Hello all,
I’m new to Unreal Engine, after I rendered a video of character animation using the sequencer, I tried to transform the world 3D positions of characters’ joints to the 2D positions on the image plane to display them in the rendered video. My animation file is in BVH format, and I first performed an FK to obtain the world 3D positions of joints. Then I used the rotations and translations read from the attributes of CineCameraActor in the sequencer to construct my extrinsic matrix of the camera, and the horizontal fov together with the image size to construct the intrinsic matrix in python. However, no matter if I transferred the extrinsic matrix into the right-hand coordinate system same as 3D joints positions, or transferred the 3D joints positions into the coordinate system of UE5(x from right to left, z from bottom to up and y from outside to inside), I cannot get the correct 2D positions on the image plane. Only when the camera had yaw, can I get the correct 2D positions.
I’m wondering whether I can apply the rotations and translations from UE5 interface directly in python to construct the transform matrices, or should I do some transformation on them?
hey, did yous solved this?