MiniMax H3 Image to Video: Bring a Still Frame to Life
Last verified: October 29, 2026
MiniMax H3 image-to-video generation works best when you already have a frame you're happy with, like a packshot, a key visual or a character still, and that opening image has to stay the same. MiniMax describes this mode as a way to control the starting frame and "bring a specific frame naturally to life." Your image sets the subject and composition, and your prompt directs what happens next.
The form works the same way. You upload the still that should open the clip and write what happens next in the prompt box. Enhance Prompt is there if you want to turn a rough idea into a fuller draft. Then you pick a clip length and one of two resolution tiers and generate. Both settings match what MiniMax lists for H3: the same range of clip lengths and the same two resolution tiers.
MiniMax builds H3 for advertising, branding, e-commerce, product design, UI/UX and gaming. For those teams, starting from an image means the model gets the approved asset itself, not a description of it. Your prompt can then go entirely on motion, timing and camera.
Explore more image to video
The H3 Form at a Glance
What you upload and set for each run, and what you get back.
Input
Image (JPG, JPEG, PNG or WEBP, up to 10 MB) + a text prompt of up to 7,000 characters
Video Duration
4 to 15 seconds
Resolution
768P or 2K
Output
1 video per run
How Should You Write the Prompt When the Image Sets the Scene?
Describe What Happens After the First Frame
MiniMax's prompt guide for this mode works forward from the picture. The description first sets out the style, subjects, composition and scene anchors, and only then the action. In the Prompt box, that means one or two sentences naming what is in the frame and where it sits. After that, write the action in playback order: what starts moving, how it builds and where it ends. Don't rewrite the look with new adjectives. Anchoring means naming what is already there, not redesigning it. If you start from a rough idea, Enhance Prompt can draft a longer version for you to trim back down to this shape.
Give Each Camera Move a Type, an Amplitude and a Speed
The same guide defines a complete camera instruction in three parts. The motion type says how the camera moves, the amplitude says how much the composition changes, and the speed sets the pace. It also separates moves that are easy to mix up. A zoom changes focal length while the camera stays still, a pan turns the lens in place, and a static shot keeps both still. Write lines like a slow, small push-in toward the product, and ask for a static shot when the framing itself must not change. Then match the move to the Video Duration you picked. A large, fast move packed into a short clip looks like a jolt rather than a camera move.
How Do You Animate a Photo With MiniMax H3?
Five steps from a still image to a downloaded clip.
Add Your First Frame in Image
Click "Image" (the slot reads "Click to upload the first frame") and add the still the clip should open on. Use the final, approved version rather than a draft, and fix any crop or retouching before you upload. Any flaw in the still becomes frame one of the video.
Describe the Motion in Prompt
The "Prompt" box is required and takes up to 7,000 characters. Briefly anchor what is in the frame, then write the action in the order it should happen, plus one camera move. If you only have a rough idea, the "Enhance Prompt" button under the box suggests fuller prompts. Your text stays as you wrote it unless you pick one of the suggestions, so you can compare them first.
Choose Video Duration and Resolution
"Video Duration" runs from 4 to 15 seconds in one-second steps and is set to 4 by default. "Resolution" offers 768P (the default) or 2K. Match the length to the amount of action: a glance or a slow push-in fits a short clip, while a chain of actions needs more seconds. Stay on 768P while you test motion and timing, then switch to 2K, the larger frame, for the take you plan to deliver.
Run It With Generate
Click "Generate" once the image, prompt and settings fit together. Every run is a new take, so when a result is close, change only one thing (the camera line, the action or the length) before you run again. That way you can tell which change made the difference.
Download the Result
Watch the whole clip before you download it. Check that the opening matches your still, that faces, hands and logos hold up while things move, and that the action ends where you wanted. If it passes, download it. If the ending misses, rework the last part of the prompt or the length and run it again.
Starting From an Image or From Text
Two ways into the same model, compared on where the clip starts and what your prompt has to do.
| Criterion | MiniMax H3 | Alternatives | Best For |
|---|---|---|---|
| Where the clip starts | Your uploaded still becomes the actual first frame at the zero-second mark, so the opening composition is fixed before the run starts. | In the text-to-video mode of the same model, the clip is generated from a written description alone, so the opening image is whatever H3 composes from your words. | Start from an image when an approved still has to open the shot, and from text when you're still exploring and have no visual asset yet. |
| What the prompt has to cover | The prompt builds forward from what the picture already shows: it briefly anchors the style, subjects and composition, then spends the rest on action and camera. | A text-to-video prompt has no picture to line up with and goes straight into describing the scene, so every visual detail has to come from your words. | Choose image-to-video if you'd rather write about motion than about the look, and text-to-video if you want the model to design the look as well as the action. |
Common Problems and How to Fix Them
Upload, framing, prompt and timing problems that come up when animating stills with H3.
My image won't upload or gets rejected.
My ultra-wide panorama or long vertical screenshot fails or comes out wrong.
My prompt is too long to fit in the box.
The action feels rushed, or the clip ends before it finishes.
Frequently Asked Questions
What image files can I upload, and how big can they be?
The Image slot accepts JPG, JPEG, PNG or WEBP files up to 10 MB, at least 300×300 pixels in size. That picture becomes the first frame of the clip, so upload the cleanest, highest-quality version you have within those limits.
Can I make a MiniMax H3 clip without a starting image?
Yes. Use the Text To Video page instead. In that mode H3 generates the clip from a text description alone, which suits ideas that don't have a visual asset yet. Come back here once you have a still that should open the shot.
What makes one generation cost more than another?
The credit cost of a run is worked out from the options you choose and shown above the Generate button once you're signed in. It changes with Video Duration and Resolution, so check the figure again after changing either one. Getting suggestions from Enhance Prompt doesn't use credits.
What frame rate do MiniMax H3 clips come out at?
MiniMax lists H3's output at 24 fps for both resolution tiers and every clip length. Set your editing timeline to 24 fps when you cut H3 clips together so they play without frame-rate conversion.
How many videos does a run give me?
Each run returns one video. Plan one run for each variation you want to compare, such as a different camera line or a different length, and save the higher resolution for the take you keep.