gray
From: softimag...@listproc.autodesk.com [mailto:softimag...@listproc.autodesk.com] On Behalf Of David Saber
Sent: Thursday, March 19, 2009 13:55
To: soft...@listproc.autodesk.com
Subject: Re: Freeze modeling and keep the texture support?
--------------------------------------------
Eric Thivierge
Technical Director
Speakeasy FX
I thought it was a bug too so I'm interested in anything anybody can remember about that earlier discussion.
BTW to clarify, FreezeM does not depend on which operators you have selected but Freeze (no "M", right above FreezeM) does. The Freeze button is the same command as on the right-click menu.
gray
-----Original Message-----
From: softimag...@listproc.autodesk.com [mailto:softimag...@listproc.autodesk.com] On Behalf Of Eric Thivierge / XSI Database
Sent: Friday, March 20, 2009 11:27
To: soft...@listproc.autodesk.com
Subject: Re: Freeze modeling and keep the texture support?
The current behaviour is not consistent with what FreezeM is supposed to
do - that is: freeze the modelling part of the stack and leave all else
untouched.
If you will, the texture op sits outside of the stack in the explorer so it
could even be left out when freezing alltogether.
If you want to freeze it, you can allways just select it and freeze it.
----- Original Message -----
From: "Grahame Fuller" <Grahame...@autodesk.com>
Freezing the TextureOp itself will always freeze whatever is below it, that is, the generator if it's still exists and anything else that happens to be there.
I agree that it makes sense for FreezeM to be exactly like freezing the topmost operator in the Modeling region.
-----Original Message-----
From: Grahame Fuller [mailto:Grahame...@autodesk.com]
Sent: Friday, March 20, 2009 2:28 PM
To: soft...@listproc.autodesk.com
Subject: RE: Freeze modeling and keep the texture support?
My understanding is that the TextureOp is always somewhere in the stack,
but you just don't see where it reads and writes from in the explorer.
Making the texture swim is equivalent to moving the TextureOp to the
very top above Secondary Shape, and making the texture stick is
equivalent to moving the op to the bottom of Modeling (just above any
generator op if there is one).
Freezing the TextureOp itself will always freeze whatever is below it,
that is, the generator if it's still exists and anything else that
happens to be there.
I agree that it makes sense for FreezeM to be exactly like freezing the
topmost operator in the Modeling region.
Does anyone know of any documentation (for the average artist) that can
give me a decent grasp on Groups, Layers, Partitions, and how they all
interact with passes? I have read the help docs, over and over, and feel
that I'm missing a piece of the big puzzle.
Best Regards,
Steve Davidson
(954)552-7956
www.3danimationmagic.com
I dont think a very generalised explanation is going to be very
enlightening.
to an extent, they are "general purpose" organisational tools, and what they
do depends on what you make them do - or in other words: different workflows
make a different use of them. One could show you his way of using them, but
that wouldnt be a complete overview of what they are and do.
This is a bit vague isnt it?
----- Original Message -----
From: "Stephen Davidson" <magi...@bellsouth.net>
To: <soft...@listproc.autodesk.com>
I would be interested in rendering different objects in different passes.
What would be the best way to pull that off?
When I create a new pass, the same object, lights, and cameras
are shared in the separate passes. How could each pass
have unique models, lights and/or cameras yet retain some common elements?
Groups? Layers? Partitions? Am I asking a clear question?
Best Regards,
Steve Davidson
(954)552-7956
www.3danimationmagic.com
-----Original Message-----
From: softimag...@listproc.autodesk.com
[mailto:softimag...@listproc.autodesk.com] On Behalf Of peter boeykens
Guillaume
-----Original Message-----
From: softimag...@listproc.autodesk.com [mailto:softimag...@listproc.autodesk.com] On Behalf Of Stephen Davidson
Sent: March-20-09 4:45 PM
To: soft...@listproc.autodesk.com
The "chain of priority" goes like: Layers > Partitions > Groups, where
the first is the highest priority and the last is the lowest. In other
words, if a layer says that some objects won't render, they WON'T,
regardless of what the groups or partitions say.
Does that make things any clearer?
On Fri, Mar 20, 2009 at 4:45 PM, Stephen Davidson
<magi...@bellsouth.net> wrote:
Group - a generic container for associating scene elements for a common
purpose. There is no hierarchical precedence for groups compared to
other groups. The behavior most often observed is that scene elements
will take on the properties of the group it was most recently assigned
(if it belongs to any). scene elements can be assigned to more than one
group simultaneously.
