For instance let say you begin production, all goes well, you do a
significant bunch of shots. Then someone believes there are things
he/she is not happy with, and changing it would require major changes
in the rig a certain characters. How do you deal with this?
- if you rigged everything the same way so that animation is entirely
reusable, then ideally you have to propagate the change to every
character, otherwise you can no longer exchange data between those
characters and the others.
Yes, absolutely true. Same would be true with local data, no?
How do you solve this with any solution!?
- there is only so much you can change without breaking
from existing
assets. For instance, if you need to modify a leg rig so much that you
don't see how existing animation could be applied on the leg and have
good animation, then all animation done on this rig will not be usable
anymore with this leg.
Yes, absolutely true. Same would be true with local data, no?
How do you solve this either?
- of course, clever td can work out the changes so that
custom tools
can be written to patch the animation on the new rig after it's been
applied. In my case I don't think I fall in this category of td.
I certainly couldn't do it.
- finally, if you read the reference model always from the
same
location (that is, you don't expose model versions), then upgrading
this model may break things right away.
This is true, but can be overcome with a little discipline and a good
back-up before replacing work flow.
I believe you can go a long way with ref models, but it
becomes hard
to deal with when you need to fix problems that arise later in the
production.
I think that's what opened up this whole can of worms in the first
place.
The limitations of the Delta's don't allow us to fix the problems, and
force us to
localize data.
No matter what I don't think there is a perfect solution.
Here we have
a database-driven pipeline for models, which includes custom tools to
replace them, upgrade them, export and import their animation. We
don't need to localize things on demand. But most importantly, it
means that data made with these assets keep working as long as the
same assets are used.
That sure sounds pretty frickin perfect to me. :-)
That said, there are errors, you can't offload anything, you need to
be tight on scene cleanup, etc. Personally I find easier to handle
these problems than those incurred by the usage of ref models.
Well, I would to, but not everyone has access to a database-driven
pipeline.
And Reference models give us some of that flexibility with the off-the
shelf software.
I just would like the Delta to incorporate more, as Todd said.
But what you are describing sounds pretty damn good to me, sign me up!
When are you going to release that, so I can scrap my Referencing
ways!? ;-)
|
