MiniMax H3 source prompts: choose the right input mode
628 source examples; not independently reproduced and no identical-result promise.
How to choose an example
This collection contains the main H3 model, not Fast or Max. Its current Studio settings are 4–15 seconds and 768P/2K.
Choose text, frame or reference from your actual inputs; frame roles and reference roles belong to separate tasks.
Check each model, input role and known setting. Attribution is not a media reuse grant; supply your own inputs.
- Hand-Painted Blizzard Survival
The creator publishes the full prompt for a frame-by-frame oil-on-paper animation with a 12 fps stop-motion cadence, cold moonlight, and a violent blizzard, depicting a mother and
MiniMax H3 · text-to-video · 0 recorded input roles
- Cloud-Head Subway Transformation
A locked-off, symmetrical subway shot glitches the passengers' heads between human faces and distinct weather clouds, then converges on the reference composition at eight seconds.
MiniMax H3 · reference-to-video · 0 recorded input roles
- Handheld Space-Cruise POV at a Black Hole
A tourist records a handheld view from a space-cruise cabin as a giant black hole fills the window. They switch off the light to improve the view and speak a short line. The post p
MiniMax H3 · text-to-video · 0 recorded input roles
- Twenty-Second H3 Test
Runs a twenty-second H3 test with the same underlying idea adapted to the H3 guide, providing a traceable comparison with an earlier LTX 2.3 version.
MiniMax H3 · reference-to-video · 1 recorded input roles
- Gothic Dracula Horror Scene
Uses MiniMax H3 inside PixVerse with a short text instruction to create a Dracula scene focused on fog, bats, supernatural reflections, and dramatic lighting.
MiniMax H3 · text-to-video · 0 recorded input roles
- Neon Mountain Hover-Bike Race
Generates a fifteen-second two-rider hover-bike race from a supplied first frame, with text constraints that lock the two riders and two bikes while maintaining high-speed continui
MiniMax H3 · reference-to-video · 0 recorded input roles
- Widescreen Character Introduction
A 15-second widescreen character-introduction clip explicitly tagged MiniMax H3 and Hailuo AI. The creator published the complete prompt in a reply, preserved here verbatim.
MiniMax H3 · reference-to-video · 3 recorded input roles
- A Cybernetic Operative Appears on a Stat Panel
A figure emerges from a dark silhouette into a black-suited cybernetic operative centered on a futuristic stat panel surrounded by attributes, equipment, and body diagrams. Output:
MiniMax H3 · text-to-video · 0 recorded input roles
- Two Glaciologists Enter a Blue-Ice Cave
The sequence opens above a small entrance in a glacier before two helmeted researchers carrying equipment move into a passage enclosed by translucent blue ice walls. Output: 15s ·
MiniMax H3 · text-to-video · 0 recorded input roles
- A Lacquer-Thread Vase from Clay Rolling to Display
An artisan rolls material into fine strands, applies and carves gold ornament across a red vase, and finally places the completed piece on a display stand. Output: 15s · 2560×1440.
MiniMax H3 · reference-to-video · 3 recorded input roles
- A Morning Begins from Bed to a Seaside Terrace
A long-haired woman first sits on a bed in a bright bedroom and adjusts her hair, followed by a seaside terrace with wooden furniture and a closing sunset view reading “Good vibes,
MiniMax H3 · text-to-video · 0 recorded input roles
- Legolas and Gimli Sit Across a Living Room
Gimli sits beside a green armchair while Legolas reclines on an orange sofa holding a stemmed glass, with the frames alternating between a shared view and individual close-ups. Out
MiniMax H3 · text-to-video · 0 recorded input roles