It looks like Blend Linear and Quat option for Bone Deform skinning method is not supported in Mantra.
If I bring an FBX in thst has blend lin quat as the method, Mantra will render the rig in Tpose when anim is present, if this is changed to linear, or quat the weighting is read and its good again…
Is this a known issue in H18? That Blend Lin Quat is not supported?
Please advise…
Issues with Bone Deform and Mantra
1034 4 2- rossm
- Member
- 3 posts
- Joined: June 2013
- Offline
- jsmack
- Member
- 7803 posts
- Joined: Sept. 2011
- Offline
This shouldn't have anything to do with Mantra. The skinning is evaluated in sops and generic polygons are sent to render. If the Blend Linear and Quat mode is selected, a weight attribute is expected to control the blending between linear skinning and quat skinning. If no weighting attribute is specified, the skin is output at t-pose and no deformation is applied.
- rossm
- Member
- 3 posts
- Joined: June 2013
- Offline
Okay…. Does anyone know if FBXs generated from Maya with Weight Blended skinning pass this attribute correctly, or not? because I do have a FBX attribute automatically bound to that attribute parameter in Houdini, but possibly it has incorrect data?
Also I do want to mention, that the reason I specifically mention Mantra, is because the issue only occurs during rendering. The imported FBX with blended quat/linear works perfectly fine in the view-port, and if used in a crowd/animation…. the loss of weights and t-pose default only occurs AT Render time…. so this attribute you mention seems to only be relevant at render time…. or possibly it is just disregarded for view-port calculations?
Also I do want to mention, that the reason I specifically mention Mantra, is because the issue only occurs during rendering. The imported FBX with blended quat/linear works perfectly fine in the view-port, and if used in a crowd/animation…. the loss of weights and t-pose default only occurs AT Render time…. so this attribute you mention seems to only be relevant at render time…. or possibly it is just disregarded for view-port calculations?
Edited by rossm - May 21, 2020 16:43:44
- rossm
- Member
- 3 posts
- Joined: June 2013
- Offline
Okay…
so I have created a test file, and pretty much got to the root of the issue… it seems that the agentLayer operator is blasting my weights away and I definitely need to use the agentLayer when supplying a rig that has multiple geo options attached to a single skeleton, as common with many crowd assets….
If you take a look at the file there are 2 outputs, one reading a simple chain with dual quat (this works perfect in the view-port, and in mantra), but if you view the “blended” setup, this has weight blended skinning method in Maya, and correct comes into Houdini as dual quat blended, the view-port looks correct, but if you fire a mantra render you will quickly see that the result is back to default rest position…..
-Ross
so I have created a test file, and pretty much got to the root of the issue… it seems that the agentLayer operator is blasting my weights away and I definitely need to use the agentLayer when supplying a rig that has multiple geo options attached to a single skeleton, as common with many crowd assets….
If you take a look at the file there are 2 outputs, one reading a simple chain with dual quat (this works perfect in the view-port, and in mantra), but if you view the “blended” setup, this has weight blended skinning method in Maya, and correct comes into Houdini as dual quat blended, the view-port looks correct, but if you fire a mantra render you will quickly see that the result is back to default rest position…..
-Ross
- jsmack
- Member
- 7803 posts
- Joined: Sept. 2011
- Offline
-
- Quick Links