MoCha performs Video-To-Video character swaps using reference images, replacing a video's character without per-frame pose or depth maps. Ready-to-use REST inference API, no coldstarts, affordable pricing.