Watch the outputs, then explore the prompt behind each scene.
Clay animation: landscape and portrait
The same prompt, two independent generations. Compare framing, motion and texture.
A wider view of the scene
The horizontal frame gives the whale and surrounding waves more room. Use it to judge how the scene reads across a wider canvas.
A vertical view of the action
The vertical frame concentrates attention on the whale's rise. Compare how the same prompt is interpreted in a taller composition.
Prompt
Claymation. A whale breaches from a stormy clay sea and rises in a slow-motion arc above the waves, body fully extended, water sheeting off its flanks. At the top of the arc it rolls onto one flank, hangs for a beat, then falls back toward the surface. Midair, the camera sweeps beneath its belly in a bold dynamic move, revealing the vast churning ocean far below and the full stretch of its clay body against the sky, then holds as it crashes down and white clay foam bursts outward across the frame. Visible fingerprints, tool marks, and material imperfections in the clay. Slightly stepped, handmade movement. The sea is sculpted clay with rolling modelled waves and white clay foam. Audio: a deep rush of water, wind, the dull slap of clay on clay, distant gulls. Composition: subject centered, action contained within the central third of the frame.
Choose what guides the video
Start with the amount of visual direction you already have. Each mode uses the same duration, resolution and credit tiers.
Describe a scene
Build a new scene from a written brief: useful for exploring a story opening, a visual concept or a stylized short before preparing image assets. Start with a subject and a clear action.
Animate an existing image
Use image-to-video to keep a starting composition in view. Upload your own image, then describe what should move and how the camera should behave. The output follows the starting image's proportions.
Guide the ending
Add an optional end frame to indicate where a transition should finish. Keep the subject and framing reasonably compatible. The frames guide generation; they do not guarantee exact timing or pixel-perfect preservation.
Settings and credits
Compare MiniMax H3 resolutions, durations and credit costs on FluxContext. The generator shows the total before you submit.
Credits per video, with the same tiers for text, one starting image, or a start/end-frame pair.
Resolution
5 s
10 s
15 s
480PNative generation
100 credits
200 credits
300 credits
768PNative generation
120 credits
240 credits
360 credits
2KUpscaled from 768P
250 credits
500 credits
750 credits
4KUpscaled from 768P
300 credits
600 credits
900 credits
A second frame does not add credits. Resolution and duration change the total; changing a text-to-video aspect ratio does not.
Inputs
Text, one starting image, or a starting image with an optional end frame.
Text-to-video ratios
21:9 · 16:9 · 4:3 · 1:1 · 3:4 · 9:16
Image-to-video ratio
Follows the starting image. There is no separate aspect-ratio setting.
Image uploads
PNG, JPG or WebP; up to 4 MB and 40 megapixels per image.
Output
MP4 video. Actual file dimensions may differ slightly from the selected aspect-ratio preset.
Audio controls
Generated videos can contain audio. This workspace does not offer a sound toggle or audio upload.
How to prompt MiniMax H3
Start with the mode you are using. These practical notes adapt MiniMax’s official guide to this workspace: text, a starting image, or start and end frames. Reference video, reference audio and end-frame-only inputs are not available here.
Text: establish the scene
Name the subject, setting and visible action. Fit the action sequence to your selected 5, 10 or 15 s.
Starting image: describe what happens next
Begin from the supplied composition; describe subsequent motion without contradicting the image’s subject, clothing or surroundings.
Two frames: explain the transition
Describe the intermediate changes connecting the opening and ending. Prefer a continuous shot for a continuous transition.
Camera: specify movement and pace
Choose a camera move, its range and speed when relevant. Avoid stacking conflicting directions.
Sound: distinguish ambience and music
Describe environmental sounds separately from background music. Prompted sound is an intention, not an exact timing guarantee.
Iterate one choice at a time
For a first draft here, try 5 s at 768P. Review the action before spending more credits on a longer or larger output. Keep a note of the prompt and settings; change one element so you can judge what helped.
Two briefs to adapt
Writing examples, not tested outputs. Copy a brief, adapt it to your scene, and select the matching mode and duration before generating.
Text to video · one camera move
Try 5 s. Replace the object and setting while keeping one clear movement.
A red paper boat floats through a shallow puddle beside a stone curb after rain. The camera tracks slowly alongside it at water level as it passes a fallen leaf. Overcast daylight, visible paper folds, small ripples. Soft water sounds and distant traffic; no background music.
Start and end frames · a continuous change
Choose Image to video and 5 s. Supply two compatible photos of the same box: closed first, open last. Copying this brief does not add images.
Begin with the closed box in the starting image. In one uninterrupted shot, the lid slowly lifts to reveal the contents, then settles into the open position shown in the end image. Keep the camera fixed and preserve the box, table and lighting throughout. A soft hinge creak accompanies the movement; no background music.
Frequently Asked Questions about MiniMax H3
Can I use an image created in FluxContext as the start frame?
Yes. On any image in your creation history or editor, click Generate video to load it directly as your starting frame in MiniMax H3. You can also upload any clear JPG, PNG, or WebP from your computer.
Can I use both a start frame and an end frame?
Yes. In Image to Video, upload your start frame, then click + End frame to upload your target ending image. The model will calculate smooth visual transitions and camera motion between both keyframes.
Are 2K and 4K native resolutions?
MiniMax H3 generates natively at 480P or 768P for fast, natural physics. The 2K and 4K options enhance and upscale the output, providing sharp, high-pixel video exports suitable for high-resolution displays and post-production editing.
Does video generation include sound and audio?
Yes. MiniMax H3 natively generates matching environmental sound effects and ambient audio synchronized with your visual prompt and scene motion.
What happens if video generation fails or times out?
Your credits are automatically and immediately refunded. If a video task encounters a server timeout, provider error, or content safety review block, deducted credits return to your account balance instantly. You are never charged for incomplete generations.
Where can I view and download my completed videos?
All completed videos are saved in the generator's History tab and under 'My Creations' in your account menu. You can preview, play, and download standard MP4 video files anytime.