RE: Reference model limitations?

["Vincent Fortin" <vincent.fortin(at)ubisoft.com>]

Hi Jeffrey, thanks for the feedback.

 

I am still interested in knowing what kind of pipeline other people use for character finaling.

 

Here’s what I have so far (I should have mentioned it in the first place, that’d have avoided confusion):

 

I am able to sim from within a reference. In fact I already have an optimized and fully-workable fx rig linked using Delta to the original animation. The rig is a mix of envelope and Syflex and I wouldn’t be afraid to add particles in there if I needed to (although I haven’t tested it). All this is very neat and procedural and most of all it’s ready to deploy over several shots without any further need for modifying the rig. Updated animation follows naturally on the fx rig since both rigs are part of the same Delta reference. The cache isn’t a problem since it can be specified per shot like any other parameter in the setup. I am confident from experience that 50% of the work can be saved that way and I’d push my craziness further by stating that even a “normal” animator could submit for approval some of the simple shots that don’t require tweaking the cloth settings.

 

Now this all falls apart when someone in the pipe tries to add a collision object to the setup (which is likely to happen) or change the content of a cluster (for optimization/tweaking purpose). I’m quite open to use my imagination but so far I am forced to admit that there is no workaround to my problem except committing to local which is something I translate into pipeline failure.

 

Vincent

 

---

From: owner-xsi(at)Softimage.COM [mailto:owner-xsi(at)Softimage.COM] On Behalf Of Jeffrey Dates
Sent: Wednesday, July 25, 2007 5:38 PM
To: XSI(at)Softimage.COM
Subject: Re: Reference model limitations?

 

I have dreamt of a 'scene stack' for referenced models so that you could manipulate them as you need, but the original model's integrity is maintained.  I am hoping this 'stack' mentality will propagate into the Delta paradigm.

6.0 has made some improvements with Delta's.. but I still find them extremely limiting.  And as you've mentioned here, at some point you might as well just localize your model to get into finessing.

>...referencing after all is mainly for the benefit of the modelling/animation stage.

Unfortuantly, that is the current paidigm we're forced to adopt if we use Ref's.  However, the concept can be imagined throughout the entire pipeline.
I'm not sure if anyone has really sat down with Softimage and illustrated how an ideal non-linear pipeline utilizing referencing models would work through production to render.  I'm amazed at how many studios just avoid them completely, either for fear of bugs/limitations, or just lack of really understanding them.

I am a huge advocate for reference models during production, and even with their obvious limitations I find them helpful in most cases.  It's just at some point on production you're forced to localize them for local scene adjustments, and then that's it...   You're local.  This only becomes a chore if you have a global change to your model that needs to propagate across your entire production.

Things like move components, clusters, shaders, groups..etc would be extremely valuable at the Delta level.  Ideally, I'd like to have my Geo independent of my rig, independent of my animation, and have the Delta bring them together upon loading the scene.

But as Andre has said, you are able to Sim from within a Reference and utilize it's cache at rendertime....  

I couldn't live without them...  limited or not. 
I have faith that Softimage will continue to push Delta's and open them up in the future. *crosses fingers*.  Assuming that more studios lean on Soft for the need.

--
Jeffrey Dates
www.kungfukoi.com
214.280.4353

Vincent Fortin wrote:

Hi there,

 

How come one can’t add operators and clusters to reference models? I’d have expected to be able to stack stuff on top of them while keeping the original file’s integrity.

I can only guess this is somehow due to XSI’s object structure that doesn’t have any special mechanism for handling this. Otherwise, from a user point of view, it just doesn’t make sense.

So assuming I didn’t overlook some basic info in the doc, how would you guys go about implementing a character finaling workflow in which an animator could do something as simple as connecting a ground collision to a referenced character that has Syflex on it?

 

Thanks

 

Vincent

 

No virus found in this incoming message.

Checked by AVG Free Edition. 

Version: 7.5.476 / Virus Database: 269.10.19/917 - Release Date: 7/25/2007 1:16 AM