Just ran across this, wondering if It's a feature of some sort? or I found an obscure bug?
If you build a fairly default vellum cloth simulation example like so:
grid -> vellum config cloth -> vellum solver
and your geometry has a primitive attribute called "color" on it,
the sim cloth geometry will absolutely freak out and worm its way all over the place in an indeterminate way.
removing that attr will restore normal operations...
is this a sacred attr name that should be avoided for vellum? (or a bug?)
I searched the docs and couldn't find anything regarding an attribute named "color"
regarding vellum constraints nor vellum solver.
Vellum "color" attribute
1058 2 0- redpaw
- Member
- 75 posts
- Joined:
- Offline
- tamte
- Member
- 8786 posts
- Joined: 7月 2007
- Offline
redpaw
and your geometry has a primitive attribute called "color" on it,
the sim cloth geometry will absolutely freak out and worm its way all over the place in an indeterminate way.
this should not happen with color attribute on the Geometry
however on ConstraintGeometry yes
i@color attribute on constraint geometry is computed graph coloring value that marks independent constraints that can be solved in parallel
it is created/updated internally whenever ConstraintGeometry changes topology or if it doesn't exist
so if you are creating it explicitly then you need to make sure it's correct, you can use Graph Color SOP for that
you can of course submit a bug as it may be safer if the attribute name used internally is more protected, like __color__ to minimize a chance of user unintentionally running into this issue by just creating an attribute called color on ConstraintGeometry
Edited by tamte - 2024年1月17日 01:25:00
Tomas Slancik
FX Supervisor
Method Studios, NY
FX Supervisor
Method Studios, NY
- redpaw
- Member
- 75 posts
- Joined:
- Offline
Yah I may submit a bug on that. If the input geom you feed into the constraint node has "color" on it already, and usually its not a single integer attr, just by the nature of what most would consider a "color" it seems to just pass that thru to the constraint geometry and its definitely bad.
Its definitely a "simple" attribute name that I feel like would/could get used a lot, and unfortunately is part of our default pipeline currently.
Thanks for the quick response.
Its definitely a "simple" attribute name that I feel like would/could get used a lot, and unfortunately is part of our default pipeline currently.
Thanks for the quick response.
-
- Quick Links