Can I cut the ROI into smaller segments?
My actors did the T pose at the end of the action sequence, so I have to track backwards several useless frames just to get to the actual performance. It would speed up the process if I could just track the t pose and go straight to the action scene.

What I usually do is track from the last frame of the T-Pose to the first frame of the motion, and then I copy the pose from this frame. Then, I can go to the first frame of whichever Take I want to track, paste the pose there, click Refit Pose a few times, and I'm ready to track. Of course, this assumes that all the takes in a recorded video start off with similar poses.

When I want to record a completely different motion, I perform that in a new/different recording.

Thank goodness for the Start/Stop button on the motion controllers. This feature saves me from recording much 'junk' footage anymore.

In general, I try to keep my recordings as short as possible, rarely doing more than three or four takes, and doing more Takes only if the motion is very brief. I find it a lot easier to manage shorter videos with fewer motions in Studio instead of working with really long videos containing a lot of different actions.

Also, working with shorter clips makes it easier to toss away junk or rejected footage, potentially saving a significant amount of disk space. For example, if you have a really long video with just a few seconds of good footage, it's not possible to trim that video to reduce disk space--recording long videos just isn't very efficient.

I find that if I plan my actions and rehearse the motions a few times before recording, sessions tend to go very smoothly without errors or surprises. I wind up recording a lot of short videos but, like I said, this makes it really easy to toss out the 'bad stuff' when the session is done.

G.

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

Watch a one minute excerpt on Vimeo now!

Our tracking algorithm is incremental. So you cannot just track one segment, then jump over several frames and continue tracking. The tracking works only within continuous segments.
In each segment, you need to find a frame with a simple pose, which can be easy approximated with manual editing tools. Then you do the Refit Pose, and now you can track in any direction (forward or backward) from that frame.
Actually, this simple pose don't necessarily need to be the T-pose. The T-pose is just the simplest, which requires minimum edits to fit the model to the actor. Also, it is convenient to make adjustments to the model itself (height, body weight, colors etc). When you already have proper model, you can go right into the action part of your video, manually set approximate model pose to match an actor, Refit Pose, and continue with tracking.