Editing semitransparent effects in video usually requires layering and precise integration. Inpainting alone isn’t enough as content behind the mask shifts. We found Aleph can handle this zero-shot. No task-specific training. Truly a generalist model. Can’t wait to see what else it can do!