Change a talking-head scene with Gemini Omni

Step by step

Run the workflow

1. 01

   Prepare the source material

   Use the source inputs to complete this stage and produce a result that can be checked before continuing.

   • Click here to open Google Flow (opens in a new tab)Official tool link added because the source does not include a working tool destination.
   • Click here to open ChatGPT (opens in a new tab)Official tool link added because the source does not include a working tool destination.

   1. Record a talking video of yourself.

   

   Common failure modes

   • The generated scene changes the person, camera, room, product, or timing beyond the requested effect.
   • Fast motion hides face, hand, clothing, text, or continuity errors that become visible when the clip is paused.
2. 02

   Copy the generated prompt for Phase 3

   Use the source inputs to complete this stage and produce a result that can be checked before continuing.

   1. Personal Creative Director
   2. Drop my prompt (Next Page) into ChatGPT and modify the timing and the changes to your desired scenes. Copy the generated prompt for Phase 3.

   

   Common failure modes

   • The generated scene changes the person, camera, room, product, or timing beyond the requested effect.
   • Fast motion hides face, hand, clothing, text, or continuity errors that become visible when the clip is paused.
3. 03

   Start the video with the original clip with no changes

   Use the source inputs to complete this stage and produce a result that can be checked before continuing.

   1. Copy the prompt below and replace any project-specific subject, product, location, or format details.
   2. Attach the source assets in the same order described by the prompt, then run one controlled test.

   Working asset

   Start the video with the original clip with no changes. The face and hair must look original.
   
   At [x] seconds, after he snap his finger, transform the hairstyle into a nice looking korean parting hair.
   
   At [x] seconds, transform the clothing into a luxury tailored black suit.
   
   At [x] seconds, after he snap his finger, replace the background into a nice looking bar that matches the video lighting.
   
   * Preserve the exact same person. * Keep facial features unchanged. Face must look exactly like the same person * Maintain facial proportions, skin texture, age, ethnicity, and expressions. * Preserve eye shape, nose, mouth, jawline, and hairstyle transition realism. * Keep lip sync perfectly aligned with the original speech. * Do not modify the voice, dialogue, pacing, or performance.
   
   * Match lighting between subject and environment. * Maintain realistic shadows and reflections. * Preserve camera angle, framing, movement, and depth of field. * Keep all non-specified elements unchanged. * Photorealistic result with seamless transitions and no identity drift. * Do not change the framing throughout the video
   
   Generate a Gemini optimized video to video edit prompt for me based on these details.

   Common failure modes

   • The generated scene changes the person, camera, room, product, or timing beyond the requested effect.
   • Fast motion hides face, hand, clothing, text, or continuity errors that become visible when the clip is paused.
4. 04

   Assemble the video

   Use the source inputs to complete this stage and produce a result that can be checked before continuing.

   1. Now Edit your video with your newly generated image.
   2. With your assets ready, it's time to edit your footages!

   

   

   

   

   Common failure modes

   • The generated scene changes the person, camera, room, product, or timing beyond the requested effect.
   • Fast motion hides face, hand, clothing, text, or continuity errors that become visible when the clip is paused.

Before you use it

Final checks

• Check the full result for changed people, products, text, timing, or layout.
• Confirm the tool, model, controls, and cost before generating.
• Keep private or unreleased material out of third-party tools.