Start with a clear reference
Use a bright, sharp photo with one face and the upper body in view. For Both of us, select two images in order: table-sitter first, arriving guest second.
First Time in a Fancy Cafe
These reference-site clips illustrate the joke, not this model’s performance. Their original lengths are unchanged. Check the seating, faces and reaction in your new five-second version.
Original → Reference preview
Original → Reference preview
Original → Reference previewFrom photo to idea
Keep the camera still and let the reaction carry the scene: a confident perch, a glance at the other guest, then a discreet slide onto the bench. Choose one role for a single photo, or use two ordered image references to cast both. Five seconds calls for one readable beat, not a shot-for-shot recreation.
Use a bright, sharp photo with one face and the upper body in view. For Both of us, select two images in order: table-sitter first, arriving guest second.
MiniMax-H3 creates one five-second video per video allowance. The trial and all packs allow 768P; the 10- and 30-video packs also offer 2K, subject to provider acceptance, which is not yet tested. Results are saved to your library for preview and download.
These reference-site clips illustrate the joke, not this model’s performance. Their original lengths are unchanged. Check the seating, faces and reaction in your new five-second version.
One-time payment
One allowance makes one 5-second video, shared across all scene video tools.
1 × 5s · 768P
Try one video3 × 5s · 768P
10 × 5s · 2K
30 × 5s · 2K
2K is available with the 10- and 30-video packs. Provider acceptance and final output quality have not yet been tested.
Choose Both of us to switch to image references. Add the table-sitter first and the arriving guest second. Use separate, clear photos with permission from each adult. This directs the model; it does not guarantee an exact likeness or frame-by-frame timing.
Each signed-in account gets one free five-second 768P video across all scene video tools. The $4.90 pack includes 3 videos; $9.90 includes 10 and $19.90 includes 30. The latter two also allow 2K, which has not yet been tested. All are one-time purchases.
Use a single fixed shot and ask for the moment of realization. Avoid packing several camera cuts into the prompt. A confirmed failed or cancelled task returns its allowance once. A completed video with an imperfect performance still counts as a generation.
Each generation uses one video allowance. 768P is available for the trial and all video packs; 2K requires the 10- or 30-video pack. Confirmed failures return the allowance. Pricing
No results yet. Start a task in a creative tool; your saved results will appear here.