Have you guys seen this:

http://www.clip-studio.com/quma/en/ [clip-studio.com]

http://quma.jp/en/quma/ [quma.jp]

A real pose able character to capture animation.
Do you think this could work for Houdini?

Would love to learn how to use this, only for sale in Japan at the moment.

It uses BVH format.

~Shawn

It looks interesting. If it outputs BVH, then you can import that into Houdini using the mcbiovision [sidefx.com] command line tool that comes with it.

I wonder how it works out in practice though with this type of posing a doll. Note that there was also this project [vimeo.com].

Having a “doll” to pose would be much quicker than using a mouse.
Think about it: One 3D handle at a time or two real hands? There are many other advantages: like my eyes being the camera. The model(doll) can be in any orientation and space without me having to constantly gyro around it in 3D environment. Instead, I can keep hands on the model; much faster than traditional animation. I would set up a foot pedal to key frame and fly through some complex sequences. Secondary bones can be added later to refine 3D character. I like this approach over human motion capture, which can seem too refined. Houdini could use something like this to boost character animation. Would be great to create all sorts of models like this product. Think of a quadraped where you can animate each leg seperately. The only problem is getting the model back into a pose to continue where you left off. Which is why the servo model you linked is neat. I'm very interested in this new Japan product. Maybe they can lend some outputs to practice bringing into Houdini.