How would I go about modifying the tracked skeletons’ transforms (without Fusion)?

Same question as How would I go about modifying the tracked skeletons' transforms? - #8 by ZanderCodes , but for the standalone BodyTrackingSingle project, not the Fusion receiver.

I’m coming off the Unity sample projects in which I was fairly quickly able to find and remove the offsets so (assuming an equal scale) the debug skeleton always rendered in the same place and showed the raw-data ground truth of the pose the avatar was “supposed” to be in (modulo the Unity project having avatar pose bugs with almost any avatar but the pack-ins).

In UE5.3, I’ve uncovered Spawn Transform parameters in both the BP_BodyTracking_Visualizer and BP_SDK_Skeleton_Visualizer, but they both appear to be 0,0,0 and yet the debug skeleton is reliably a meter or two left of the avatar.

Where is the data that controls the actual respective visual offsets? Is it even in a blueprint or will I need to crack open some cpp files?

This seems weirdly disorganized, but I’m seeing a hard 0,-100,0 applied in BP_SDK_Skeleton MoveNodesToKeypoints (but not MoveNodeTo) and another pair of hard 0,-100,0s applied in MoveBoneTo (but not MoveBonesToKeypoints). It shouldn’t take much to replace these with a Vector variable and wire it up and out to the Event Graph of BP_SDK_Skeleton_Visualizer (Event BeginPlayCreate the avatar if it does not exist yet block). I’m hoping this is the extent of the offsets, though, because I’m not sure there’s a way to search blueprints for “all occurrences of a vector matching (x,y,z)”

Hi,

I completely agree with you, having a hardcoded offset like is a mistake.
I’m adding this in my backlog so I can update it later.

Thanks for the feedback.

1 Like