Layer - an organizational tool primarily for working in the viewports
and should not be confused with passes/partitions workflow. Layers have
precedence over groups and are most often used for assigning visibility
properties and wireframe colors. One example of using a layer involves
organizing a character into skeleton, geometry, controller layers for
easier manipulation. Layers are mutually exclusive - scene elements can
only exist in one layer at any given moment.
Pass - an instance of the scene organized for rendering purposes. Every
render pass contains every renderable element in the scene. Each render
pass is a clean slate and not tied to any other render pass. You can
have as many render passes as you like. Each render pass generally
generates one set of images related by a common purpose. Examples:
ambient occlusion, motion vector, diffuse, albedo. Each rendered image
is associated with a specific scene camera. In short, each render pass
has it's own render settings.
Partition - a mutally exclusive group within a pass. Each pass can have
any number of partitions. A partition is mutually exclusive - meaning,
elements within a render pass can only belong to one partition.
However, that same element may belong to different partitions in other
render passes. Partitions are most commonly used for applying material
overrides, visibility properties, and other wholesale changes to the
scene in the rendering context. At least one object partition and one
light partition must exist in each render pass.
summary:
Passes have precedence over partitions.
Layers have precedence over groups.
Layers and render passes can work together, but are generally targetted
for different purposes.
Matt
> -----Original Message-----
> From: softimag...@listproc.autodesk.com
> [mailto:softimag...@listproc.autodesk.com] On Behalf Of
> Stephen Davidson
> Sent: Friday, March 20, 2009 1:45 PM
> To: soft...@listproc.autodesk.com
Its all about propagation of properties.
Groups, passes both allow for applying properties/overrides to objects in a
non destructive way, but with a very different scope.
The group lives on the scene level. You can put it in a model, but that
doesnt prevent objects outside the model to also belong to the group). If
you want to apply some property to a lot of objects for the whole scene, you
can use a group. Eg. apply a material to a group, and add all objects and
clusters to receive this material to the group. another common one is
geometry approximation. The properties/overrides you apply to the group will
apply to all its members, for the *whole scope of the scene*, and will take
precedence over properties applied to the individual objects.
Groups are also commonly used for selection, eg: all deformers for a
character, to be used for enveloping, can be put in a group, so you can
select all of them at once.
Passes and partitions are a very comparable mechanism, BUT you have several
(unlimited) in the scene that live NEXT to one another.
What you do in one pass doesnt affect the other passes. Typically this is
geared towards the output of images (layers and channels) but its not an
absolute requirement. Sometimes I use passes to have different setups of the
same scene to work on different things: eg: for games: one pass would have
the scene setup for baking Hires to Lowpoly, and another pass would have the
lowpoly objects with the baked textures applied. By switching between passes
you can create the bakes on one hand, and verify them on the other - rather
than doing this in two seperate scenes, or having to switch a lot of things
manually all of the time.
Think of the properties and overrides you apply in a pass as "modifications"
on top of the properties the objects have in the rest of the scene,
specifically for the scope of this pass.
Eg, you're rendering an ambient occlusion (AO) pass - you dont want to lose
the material assigned to the object (say: a metal material) in the rest of
the scene. But just when rendering this AO pass you dont want to use that
metal material, but rather an AO material. So in this pass, you would put
that object into a partition, and modify the partition to render as AO.
There's a few possibilities:
You can apply an AO material to the partition and this material will now be
used for the objects in this partition in this pass.
But you could also apply an override to the surface port, and put an AO
shader in there. That way, only the surface part of the materials will be
replaced with the AO. This is usefull, because all the bumps and
displacements of the original materials are still there!
A third possibility is to construct a switch in your original shaders, that
allows to change between different looks, the standard metal look, the AO
look, a specular only look, you get the idea. Now with an override on the
partition, including just the "switch" parameter, you can set the object to
the AO look for this pass.
Layers are a visibility/selectability switching mechanism only - and take
precedence over the other ones.
Think of a car. You might put the main surfaces into a layer, and all small
details into another layer. That way you can quickly hide the details in
order to clean up your view, speed up your work.
A common approach is to use them while building a scene but then to totally
get rid of them when rendering your scene, and handle all visibility within
the passes/partitions.
So you could say layers are more geared towards the modelling/animation and
less towards rendering.
A word of caution: all these mechanisms can coexist and can be used in a
variety of ways.
You can get very smart and set up your scene to help you to do your tasks.
But then you hand the scene over to someone else, and he can't get anything
done in it, since he doesnt know the way you organised things.
So in a collaborative environment, use of these mechanisms can be subject to
convention.