Still playing with Ltx & Wan 2.1 in @comfyui . I think models can be swapped out easily especially for API ones. Better accuracy can be achieved if needed but for now, testing the workflow of using depth, crypto and motion vector passes from your 3D file to composite an ai output. It’s quite feasible, despite this Wan2.1 output being a little off from the playblast (nothing a few blurs can’t fix)

Tbh still not too sure how i feel about this, looks abit like slop, BUT this workflow definitely has potential with other models.