Separate portraits
Two inputs, explicit roles, independent photo choices.
Already have a good photo together? Use it. Prefer separate portraits? Add one for each role. Create a shared elevator moment without first filming the two of you in the same room.
The selected preset guides the new scene. The example and your render can differ in timing, sound and appearance.
Choose the input with the clearest faces. Neither arrangement guarantees likeness.
Two inputs, explicit roles, independent photo choices.
One input with both faces and their left-to-right relationship.
This source reference demonstrates the headphone exchange. Your photos supply the cast; the generated scene, timing and sound can differ.
Reference media from the source site shows the creative direction. It is not a result generated by this service; your output may differ.
Separate portraits give you control over each role. A group photo gives the model both people’s relative positions and lighting in one image. Choose the input you can make clearest, then keep both faces visible through the encounter.
Begin with the listener absorbed in the headphones.
Let the second person interrupt with a glance and a short remark.
Finish with a smile or small reaction that reads even without sound.
Pick the input arrangement before you add files; changing that arrangement clears the photo slots.
Separate needs two photos. Together needs one existing photo with both adults visible.
Photo 1 supplies the listener in separate mode. Choose a supported aspect ratio and output setting.
Watch the full clip for role consistency, faces and audio. Add your own caption and an AI label when sharing.
Avoid a group shot where one person is tiny, hidden or turned away.
A clean source photo usually preserves more detail than a compressed screenshot.
One expression or a simple camera instruction is clearer than several unrelated actions.
MiniMax H3 starts at 15 seconds and 768P for this scene. Change duration or quality to update the general-credit quote, and review it before submitting.
MiniMax H3 · 480P / 768P / 1080P: 4–30 whole seconds · 2K: 4–15 seconds.
| Resolution | Duration | credits |
|---|---|---|
| 480P | 15s | 16 |
| 768P | 15s | 24 |
| 1080P | 15s | 38 |
| 2K | 15s | 40 |
The quote updates with duration and quality. For 16–30 seconds, the whole video fee doubles, including any video-reference reserve. Up to five photos and audio add no extra fee. Standard queue is used.
Trial and video-pack units cover only five-second outputs at their matching quality, without a video reference. Other settings use general credits.
Yes. Choose Together and upload one image containing both adults. The prompt asks the model to preserve both faces and use their left-to-right order.
In separate mode, the first reference is the listener. In Together mode, it is the person on the left of your photo. Check the final video because AI may still confuse positions.
Yes, when the selected model offers 16:9. Choose the aspect ratio in the editor; the reference example does not lock your output frame.
No exact soundtrack is promised. The prompt requests an original exchange. Use platform-licensed or otherwise permitted audio when finishing your post.
The editor shows the HiHD credit quote for your selected model and output settings before you generate. The Pricing page lists the available plans and credit packs.
Choose the input arrangement, then cast the moment.
Start creating