If the AI doesn't completely remove the target object, first review your prompt for clarity. Vague descriptions like "thing" or "stuff" can confuse the tracker—try being more specific (e.g., "blue car in the background" instead of "car"). Also, check your footage: shaky video, rapid motion, or objects that blend into the background can reduce accuracy. If the issue persists, try processing a shorter clip or adjusting the camera angle in your source footage. For complex scenes with multiple overlapping objects, you may need to run the model multiple times, removing one object per pass. Alternatively, consider using
Wan VACE Video Edit for more advanced control over masking and frame-by-frame adjustments. Testing with a 5-second sample before processing the full video can help you refine your approach and save credits.