Mo-cap Question

[brucelay]

Hi there

I have some facial mo-cap (fbx) data files. They are just the movements of a collection of nulls. But I wonder how I could use those data to move the geometry in Houdini? How can I use mo-mapped data to move Houdini's facial Geometry? Does anyone have any idea? I urgently need help! Please help.

thank you for your attention!

[MatrixNAN]

Hey Brucelay,

Wel you have 2 options or for a 3rd one you can do an combination of the 2.

1. You can lay out bones in the face and the nulls drive those bones to animate the face. Thanks to the new Houdini you can also layout volumetric muscles to go along with the skull bones and mandible if you feel so inclined.

2. You can use blend shapes for the face and then tie different blend to the nulls to be driven in a custom blending system. I would bind the blend shapes to the translation of the nulls to drive the animations. For instance if I had an animation of a sneer and then I created a blend shape for the null marker on the lip going up on that part of the lip then I would get the desired reaction the animation. Side to side for the nulls etc. You end up doing a lot of blends as you can imagine. This takes some time to setup. If you have that time then go for it. If you do not and don't want to do bones then I recommend trying Face Robot because it has a very fast setup for this. From what I understand though it does not take FBX it uses another format thats not that hot.

3. Do a combination of the 2 above.

Cheers,

Nate Nesler

[edward]

You need to output .bvh files from MotionBuilder and then use the standalong mcbiovision tool to convert it into script for use with Houdini.

[Jim Su]

Hey Brucelay,

Wel you have 2 options or for a 3rd one you can do an combination of the 2.

1. You can lay out bones in the face and the nulls drive those bones to animate the face. Thanks to the new Houdini you can also layout volumetric muscles to go along with the skull bones and mandible if you feel so inclined.

2. You can use blend shapes for the face and then tie different blend to the nulls to be driven in a custom blending system. I would bind the blend shapes to the translation of the nulls to drive the animations. For instance if I had an animation of a sneer and then I created a blend shape for the null marker on the lip going up on that part of the lip then I would get the desired reaction the animation. Side to side for the nulls etc. You end up doing a lot of blends as you can imagine. This takes some time to setup. If you have that time then go for it. If you do not and don't want to do bones then I recommend trying Face Robot because it has a very fast setup for this. From what I understand though it does not take FBX it uses another format thats not that hot.

3. Do a combination of the 2 above.

Cheers,

Nate Nesler

The problem with 1. is unless the face of the actor is the same as the face geomety, then you will run into problems. So you wouldn't be able to rely on the global position of a null. I recently worked on a model that uses Hillary Duff's facial mocap. I didn't do the setup, but my understanding is the translations drive the shapes. And I modelled over 100 shapes! So be prepared for lot's of blendshape work!