PDA

View Full Version : PrMan displacement mapping and hair


mjs1
01-17-2004, 09:51 PM
Hello all!

I'm new to PrMan and i have a question regarding displacing a surface that has hair geometry atteched to it.

Say you have a short-furred character like a lion that is displaced for features such as veins, tendons, etc. That means displaced geometry only exists within the current bucket x/y dimensions. With hair you will have strands that originate above, below, and to the sides of the current bucket that may be visible within the current bucket. If the geometry for the offscreen-originated strands is not displaced since it's not in the currently active bucket, how do the hair strands match up between bucket borders?

I imagine shadows might have a similar issue. If you had a long vein running across many buckets you might get shadow inconsistencies. How does PrMan handle this?

Currently I'm only aware of being able to alter bucket size. The larger the bucket size the less efficient the on-the-fly displacement becomes.

gga
01-18-2004, 10:41 AM
Originally posted by faceForward
Hello all!

I'm new to PrMan and i have a question regarding displacing a surface that has hair geometry atteched to it.

the geometry for the offscreen-originated strands is not displaced since it's not in the currently active bucket, how do the hair strands match up between bucket borders?


Simple. Prman does not create hair from displaced surfaces (or from any surface, for that matter).
It is up to the RIB translator (mtor, houdini, etc) to place each hair strand appropiately. Most likely, the translator does not bother about the displacements, and just grows hairs from the non-displaced object surface.


I imagine shadows might have a similar issue. If you had a long vein running across many buckets you might get shadow inconsistencies. How does PrMan handle this?
[/B]

Using shadow maps, mainly. Shadows are generated as a prepass, taking into accout what each light sees, not what the camera sees, so whether the object is in view is no problem. You can, however, get small tesselation and floating point discrepancies between the light view and the camera view due to the size of your shadow map. This can lead to the usual artifacts of shadow maps (bias issues, lack of detail, too blurry, etc.). In practice, with decent settings, this is rarely a problem for most scenes.
When dealing with raytracing (like traced shadows), prman keeps two tesselations, both done in a view dependent manner (one for what the camera sees and one for what each ray sees, basically). You can find more info about that in the recent paper of Julian Fong, Per Christiansen and Dana Batali (available on Christiansen's web site).

playmesumch00ns
01-19-2004, 03:17 PM
What's the website address? Google doesn't turn up anything of interest

gga
01-19-2004, 04:26 PM
No surprise, as I mispelled Christensen's last name :)

http://www.seanet.com/~myandper/abstract/eg03.htm

CGTalk Moderation
01-17-2006, 04:00 AM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.