hello,anhungxadieu wrote: one more question, i have entire rig have done via ice, and for every matrix output from the rig i need to create a null in order to do an envelope i already done it, but it quite slow(about 4 fps) so i need to find a way to solve it. i think about point cloud rig. But i don't know how to dealing with envelope weight. Do you think i keep finding or stop? i know softimage is discontinue by Autodesk and more difficult to find a way to resolve my own problem.
Here is scene with walking Martians. Each one utilizes only a weighting part of SI envelope operator, while transformations are coming from point cloud. It has a classic rig, but only for building the basic structure and enveloping, rig is not animated. You can duplicate some of Martian, and change the 'driver ID' parameter, just to see what happens.
Regarding optimizations, I'd be free to point again, to structure of ICE enveloping compounds:
- completely static ( evaluated only once, probably on scene load) weighting, provided by SI envelope operator
- per object arrays of transforms - that is, evaluated each frame, but only for entire object, still not 'actualized' on point level
- per point evaluation of only small part of transform arrays, defined by array of indices
Beside standard rule of thumb, that everything which is possible to evaluate only once, it should be evaluated just and only only once - in systems like ICE or Houdini VEX, I think it worth to checkout, is it the thing *really* evaluated or not. For small example, even if Houdini has lazy evaluation of nodes, on deformation system I'm working, I'm constantly getting a small boost when static part is locked, manually, and scene is re loaded. 'Lock' in H = sort of non destructive freeze.
About ICE kinematics unfortunately I have zero experience.
Does it worth to still use SI - I'd say, your choice. Just personally I'm on Houdini right now, but, my Softimage version is 7.01 - after seven years perhaps there's time for change . Now serously, if you're making a system like body rig, hair system or so, imho most important is to get feedback of skilled artist, to get a picture, what to add or remove. Let's say, you're working as TD somewhere. If not... from my knowledge, only place on planet earth, where it was possible to get an skilled artist, ready to play with something created with visual programming system like ICE - it was Softimage community in times of what I'd call 'Golden ICE Age' around 2009 - 2012. In Houdini crowd, everyone seems to be an author, artists are rare animals, especially skilled ones.
However, there's chance in future, with Fabric Canvas and Implementation of H engine. I think we will have some more exact info by the end of this year.