MiniMax H3 example
A golden retriever at the wheel
Taking you to the studio...

- Reference sheets
- Keeps characters on-model
- 2K
Settings
- Model
- MiniMax H3
- Aspect ratio
- 9:16
- Duration
- 10 s
- Resolution
- 2K
Prompt
Photorealistic cinematic still, vertical 9:16, natural true-to-life color with neutral slightly flat contrast. Shot from the small camera mounted on the passenger headrest, looking forward past the driver's seat. GOLDEN RAMSEY from the dog sheet sits in the driver's seat seen from his right side in three-quarter view: red harness clipped in, glasses on, navy bandana, both front paws on the black yoke, eyes wide. Through the windshield and glass canopy: the Mojave desert road and Joshua trees just a few meters below from the LEFT panel of the sky sheet, the car just lifting off, faint white vapor curling past the bottom of the side glass. Morning daylight. His head sits in the upper two-thirds of the frame. Exactly one dog. No text, no logos, the touchscreen blank, no human hands. The image is the first frame. 10 seconds, vertical 9:16. The camera is fixed to the passenger headrest and moves only with the car. Photorealistic, a real golden retriever with natural dog anatomy and real paws. Natural true-to-life color, neutral slightly flat contrast, no color filter. Keep his head in the upper two-thirds of the frame. 0-2s: The car lifts straight up. A soft vibration shakes the cabin. Ramsey stiffens, paws gripping the yoke, eyes wide. 2-6s: The car climbs fast. The force pulls his cheeks, jowls and ears downward and his eyes bulge. Through the glass the desert drops away below, the road becoming a thin line. In a strained posh British voice: "Smooth… very smooth…" 6-9s: The climb gets steeper. His ears flap upward, his lips flutter, and his voice cracks into panic: "oh, that's TOO smooth—" 9-10s: He clamps his mouth shut, ears still flapping, staring straight ahead. The same dog, glasses, bandana and harness throughout; the harness stays clipped. White vapor visible outside the glass. No text, no logos, no people, no human hands. Audio: deep rising hiss of cold-gas thrusters, cabin rattle, rushing wind, his voice, a nervous whimper at the end. No music.
About MiniMax H3
Natural movement and sound, without the wait.
MiniMax H3 (Hailuo 03) ranks among the top text-to-video models on Artificial Analysis. It renders 6, 10 or 15 seconds at 768p or 2K, always with a stereo soundtrack generated alongside the picture — describe the dialogue, music or effects in your prompt. Guide it with a first and last frame, or with up to 5 reference images plus reference video and audio.
