Hi Gene,
Computationally, the real benefit of render channels is that it allows
you to save off data that has already been calculated into multiple
images. Whereas to create a pass with the same thing means
recalculating that information.
There are two situations that I use render channels for - information
that's already calculated and images which can't be created in a
standard pass.
The first one is pretty easy - depth is already calculated, diffuse,
albedo, specular, matte etc. Though you may have particular parts
you're using in a shader like incidence that you may want to extract
as well - anything from your tree really. Things like AO are a
different beat - unless your beauty pass shader is already using
ambient occlusion, then it should go in a separate pass, as you're not
gaining much computationally (someone is going to point out
tessellation or something if I said not gaining anything). In the case
of AO you're also looking at costing yourself more than you're gaining
if you need to rerender just the AO for instance.
For things you can't get otherwise a reflection pass is a good example
(ok you could get it, but with the current XSI setup you couldn't do
so in a non-destructive manner). By reflection pass I mean a real
reflection pass which includes self-reflection and other objects and
environment all in one pass.
Anyway - hope this helps explain where you can make the best use of
render channels and passes.
Cheers,
Alan.
On 7/28/07, Gene Crucean <emailgeneonthelist(at)gmail.com> wrote:
> Who here uses Render Channels? I do... but I mainly use them for small
> things like a matte for an object, a singled out rendered object or quick AO
> pass for a single object. I see these as a HUGE time saver and really want
> to see it progress into a mature feature. Everyone rendering passes should
> be interested 110%. For those of you who use them, feel free to post your
> ideas on how to improve integration and workflow. For those of you who DON'T
> use them, why?
>
>
> Here are a few of my comments:
>
> Better management control. Instead of mushing the current Render Channels
> Output ppg group into the Default_Pass Output tab, why not create a whole
> management ppg for them like the Render Manager. Make it easy to control
> every aspect of each channel. What objects are assigned to each channel.
> What shader and overrides are contributing to each channel.
> The default channels NEVER work. Make it so you can tweak them at a minimum
> to actually get them to work.
> Is there a reason to not have a Render Channel setup exactly like passes? To
> keep things consistent with the rest of XSI maybe integrate them with a
> similar setup. IMO, this takes precedence over the Render Channels PPG (1st
> bullet point).
> I like how you can break out parts or all of a shader and make that into a
> channel... and would hate to see that get the boot.
> Something I forgot about was how the channels are put together in this
> video.
> http://www.softimage.com/products/xsi/videos.aspx?video_id=27
> I watched the video when it was released and quickly forgot about the
> technique because it's not implemented like it should be. Obviously it's a
> wonderful thing that you can use code to squeeze more out of the channels
> but just to get a global AO pass? Hmm. We shouldn't HAVE to though. IMO, SI
> has the best render channels setup available so far. There needs to be an
> interface and intuitive workflow that lets everybody actually use these
> powerful channels. Personally I see passes fading off into oblivion. Maybe a
> legacy feature thats nice to have if you just want to quickly render a matte
> pass or whatever. That way you don't have to render the whole wad.
>
> Those of you who don't use them. Would you if they were implemented like
> passes?
>
>
> [/gene]
>
---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
unsubscribe xsi
