First i'm not an animator so please don't assume. If I use a keyframe software tool, the T pose is the reference frame from which the bones move.

Is there any benefit in doing the T pose with the human actor in DMC? I ask because having established a T pose and analysed the actor etc. the next thing we do is just move from that pose to a different part of the video and then have to align the character all over again so why did we do the T pose?

Also, I did a fairly subtle movement and it got totally wiped out by the jitter removal. It wasn't that subtle though, a stirring type movement of the hand. I'm not sure of the best technique to remove jitter yet not affect this movement.

Maybe an enhanced jitter removal where we can choose which bones should be filtered. Could this also be enhanced to say define a frame range where feet are planted and thus remove jitter in other limbs to avoid the foot sliding, would this work?

I've used 4 Playstation cameras @ 320x240 @ 50fps and apart from the above get good results. Happy so far!

The T-Pose gives the tracker a mostly consistent starting point to track from. It's assumed that you have a T-Pose at the head of each performance, but if you're doing a lot of takes after the T-Pose and want to skip over a lot of video then, yes, the T-Pose may not seem very useful.

Ideally, you would have the performer do a T-Pose at the head of each motion. Then all you need to do is move the saved T-Pose (just load the saved actor) and reposition it. Clicking Refit Pose once or twice is usually all it takes to get things aligned again. This is normally faster than trying to manually pose the rig into a non-standard pose and refitting it for each capture. (Which, I'm guessing is what you're doing?)

When I'm in the situation described in the first paragraph, my workaround is to turn off Shoulder and Feet Tracking, and to set the Tracking Resolution to Low, and then 'fast track' the video almost up to the performance I actually want to capture. (There's not much point in tracking motions you don't want in high quality.) With Low Quality settings, tracking will go faster and the rig will be more or less in position when it gets to the good stuff. Then I'll turn Shoulder and Feet Tracking back on, and switch the Resolution back to High, hit Refit Pose a couple of times, and then get on with my real tracking.

G

Edit: I should point out that I've made it a habit to do a T-Pose for each action. It may seem like a hassle on stage, but the practice can save you a lot of time during tracking.

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

Watch a one minute excerpt on Vimeo now!

Thanks. I think if the action all starts as standing then yes, doing a T at the beginning of each action woudl work, some of my "acting" is starting in some awkward positions so the transition from a T to the position is a bundle of limbs although my lighting is not great, something I need to work on (including a bigger house!)

I'll remember to set tracking resolution to low and turn off the shoulder tracking next time. So far for the main, i've been letting it track the whole way through. My graphics card does each frame in about 1.5 seconds so the overhead of just leaving it for an hour and a half for my fairly short stuff hasn't been a problem.

I guess I could throw in a T every now and again then :)

For me, the T pose is extremely important because I always set up my cameras manually, but I do not go into a T pose for every recording. Basically, inside my work area, I have an X on the ground. That is always my starting point. I record 1 Tpose, all by itself. Then I will record dozens of movements, and all those stay in the same folder as the Tpose recording.

When I process all the video, I do the Tpose and camera set up first, creating the Actor and Scene files. I then just load the videos, set up the tracking area, Refit the 1st frame a couple of time, and track all the video at low res. That usually takes a couple days and then the real work starts. I go frame by frame from then on, adjust the head and refitting each frame.

Jitter Removal is a dang handy tool that can really make a huge difference in the quality of the animation, but yes it will remove slight movements. When I think the animation is almost done, I run the jitter removal, only once. I never run it through an animation more than once. After jitter removal, I will go through the animation again, and look for areas that got lost or could be smoother and adjust those manually. If anything looks jittery at the end, I will smooth those frames out by upping the smooth setting on that frame, then copy and pasting it on the same frame. This saves the frame without affecting the frames before or after it.

When I export, I usually have smooth at 1 or 2 depending on the action.

I use T-Pose after actions that might be problematic, like fast movements or hidden body parts, just to reset the body position.