Re: Replacing the shadertree? Possible?

Date : Thu, 29 Jun 2006 11:26:04 -0700
To : XSI(at)Softimage.COM
From : Andy Jones <andy(at)thefront.com>
Subject : Re: Replacing the shadertree? Possible?
Yeah, that's the idea. What's funny is Mental Ray is already set up to do this. All the surface shaders in XSI are examples of shaders built around several building blocks. That's why I think a lot of this stuff isn't a hallucination at all -- Mental Ray totally supports pretty much every feature we've talked about in some way or another.

As far as the "black box" debate goes, right now the biggest reason to do it is that you can't easily share intermediate calculations to be reused by other shaders. Even if there were a great phenomenon-generation tool, separate nodes doing the same calculations is an efficiency problem, especially if there's ray tracing involved. For example, try making a shader with two specular algorithms combined behind a mixer, like you would with a clear-coated surface. Turn shadows and some verbose messages on and notice that Mental Ray sends outs twice as many shadow rays. Sometimes this is addressable with multiple outputs, and other times it's not. In this case, it's particularly tricky because what you need to do is compute an array of light samples and send it to both shaders so they can attenuate the highlights each their own way. In a shader like T2S, this is taken care of by having multiple specular models you can turn on and off within the shader (though I actually preferred a more general "primary, secondary, tertiary" solution where you pick which model to use each time.) Rendering in passes alleviates some of these ploblems, but not all. For this reason, we structure our shading behind a fairly monolithic shader node. But we try to allow things to be overridden by custom solutions wherever it might be needed. And once the customization starts, smaller nodes become very useful.

-Andy

kim aldis wrote:



-----Original Message-----
From: owner-xsi(at)Softimage.COM [mailto:owner-xsi(at)Softimage.COM] On
Behalf Of Tim Leydecker
Sent: 29 June 2006 09:48
To: XSI(at)Softimage.COM
Subject: Re: Replacing the shadertree? Possible?


But seriously, Andy Jones had pointed out the following here:



Ultimately, what has benefited us far more is to find ways to reduce
the building blocks of the shaders to a smaller set which we can then
change dynamically and have the changes propagate through the job.



[kim aldis]


But someone has to build that reduced, smaller set in the first place. I think what Andy's asking for is the best of both worlds; more of the smaller building blocks but at the same time something that will allow him to pull those small building blocks together himself into something simpler. Correct me if I'm wrong, Andy.



---
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.