normal mapping valid in renderman?

[Anthony Rosbottom]

Hello again,
Just wondering, if all the renderman compliant renderers do micropoly displacement for free, is it worth using normal maps at all? I use normal maps in mental ray because they are easier to apply and render.

Do normal maps have any advantages over displacement maps in a rm compliant renderer like Aqsis or Pixie?

Cheers,
Anthony Rosbottom

www.bliz.co.uk
 

[Paul Gregory]

Every RenderMan renderer in common use excels at subpixel displacement. They mostly implement a variant of the Reyes rendering architecture, and so deal directly with some sort of micropolygon.

The only time there will be a significant hit for displacement is when you have large amounts of displacement, you'll need to adjust the bound and that can sometimes result in a performance hit. But then, if your displacement is that big, normal mapping ain't gonna cut it anyway.

Cheers

Paul

--
You received this message because you are subscribed to the Blender to RenderMan Forum Group.

To post to this group, send email to blenderto...@googlegroups.com
To unsubscribe from this group, send email to blendertorender...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/blendertorenderman
To view the Blender to RenderMan Blog, visit http://blendertorenderman.blogspot.com

Any questions about membership of the group or blog can be sent to blenderto...@gmail.com

--
Paul Gregory
http://www.aqsis.org

[>WHiTeRaBBiT<]

From how I understand it all surfaces are split and diced regardless
of the geometry type therefore even if you don't displace the overhead
is still there (so why not take advantage of it). This has been the
hardest thing to explain to users comparing performance to something
like Blender's internal as RenderMan is doing more work even if it's
not visible on a box and plane.
One area were performance can be hit using displacements is when
including them in raytracing. It can become even worse if the scene
isn't carefully tuned as sampling, inter-reflection/refraction and
volumes can begin to exponentially slow the renderer down (one big
advantage for using raster based techniques in complex scenes). In
these cases it's customary to tune the ray depths, limit ray distance,
control object visibility with ray groups, make use of point clouds
for baking and even mix traditional methods too.
Also something else worth mentioning about displacements is Air by
default doesn't displace without a special attribute, instead it
converts displacements into normal maps "on-the-fly" to squeeze out
more speed (I'm assuming in this case it doesn't do sub-pixel
calculations unless enabled?).

Anyway that's my 2 cents :)
Eric Back (WHiTeRaBBiT)

> > blendertorender...@gmail.com
>
> --
> Paul Gregoryhttp://www.aqsis.org