How do I turn a script into a narrated explainer video in ChatGPT?

Split the script into three or four beats, generate a start frame for each with Nano Banana 2, render the scenes in one MiniMax H3 Max generation of up to 15 seconds, then add an ElevenLabs voiceover and captions.

By Itay Dressler, Creative Claw · Updated

Steps

  1. Keep the script to about 35 words for 15 seconds, split into beats such as problem, product, how it works, and result.
  2. Generate a product reference, then one frame per beat that uses it, so the product looks the same in every scene.
  3. Ask for a 15-second MiniMax H3 Max video with timecoded scenes and hard cuts, passing the frames as references.
  4. Generate the narration with an ElevenLabs voice, place it so each sentence lands on its scene, and caption it.

Tips

Show icons rather than words on phone or laptop screens, because generated video blurs small text. Put exact wording in the captions instead.

Still stuck? Email [email protected] with the job ID, or browse all help articles.