01Tidal light
H3 MaxSea-glass water. A slow coastal glide.
16:9 · 5s · 480P · Balanced
Turn a written direction or starting frame into a complete shot. Use H3 Max for the final pass, or H3 Max Turbo to explore motion at lower cost.
Made with MiniMax H3
From natural light to handmade worlds. Original 5-second text-to-video studies, created for MiniMaxVideo. Press play to explore.
Sea-glass water. A slow coastal glide.
Soft reflections, a quiet product portrait.
A little light, a little movement.
Folded petals find their own rhythm.
Tiny bubbles. Fresh afternoon light.
A handmade world, just above the clouds.
MiniMax H3 is most useful when a prompt does more than name a subject. A strong direction explains what moves, how the camera responds, what changes over time, and where the shot should end. MiniMaxVideo keeps those decisions in one workspace, from idea to private result.
The interface separates two jobs that are often mixed together. Text-to-video gives you control over the frame before any image exists. Image-to-video starts with a visual truth: a product photograph, portrait, illustration, environment, or storyboard panel. In image mode, the source frame determines the aspect ratio, while an optional final frame can define the destination of the movement.
Both MiniMax H3 tiers use the same controls, private history, and generation workflow. The choice is about how you want to spend time and credits during the creative process.
01 · Quality-led
From 30 credits / 5s
Use H3 Max when the clip is moving toward publication. It is the stronger choice for a final product shot, cinematic concept, polished fashion movement, or any scene where visual finish and prompt adherence matter more than the cheapest iteration.
02 · Faster iteration
From 15 credits / 5s
Use H3 Max Turbo when the important question is whether the idea works. Test a camera move, compare prompt wording, explore a vertical crop, or validate the relationship between a first and last frame before spending more on the final pass.
MiniMaxVideo follows the MiniMax H3 API instead of presenting a generic wall of switches. Each control appears only when the selected mode can actually use it.
Choose from 21:9, 16:9, 4:3, 1:1, 3:4, or 9:16, then write the action in temporal order. Define the opening image, the primary movement, the camera response, and the final state. This gives MiniMax H3 a coherent path through the shot instead of several unrelated visual requests.
Upload a first frame to preserve the subject and composition. Add an optional last frame when the ending must land on a particular product position, pose, reveal, or transition. The generated video inherits the first image ratio, so there is no misleading aspect-ratio control in image mode.
A five-second 480P Turbo clip is useful for inexpensive motion tests. A longer 768P H3 Max generation is better suited to a final candidate. The live estimate changes with model, resolution, and duration before the task is created, so the cost is visible before credits are reserved.
Follow your progress in the workspace after starting. Completed videos are saved to your history. If generation fails, your reserved credits are returned automatically.
Begin with the subject and setting, then describe motion as a sequence. Add camera language such as a slow push-in, locked-off wide shot, handheld follow, or low orbit. Finish with the intended final frame. A concise shot direction is easier for MiniMax H3 to follow than a dense list of visual adjectives.
Select H3 Max or H3 Max Turbo, set a duration from 5 to 15 seconds, and choose 480P or 768P. Text mode adds an explicit aspect ratio. Image mode replaces that control with a required first frame and optional final frame, preserving the relationship between interface and model behavior.
After sign-in, MiniMaxVideo uploads references, checks the prompt, reserves the displayed credits, and opens the dedicated video workspace. Queueing, processing, archiving, playback, and download stay attached to the same task, so a browser refresh does not erase the work in progress.
Turn a clean product frame into a reveal, rotation, material study, or atmospheric advertising moment. A final frame can help the object settle into a precise end composition for an edit or landing-page sequence.
Explore camera paths, environmental movement, character blocking, and lighting transitions before a shoot or a larger production. Turbo makes comparison affordable; H3 Max carries the strongest direction into a final concept clip.
Animate a portrait with restrained expression and camera movement, add believable motion to garments, or give an illustration depth without abandoning its original composition. Starting from a frame helps keep the visual identity grounded.
Generate vertical openings, looping visual ideas, and compact story beats for social channels. Select 9:16 in text mode or upload a vertical first frame in image mode, then keep the action readable inside a short duration.
Identify the person, product, creature, vehicle, or environment that must remain readable. Include a few durable traits—material, clothing, silhouette, age, or color—but avoid adding a new style reference in every sentence.
Explain what begins the shot, what the subject does, and how the action resolves. When several subjects move, state their relationship. Conflicting instructions such as running, standing still, and turning in different directions create unnecessary ambiguity.
Use concrete language: track beside the subject, crane upward, orbit slowly to the left, hold a locked wide frame, or push toward a detail. Camera direction connects separate visual events and gives the clip an intentional point of view.
A light change, passing shadow, burst of rain, or shift in atmosphere can mark time inside the shot. MiniMax H3 may infer sound-related visual behavior from the prompt, but this workspace currently returns video only and does not expose a separate audio control.
Your credit estimate depends on the model, resolution, and duration. Review it before you begin, and use credits from your subscription or a credit pack.
The estimate shown in the generator is the amount reserved for that exact model, resolution, and duration. If generation fails, reserved credits are returned automatically.
MiniMax H3 creates video from text or images. Choose H3 Max or H3 Max Turbo in MiniMaxVideo, adjust the shot settings, and save and download your work.
H3 Max is the quality-led final tier. H3 Max Turbo uses half the credits at the same settings and suits prompt, movement, framing, and timing tests. Both support the same modes, resolutions, durations, and prompt expansion.
Yes. Text-to-video starts with a prompt and explicit ratio. Image-to-video requires a first frame and accepts an optional final frame. Its output ratio follows the first image to preserve the source composition.
Both tiers support whole durations from 5 through 15 seconds at 480P or 768P. Cost updates per second, so longer or higher-resolution tasks reserve more credits before entering the queue.
Balanced performs a quick rewrite. Quality allows more time for a richer expanded prompt before inference. Neither replaces a clear subject, action, camera path, and final state.
No. Sign in to MiniMaxVideo to generate with your account credits, then view and download your videos from your private history.
Follow your progress in the workspace after starting. Completed videos are saved to your history. If generation fails, your reserved credits are returned automatically.
History is private to the signed-in account. This release does not automatically publish personal generations to a gallery. You remain responsible for the rights to uploaded images and generated uses.
Your next shot can start here
Write the movement, choose the right H3 tier, and see the exact credit estimate before the task begins.