We've recently had a slightly unsuccessful capture of walking animations - either the sleeves were too long or light was not even enough. Maybe a little bit both, I guess.

The main problem is: arms tend to lag behind and often stick to the sides of the body. The takes are very short, so it is not the end of the world to fix those positions manually. I was thinking though, should I always run jitter removal with zero (for arms) before making any manual corrections? I'm not even trying retracking since everything else already works. I only need to fine-tune the arms to correct positions. I'm still using v1, by the way.

You can run Configurable Jitter Removal (CJR) after tracking (this is how I usually use it.) My understanding is that CJR doesn't destructively change the original or the tracked data because it is applied 'on top' of the tracking in a separate 'layer' if you will. This means you can experiment and change the CJR settings and re-run it as many times as you like without worrying about 'degrading' the data beneath. (I should double-check that but I'm almost certain this is the case.)

Sometimes I find it necessary to use different CJR settings for different ranges of a motion. If you do this, remember to 'touch' the Trajectory Filtering slider to smooth out the transition between different CJR settings. To do this, just move the slider to a different setting and then move it back to the setting you want. This too, is non-destructive to the original and tracked motion data, but obviously it does affect the JR 'layer' (as it should.)

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!

I was thinking as well, that there are two layers. Sometimes it happens that a part of the animation is wrong again that I thought I've already fixed. This lead me wondering, am I really affecting the original layer or the filtered one.

Pale skin, white background and halogen lights seem to result a lot of tracking errors to arms. Have you ever used gloves (successfully)? We tried once but it was a mess, however, at that time there were many other things we didn't quite understand yet.

I'm assuming you're using PS3 Eye. When I'm tracking PS3 Eye data, I alternate between Analyze Actor and Refit Pose a few times whenever tracking accuracy starts to drift. This helps iPi Studio adjust for lighting changes as the performer moves around within the environment.

The above is partially irrelevant for Kinect because Analyze Actor isn't used for Kinect data tracking. However, using Refit Pose repeatedly in difficult tracking situations can be very effective.

Hope this is helpful.

G.

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

Watch a one minute excerpt on Vimeo now!