Understanding PSD


Hey all.

For the last couple of days I’ve been reading about PSD, it’s differences from/ and advantages over SSD and Shape Interpolation. I tried to get as much information as I could find in the Siggraph and Disney papers and also from several other resources.

Several questions have arised and I hope I can get some answers here:

  1. Does PSD work without the initial skinning (SSD) or it is added over SSD?
    The question arised because the original paper compares the operations speeds of these methods. This confused me a bit, because if PSD is added over and needs skin to be present why compare their speeds?

  2. Is Pose Interpolator in Maya implemented according to the original paper?

  3. Is the term “pose reader” related to PSD? I haven’t seen this term in the original paper but have stumbled on it in some tutorials and scripts.

Because the deep math goes over my head and I skip the super techie parts I’d be grateful if someone could explain these points to me.


They explained all that in the super techie part. Did you skip it? :slight_smile:


actually I read everything. even the equations (though those don’t mean anything to me, unfortunately)

i’m new to the techie side of these things, though I’m generally familiar with this stuff and have used blendshape corrections in practice. so, forgive the ignorance


So what’s your problem then? Are you hitting a problem with the tutorials or scripts or in one of your rigs? Or do you just want someone to write you some theory?


It could be helpful for me to know if the Pose Interpolator in Maya is what the paper explains(to make sure that i generally comprehended the logic of PSD) . From what I understand, it should be based on the paper, but still.

As I understand PSD is just interpolation of shapes(real interpolation) + it doesn’t suffer from gimbal lock issues and the rest is the same as the old corrective blendshapes?


PSD can be relative to skinning, which of course would require the skinning, but it can also be relative to other shapes or any deformer. The pose driver/interpolator terminology just refers to whatever logic is driving the corrective PSD shape, to make sure it happens precisely as the other deformers are hitting the foundation shape that it is based on. These could be SDKs, Cone drivers, RBFs, whatever makes sense. Check out the SHAPES tool by Ingo Clemens - it’s excellent for authoring PSDs, and his tutorial vids here do a good job describing how to use Pose Interpolator or RBFs as drivers.



I’ll certainly check out the videos. thanks.

I haven’t used Shapes. What’s the advantages of it over using the new (Pose Editor) workflow in Maya?


That’s a matter of preference - I’ve been using SHAPES for about 5 years and love it for face rigging. The maya tools seem limited by comparison.