How to Replace a Person With a Robot in Video Using AI

By /

Replacing a person with a robot in a video is now possible without building a traditional 3D CGI character, but the workflow still has several stages: prepare licensed footage, remove or isolate the original subject, create the replacement character, transfer motion, and composite the result. This updated tutorial preserves the original Canva–Runway–Viggle–Filmora method while correcting the parts of the software workflow that have changed.

Watch the original AI person-to-robot video editing demonstration on YouTube. The buttons shown in the video reflect the interface available when it was recorded.

What has changed since the original tutorial?

The tutorial uses Runway’s former Video Editor Projects and standalone Inpainting tool to remove the original dancer. Runway retired that project editor and deprecated several standalone tools in 2026. Its current help documentation points users from Video Editor Projects to Timeline Studio and from Inpainting or Remove Background to Edit Studio. Do not spend time looking for an old “All Tools” page that no longer exists.

Viggle has also advanced beyond the early Mix workflow. Its current V4 model focuses on motion transfer, character consistency, facial expression, multi-person scenes, and controls such as Character Refine, Smooth Motion, and Foot Lock. The core concept remains the same: provide a driving video and a character image, then transfer the movement.

How to replace a person with a robot in video

1. Choose footage you can legally edit

Record your own performance or download footage with a license that permits modification. The original tutorial uses Pexels. Check the current license and the individual asset page before using stock footage in advertising, client work, or a monetized video.

Select a clip with a visible full body, limited occlusion, stable lighting, and a background that can be reconstructed. Fast cuts, crowds, objects crossing the performer, and heavy motion blur make both subject removal and motion transfer harder.

2. Create a clean background plate

The original method paints over the performer with Runway Inpainting. In the current Runway interface, use the available Edit Studio or equivalent object-removal workflow. Upload the source clip, mask the person, inspect the reconstructed background across the full duration, and export the clean plate.

If the camera is locked, you may also create a clean plate from a frame where the background is visible. For a moving camera, background reconstruction must follow the camera motion. Watch for warped walls, repeated textures, floating shadows, and missing reflections.

Runway’s official deprecated-tools notice explains the current replacements for the old tutorial workflow.

3. Design the robot character

Use your own artwork or generate a character in Canva. Open the editor, choose Magic Media or the current image-generation feature, and describe the robot’s body proportions, material, clothing, view, lighting, and background. A full-body character facing the camera is easier to map to a dancing person than a cropped or extreme-angle portrait.

Keep the limbs clearly separated and avoid accessories that hide the body silhouette. Match the source performer’s approximate proportions and camera angle. Export a high-resolution PNG. If you plan to use the result commercially, verify the image-generation terms and avoid protected characters, logos, or a real person’s likeness without permission.

4. Transfer the motion in Viggle

Open Viggle AI. In the current Mix or motion-control workflow, upload the original performance as the motion source and the robot image as the character. If Character Refine is available, generate the additional reference angles before the main render. This can help preserve a detailed non-human design during turns and fast movement.

Choose a transparent or green-screen output when available so the generated character can be composited over the clean plate. Generate a short test first. Inspect the feet, hands, face area, body scale, and contact with the floor before processing a longer clip.

5. Composite the robot in Filmora

Open Wondershare Filmora. Place the clean background plate on the lower track and the Viggle output above it. Use Chroma Key for a green background or the appropriate masking/matting control for another output type.

Refine edge tolerance, spill suppression, scale, position, and shadow. Match the color temperature and contrast of the robot to the background. If the robot seems to float, add or rebuild a contact shadow and align the feet with the original floor plane. Export a short preview before rendering the final resolution.

Troubleshooting common results

  • Robot shape changes: simplify the character and use Character Refine or multiple consistent references.
  • Feet slide: align the source motion, enable available foot-lock controls, and correct position with keyframes.
  • Green edges appear: reduce chroma spill and add a subtle edge treatment.
  • Background warps: improve the removal mask or use a manually prepared clean plate.
  • Character scale changes: use a source clip with less camera movement or correct scale over time.
  • Motion looks detached: add a contact shadow, match lighting, and preserve the original timing.

Consent, disclosure, and commercial use

Use footage, characters, and likenesses you own or have permission to modify. Do not use this workflow to impersonate a real person, fabricate evidence, or create deceptive endorsements. Platforms may require an altered or synthetic media label when the result appears realistic. Review the current rules for the destination platform before publishing.

Plan limits and commercial rights vary. Viggle’s current site describes commercial use, but you should still read the applicable terms for your account and project. Stock footage, generated images, music, and editing assets can each have separate licenses.

Creator resources

Some links above are affiliate links. AI Tools Arena may earn a commission if you purchase through them, at no additional cost to you.

Related AI video tutorials

Conclusion

The quickest current workflow is to create a clean background, generate a well-posed robot, transfer the original movement with Viggle, and composite the result in an editor. The technique avoids traditional character rigging, but it still benefits from careful masking, reference preparation, edge cleanup, licensing, and consent.

Scroll to Top