
>
> * Copying a "good" polygon that is lit by the headlight into a
> portion of the database that isn't lit by the headlight, causes that
> polygon to also not be lit.
> * I'm assuming that because of the above experiment and the fact
> that the problem is direction based that it has something to do with the
> hierarchy and the cull/draw traversal instead of a basic problem with
> the geosets.
> * All suggestions welcome. Please send them to sfuhrman++at++maxoutinc.com
> * Thanks in advance for any help.
>
> I'll summarize and post private replies.
>
> pfThanks,
-- ========================================================================== Steve Fuhrman ========================================================================== -- American Proverb Scratch my back and I'll scratch yours.
This archive was generated by hypermail 2b29 : Thu Mar 07 2002 - 11:53:15 PST