Hi guys!
This mocap approachs looks very interesting and prommissing to me, so I have a few questions.

1) Is a three Sony Eye camera setup enough to track motion or do I need
anything else (except of the software, of course) (just to make that clear for me ;) )

2) How do you track large scale scenarios with only one workstation? isn't the capture space limited by
the cable length of the cameras? Can I capture from multiple computers and sync the streams later on?

3) What about tracking the head movement? I've read in the wiki that's not supported yet. So the head
and the hand bones are not animated at all, right? (again, just to make sure...)

4) Can I rename bones / change the skeleton to fit certain needs (before export)

I think thats all for now... I'm waiting for your answers! :)

1) Three works but it's less accurate than four or six cameras. You'll probably need to do more motion correction and apply stronger jitter removal to the results; also, with fewer camera, occlusion becomes an issue.

A minimum of four is recommended; I've gotten very good results with four, requiring very little clean up. Obviously, six is much better.

2) With PS3 Eyes, the capture space can be quite large. (Watch the Danse Kabyle making of video to see a great example.) All you need is a set of USB Repeater (a.k.a. Active Extender) cables. These are extension cables with a small box on the end that boosts the signal. Depending on the cable and your USB controller, I think you can daisy chain up to about 80 feet. (I once used one chain of three for about 48 feet and this worked fine.)

Technically, you can capture using multiple computers but IMO this is more trouble than it's worth because you will need to composite all your video footage to a single video in post, and the frames might not sync exactly. Better to just capture using all cameras synced to one computer and deal with a single video file.

3) Head tracking is coming, hopefully soon. You can animate the head in a 3D program, which is generally not difficult if you target the head and simply point the target to what the character is looking at, or just keyframe it, or do both if you can blend IK and FK; whichever works best for your shot.

Hands are not captured but this is usually something you want to keyframe anyway, especially if the hands need to touch 3D props or the environment. Fingers even more so.

FWIW, when we get mocap from high-end facilities, we typically only get wrist rotations, and we usually wind up adjusting the rotation to suite the scene. We used to get finger capture but found that we usually wound up reanimating the fingers to suite the action in the shot, so this seemed pointless.

4) It's not necessary. Typically, you retarget the data to your own rigs, and the bones can be named however you want. That said, it's best to use conventional names to simplify/standardize workflow, especially if you're using a program like Motion Builder or Animeeple to retarget. BTW, iPi Studio can retarget too, though I haven't used this feature yet. (I typically use Motion Builder.)

Hope this helps.

G.

_________________
Greenlaw
Artist/Partner - Little Green Dog | My Demo Reels (2013,) (2015,) (2017,) and (2019)

Watch a one minute excerpt on Vimeo now!

Hi Greenlaw!
Thank you very much for your explanation! I've already found out some answers on my own, but to hear personal experience is
much better. :)
You are right, with a common optical mocap system you also don't get propper hand/finger movements so you have to do it later
on.
I wanna capture actions for massive so I have to make sure that the skeleton fits massive's needs and foods are sticking to the
ground without spending too much time in Motionbuilder etc. to clean up the data.

many thanks!

Seb.

With the PS3 Eye system, stable feet usually isn't a problem. You will see jitters in the first pass of tracking but applying iPi Studio's CJR (Configurable Jitter Removal) to the legs fully stabilizes them. (You will likely need to put at least a little CJR on all limbs, and legs typically need more.)

I think iPi Studio could be an ideal low-cost system for Massive, especially because you don't typically get too close to Massive animations to see small flaws in motion.

As for Motion Builder editing, if it's shot well I don't find I need to spend too much cleaning up the raw .bvh data. After retargeting however, I spend time more tweaking arm rotation limits but that's because my characters they have such HUGE heads. :)

G.

_________________
Greenlaw
Artist/Partner - Little Green Dog | My Demo Reels (2013,) (2015,) (2017,) and (2019)

Watch a one minute excerpt on Vimeo now!

Hi,

You can use the kinect camera and a laptop on a rolling rig like a cart or chair with wheels to cover long distances...

And also I saw that if you do a crawling motion or push ups it will track the hands..

Any limb or appendage that touches the floor in iPiSoft Kinect will track,.

Try it!