fabric engine skinning research

Forum for users who have migrated or are migrating to the Fabric Engine
User avatar
Mathaeus
Posts: 1708
Joined: 08 Jun 2009, 21:11
Location: Zagreb, Croatia

Re: fabric engine skinning research

Post by Mathaeus » 05 Mar 2016, 15:22

anhungxadieu wrote:
Mathaeus wrote:I'll try to do working example this weekend, should be a good exercise for something else close to that, where I'm personally interested. So, expect the victory :) or.... admittance of defeat.
wait to see it! :)
Defeat. While seems to be possible to avoid 2d array by mentioned unpleasant setup, problem is that array of transforms (the one you built directly in your canvas example, d1 ... d2...), when coming via Alembic, it is completely unrelated to arrays of envelope indices created by SI. And I don't see usable way to bake out this one into mesh, simply because it does not belongs to mesh. Which is expectable, as array of transforms in SI is sorted by SI group (afaik). While Alembic (afaik) don't know for let's call it complete enveloping construction (won't keep related arrays in sync). So in best case of Alembic and mentioned setup, there's need for remapping the array of transforms, and filtering the local arrays too (could be one, two, three or more transforms assigned to point). Too much of workarounds, for my taste -at least in only possible way for me, which is FBX from SI to Houdini, Alembic from H to FE (maybe some part is left on the road).
FBX is looking as much better choice for getting the all that envelope elements in FE, as it supports enveloping for decades. But, now I'll need to figure out the FBX IO stuff in FE.
The renderer "turtle" used by this scene, is not currently available. The "turtle" renderer will be used instead.

Pooby
Posts: 492
Joined: 27 Aug 2010, 22:25

Re: fabric engine skinning research

Post by Pooby » 05 Mar 2016, 20:35

Can't you force it to belong to the mesh with ice?

Edit Oh I see what you mean I think . The transform array in fabric isn't the same order as si. That's always going to be the case from whatever dcc isn't it?

User avatar
Mathaeus
Posts: 1708
Joined: 08 Jun 2009, 21:11
Location: Zagreb, Croatia

Re: fabric engine skinning research

Post by Mathaeus » 05 Mar 2016, 23:11

Pooby wrote: The transform array in fabric isn't the same order as si. That's always going to be the case from whatever dcc isn't it?
Perhaps there is some pattern, but, in any case, chances are huge for mismatch between hierarchy and envelope transforms. What's happening with SI effectors or Max Bipeds nubs, for example. That's why FBX sounds good for me, as it *has* to write special array related to envelope. In Houdini I had a custom mapping template from names to numbers, it seems it's possible to do similar using 'filter paths' in 'get paths' (Alembic nodes), maybe with something else.

P.S. for now it seems I'm in 'one answered question creates three new questions' phase, :D . Reminds me to good old first months of ICE, hehehe.
The renderer "turtle" used by this scene, is not currently available. The "turtle" renderer will be used instead.

Pooby
Posts: 492
Joined: 27 Aug 2010, 22:25

Re: fabric engine skinning research

Post by Pooby » 06 Mar 2016, 00:06

Can you write arrays of strings in ICE? That might make things clearer.

That is, If alembic supports them and subsequently could bring them into canvas.

User avatar
Mathaeus
Posts: 1708
Joined: 08 Jun 2009, 21:11
Location: Zagreb, Croatia

Re: fabric engine skinning research

Post by Mathaeus » 06 Mar 2016, 01:14

Pooby wrote:Can you write arrays of strings in ICE? That might make things clearer.

That is, If alembic supports them and subsequently could bring them into canvas.
Well just personally can't, my XSI is 7.01, that's reason for exporting FBX to Houdini, to export Alembic from there. Anyway just watching alembic tutorials, makes some things clear, perhaps I should change practice to watch tuts only as last option :D .
For things I (personally) planning to create in Canvas, actually it's enough to have an animated hierarchy, 'weighting' is procedural.
The renderer "turtle" used by this scene, is not currently available. The "turtle" renderer will be used instead.

anhungxadieu
Posts: 175
Joined: 17 Apr 2014, 10:39
Skype: nguyenvuducthuy

Re: fabric engine skinning research

Post by anhungxadieu » 06 Mar 2016, 04:47

Mathaeus wrote:
Pooby wrote:For things I (personally) planning to create in Canvas, actually it's enough to have an animated hierarchy, 'weighting' is procedural.
well, i think it's harder and harder due to FE not have implement 2d in canvas yet. Bring a mesh from ddc to fabric and do a skinning maybe more difficult than do all skinning in fabric itself.

User avatar
Mathaeus
Posts: 1708
Joined: 08 Jun 2009, 21:11
Location: Zagreb, Croatia

Re: fabric engine skinning research

Post by Mathaeus » 06 Mar 2016, 12:03

anhungxadieu wrote: well, i think it's harder and harder due to FE not have implement 2d in canvas yet. Bring a mesh from ddc to fabric and do a skinning maybe more difficult than do all skinning in fabric itself.
Agree with that, of course if we are counting nodes, only. By the way noticed some FBX related nodes like 'parse skin weights', but please don't ask me how to connect it :). In any case lack of nodal support for 2d arrays (once point is already member of 1d array) *or* nodal support for loops, leads to a lot of 'funny' cases, for example how to create a plain particle trail, and so on.
Just for reference, Softimage ICE is still only visual programming system on planet, able to deal with enveloping/skinning in full scale, using only nodes, since 2008 - so, for coders of another apps (who usually knows nothing about ICE), requests for nodal skinning probably sounds like blasphemy...
The renderer "turtle" used by this scene, is not currently available. The "turtle" renderer will be used instead.

Pooby
Posts: 492
Joined: 27 Aug 2010, 22:25

Re: fabric engine skinning research

Post by Pooby » 06 Mar 2016, 13:55

Interior graphs are coming in canvas.
If I understand correctly, This feature should allow a per point parallel process, and be able to handle this kind of thing.

anhungxadieu
Posts: 175
Joined: 17 Apr 2014, 10:39
Skype: nguyenvuducthuy

Re: fabric engine skinning research

Post by anhungxadieu » 06 Mar 2016, 15:06

"for loop" or "2d array ... " thing like this should be implement first because i think it's very need!
Last edited by anhungxadieu on 05 May 2017, 07:31, edited 1 time in total.

anhungxadieu
Posts: 175
Joined: 17 Apr 2014, 10:39
Skype: nguyenvuducthuy

Re: fabric engine skinning research

Post by anhungxadieu » 05 May 2017, 07:30

Recently i have time to do some r&d regarding skinning in Fabric Engine again, and here is how it going :D

User avatar
Mathaeus
Posts: 1708
Joined: 08 Jun 2009, 21:11
Location: Zagreb, Croatia

Re: fabric engine skinning research

Post by Mathaeus » 05 May 2017, 10:09

Nice ! Just wondering, how you got weights and bind pose, do you have some hidden connection to your FE operator.
The renderer "turtle" used by this scene, is not currently available. The "turtle" renderer will be used instead.

anhungxadieu
Posts: 175
Joined: 17 Apr 2014, 10:39
Skype: nguyenvuducthuy

Re: fabric engine skinning research

Post by anhungxadieu » 05 May 2017, 11:53

Mathaeus wrote:Nice ! Just wondering, how you got weights and bind pose, do you have some hidden connection to your FE operator.
This is Alembic attribute. Actually, i Just modify the source code from rigging toolbox has been provide by FE and try to get my head around with that and finally i got this result. The original code don't work with recent release of FE. So i remove all stuff i don't need. it's quite dirty right now but properly it works.