1. Choose two clear portraits
Use one adult person per photo, with the face clearly visible and evenly lit. JPG, PNG and WebP are supported, up to 10 MB each. The image must be at least 240 pixels on each side. Avoid tiny faces, extreme side profiles and heavy filters.
2. Pick who stands where
Upload the first portrait on the left and the second on the right. Use the exchange button to swap them. Previewing here shows your input photos; it is not a generated preview of the final video.
3. Choose your lyrics and vocal language
Let AI write creates new short bars in English or Mandarin. Write my lyrics uses the words you enter: put each line on a new line, alternating left and right performers. Follow my lyrics preserves the original language, including mixed-language lines. Keep the lyrics short enough for your chosen duration; AI wording, pronunciation and lip sync can vary.
4. Choose the aspect ratio, quality and duration
The default is 480p, 5 seconds, vertical 9:16. Choose 720p for a sharper picture, or 10 or 15 seconds for more performance time.
Choose the aspect ratio alongside picture quality and duration. The default is 9:16; 16:9, 4:3, 1:1, 3:4 and Auto (adaptive) are also available. Aspect ratio does not change the credit cost.
| Format | Credits |
|---|---|
| 480p · 5 seconds | 80 |
| 480p · 10 seconds | 160 |
| 480p · 15 seconds | 240 |
| 720p · 5 seconds | 160 |
| 720p · 10 seconds | 320 |
| 720p · 15 seconds | 480 |
5. Sign in and review the cost
Confirm that both people agree to the use of their photos. Sign in with your email and password or Google, and purchase a one-time credit pack if needed. Your photos and settings can be restored in the same browser after login and checkout. Credits do not expire and no subscription is required.
6. Generate and check back
Generation typically takes 1–5 minutes; queues may take longer. Your task is saved in My Videos after submission, so you can leave the page. A browser waiting timeout does not mean generation failed.
7. Preview and download
Play the finished clip with audio, check the two faces and the performance, then download the MP4. The track includes AI vocals and an original beat, using automatic or custom lyrics. If you want another version of a successful take, it uses credits again.
When a take does not work
A confirmed technical failure returns the reserved credits. If the clip completes but a face or gesture differs from what you wanted, try a clearer portrait after reviewing the photo tips. For a task still marked as checking, contact support with its reference rather than submitting repeatedly.
Videos are generated with the Wan 3.0AI video model. Our MVP offers the four formats listed above.