What you need before you begin

Prepare two separate JPEG or PNG images: one image for Person 1 and one for Person 2. Each image should show a single subject with a visible face, clear lighting, and enough of the hairstyle, clothing, and body to guide the selected performance. Avoid group photos, sunglasses, strong beauty filters, very small faces, and poses that hide most of the body.

1. Choose one of the five current templates

Open the generator and preview the available performance cards. The template controls the movement, scene, camera sequence, and expected framing. It does not merely add a visual effect to a still image. Check where each performer appears so you can assign your two photos to the intended roles.

Templates can use either a vertical 9:16 or wide 16:9 composition. The generator selects the template's original ratio by default, and you can review the ratio control before starting.

2. Upload one photo for each performer

Add the first reference to the Person 1 slot and the second reference to Person 2. Only upload images you own or have permission to use. A clear, front-facing portrait normally provides more reliable identity cues than a compressed screenshot or a photo with several people in the frame.

  • Use even light and a face that is not covered.
  • Keep important clothing and hairstyle details visible.
  • Match portrait or full-body framing to the template preview.
  • Use a separate file for each person instead of a group shot.

3. Review the template and output ratio

Play the preview before generating. Confirm that Photo 1 and Photo 2 are assigned to the correct roles, then choose 9:16 for a vertical result or 16:9 for a wide result. The selected template remains the motion guide; the uploaded images guide each person's appearance.

4. Sign in and use one generation credit

Raindance AI uses Google sign-in to connect credits and completed videos to your account. One completed two-person generation uses one credit. Credit packs are one-time purchases rather than subscriptions. If a generation task fails, the used credit is returned automatically so you can try again.

5. Review the generated video

Inspect both faces, hands, clothing, body proportions, movement, and scene continuity. AI output is probabilistic, so the result can reinterpret details from either reference photo. A completed result is not guaranteed to reproduce every facial or wardrobe detail exactly.

6. Download and prepare your post

Audio availability depends on the selected template. You can replace or supplement it through the licensed sound tools available in the social app where you plan to publish. If the result contains realistic AI-generated people, review that platform's current disclosure requirements and apply the appropriate AI label before posting.

Troubleshooting weak results

If a result has an unstable face or clothing, try clearer source photos with fewer obstructions and lighting closer to the template. If the two people appear in the wrong roles, return to the generator and swap the Person 1 and Person 2 files. For a technical failure rather than a visual variation, contact support with the approximate generation time and task reference.