Re: Replacing the shadertree? Possible?

Date : Tue, 27 Jun 2006 01:17:48 -0700
To : XSI(at)Softimage.COM
From : "Mike Werckle" <stumbly(at)gmail.com>
Subject : Re: Replacing the shadertree? Possible?
Thanks Bernard for your thougtful response.  And thanks to everyone for a great discussion.

Bernard, here are some more thoughts:

I respect the difference of opinon we have on whether the render tree is intuitive or not.  I feel perhaps I wasn't as clear as I could have been-- I don't think Soft3D's material (shader) functionality was better than XSI's, nor do I know enough about Max or Maya to compare.  I'm just using the render tree as an example of how I feel XSI in general feels unintuitive.   I feel the same way about the particle system, dynamics, hair, and other XSI systems in their current states.

To me, intuitive means that I can pick it up and figure out how to use by nature of its design-- that I can learn it by playing with it, by feeling it out. It means the tool gets out of the way and allows me to concentrate all of my energy on creating.  To me, this is an "artists" approach, although I undertand not all artists work this way.

Perhaps I am too idealistic.  Do you think that the act of creating shaders is so complex that it's simply unrealistic for me to wish for a more elegant solution?

Mike






On 6/25/06, Bernard Lebel <3dbernard(at)gmail.com> wrote:
I'm sorry but I have to say I completely disagree with everything you
say. See [Bernard] below.


> When I said I felt the Render Tree was a mess, I meant that I feel it's a
> mess in terms of being an intuitive tool for artists.  Currently it seems to
> be a visual tool for the technical minded person-- not that that's a bad
> thing.  It's just that I feel like if Soft is going to make a visual
> interface, it should be for the artist, not necessarily the TD.
>
> The technical minded person doesn't really need the visual interface because
> they are going to understand the inner workings of the shaders and render
> pipeline to begin with.

[Bernard] Quite untrue. I'm a TD, and yet I'm a *very* visual type of
person. Those who believe that all TDs are good at maths or are good
at thinking about complex problems are wrong. I'm not that type of TD,
and thus the Render Tree makes my job quite easier.

I think there are a lot of prejudices regarding the TDs. As artists,
TDs come from all walks of life, and some TDs are very creative and
original with existing tools but poor technicians, some are very good
technically but they need to code everything from scratch to get
anything done, others are amazing at finding and working around bugs
that one else can, and so on.



> The visual representation of this should therefore
> be useful to those who are going to benefit most from it-- the artists.

[Bernard] I'm sorry, but it IS useful for the artist. Look, I have
used XSI, MtoR, Softimage|3D and 3ds max, and I think XSI's solution
outforms them all both in terms of user-friendliness and simplicity.
YES, there are some technical aspects to the Render Tree, like the
vector shaders, and some of them are actually poorly designed
(bumpmapping comes to mind). But still I think it's quite wrong to say
it's not useful to the artist, quite the opposite in fact. When I came
to XSI I was not a TD, and I really liked the approach.

In fact, since you have used Softimage|3D and loved it so much, I
wonder how you can say XSI is worse than SI3D in that area. Layering
textures on top of each other was the only thing that was easy,
anything was quite a mess and limited if you ask me. I think the
Render Tree made the whole shading approach a lot more friendly and
interesting, I have a hard time believing you love treeless interfaces
like SI3D and 3ds max.



> I don't mean to imply that artists aren't also capable of understanding the
> more technical aspects, just that the render tree is quite convoluted and
> mystical.

[Bernard] Convoluted, not so much. Create node, plug it. Dclick on it
to edit the parameters. I hardly see how this can be convoluted.

The improvements I would see to this workflow would be:
- have floating menu for shaders
- see custom shaders in menu, or customizable menus
- revamp the texture layers implementation in nodes
- perhaps a few more features (multiple outputs for example),  and few bug fixes
Other than that, I think the Render Tree is fully functional in its
current state. It is a far better solution to any other approach I
have ever used.

Mystical, probably. Many shaders expose mathematical operations, so
the math-oriented user can indeed take advantage of it. But for less
math persons, like me, I can do the majority of the work without using
too much of the math shaders. Of course the more you know, the better.
But I think it's quite a mistake to believe that the Render Tree is
only for the math person.

In the area of "mysticism", I think the desperately needed improvement
is documentation about the math nodes, as well as shaders that exposes
these math concepts in a language the non-math guy can understand.

The whole problem, I think, of this debate is that the non-math person
doesn't know what to do with the math nodes. Environment shaders,
illumination and all this stuff is simple for pretty much everyone,
not so much documentation is needed about this. But math nodes to
extract and manipulate vectors, matrices and all kinds of
not-very-visual-aspects should be much better documented and tools
could be offered. Btw the Production Series DVD does a pretty good job
at this......


> It's not something that you can just pick up-- it requires
> advanced training which otherwise incredibly talented artists might not
> have.

[Bernard] That's not true. Most of the basic work can be done without
any advanced use of the tree. "Advanced" training will probably be
needed when doing "advanced" effects with the Render Tree. I think
it's pretty logical, and applies to any area of 3D.



> It's also my understanding
> that Soft 3D was the first to make a camera look like a camera, and make
> lights loook like lights-- I'm not sure about this, can anyone corroborate
> this?

[Bernard] I'm not sure what you mean here. You don't mean the viewport
icon, don't you? If that, I don't see where it fits in this
discussion. If you mean parameter and behavior, I assure you there has
never been more real matter in Softimage's lights than in any other
softwares. Most softwares implement lights in a very similar fashion.
I fact I think the 3ds max lights are years ahead from those of SI3D
and XSI in terms of features, if not only for the fact 3ds max doesn't
have a Render Tree.

Actually 3ds max is a good example. It has a super strong lighting
toolset, however it doesn't have a Render Tree. Everything is either
done through a panel (ie a single PPG) or through the Material Editor,
which is very similar to the SI3D Materials window. Nested shader
trees is extremely convoluted to navigate, there is not multi-ppg. But
yes, I will agree that the procedural shaders in 3ds max (with the
fact you can see them in viewports) are extremely powerful. Sitting on
a Render Tree interface, the whole 3ds max shading tools would rock.



> By using the Render Tree as an example, I'm trying to say that it seems like
> the original intent of Soft 3D has been lost or abandoned with XSI-- does
> anyone else see that?  It's a tool with great functionality, but not
> necessarily designed to empower the artist.

[Bernard] I would say the exact opposite, in fact, XSI gives so much
power to the artist that the artist is lost and feels frustrated. As I
said, I think we need a much stronger documentation, and nodes that
implement math concepts in a non-math interface.



> I just feel like it could be a factor that plays into the fact that
> XSI isn't being overwhelmingly adopted-- becasue I really wish it was.

[Bernard] I think this statement is highly arguable and subjective and
doesn't belong to this discussion.



Bernard
---
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.