Re: scale texture support to cluster - automatically?

Date : Fri, 2 Sep 2005 09:13:46 -0400
To : XSI(at)Softimage.COM
From : Bernard Lebel <3dbernard(at)gmail.com>
Subject : Re: scale texture support to cluster - automatically?
I'm simply firmly against the use of cluster material for anything
that has to go through mental ray. Not because of mental ray, but
because of passes, as Christian pointed out. Since clusters are not
exposed to partitions (and thus, partition parameters and properties
like overrides and applied materials), controlling clusters in a
context of passes is simply looking for troubles. Describing the
problems related to this issue would be long-winded.

Guillaume has come up with a clever trick involving texture layers to
workaround this problem, but quite frankly it's so dirty that I would
never rely on such technique in production.

Now to avoid using clusters, I know only one efficient solution:
weight maps. Here I mean the general term, not the XSI specific
feature. A weight map can be a XSI weight map, a texture map, color at
vertices, or simply a grayscale texture. You pipe this in the render
tree as the branch of the weight parameters of Mix_color nodes, in
order to blend various materials. The idea is to have only one
material on the object, that has all the different surface qualities
of the object.

My favorite approach is to use bitmap weight maps, with a very small
resolution of 64x64. Usually I will do a separate texture projection
for this weight map: where I want a branch of the render tree visible,
I'll throw the UVs in the center of the projection, where the weight
map is white, and for the part that I don't want visible (ie to show
another branch of the tree), I'll collapse the other UVs in the
corner, where I have a few black pixels.

More textures, more projections, overall more work, but more
straightforward and much, much less troubles in the end. Dealing with
objects is a lot more convenient than clusters.


Cheers
Bernard


On 9/2/05, Chris Marshall <chris(at)eclipsecreative.co.uk> wrote:
> Thanks Christian,
> I'll give that a go.
> It's an interesting point about cluster materials and passes. I've never
> really got into using passes, just very basic stuff, but I would think
> that about 70% of my models for broadcast work contain clusters with
> materials. I would be interested to know how other facilities, who use
> passes alot, would approach this issue.
> 
> Chris
> 
> 
> Christian Rittener wrote:
> 
> > You can CREATE a texture support fitting exactly the size of your
> > cluster by using Get->Property->Texture Projection instead of going
> > through the image PPG. I don't know how to scale an existing TP.
> >
> > However, as Bernard would say, you shouldn't use cluster materials,
> > not if you want to use passes and overrides. By the way, is there
> > anything new in that area in v5?
> >
> > Christian Rittener
> >
> >
> > Chris Marshall wrote:
> >
> >> Brain's just died!
> >> If I have a cluster with a material and texture support applied, is
> >> there a way of auto scaling the support to the boundaries of the
> >> cluster?
> >> I thought this was possible, but can't figure out how.
> >>
> >
> >
> 
> 
> ---
> Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
> unsubscribe xsi
>

---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
unsubscribe xsi


Search the XSI List archives here or use the advanced search form to search across mailing lists. Searching help is available.
This site supposedly brought to you by Benjamin Grosser and the Imaging Technology Group.