Build a short video from a written scene or a starting image. Choose H3 Max for 5–15 second clips at 480p, 768p, or 1080p with audio, and review your settings before continuing.
MiniMax H3 Max creates short videos from text or a single starting image. Use text when you want to invent the setting, subject, composition, and action together. Start with an image when a product, character, or designed scene already has the appearance you want. In either workflow, describe what happens during the shot, how the camera moves, and which sounds belong in the scene. A clear visual beat gives you something concrete to judge when the clip is ready.
H3 Max is a separate version of MiniMax H3, with its own output choices. Max offers 480p, 768p, and 1080p for clips lasting 5–15 seconds. Choose the version before setting the output size: the original H3’s upscaled 2K and 4K choices do not apply to Max. The three Max examples here use five-second, 768p requests. Compare the motion and framing with your own needs rather than assuming one example predicts every prompt.
For a product reveal, try a slow push toward the subject or a small orbit that keeps the object centered. Leave enough room around the subject for that movement. If the camera follows a person, specify the direction and pace instead of combining a zoom, orbit, and sudden cut. Inspect the edges and reflections as the viewpoint changes; a convincing opening frame does not guarantee that the product stays consistent throughout the shot.
A short action is easier to direct when its conclusion is clear. A skateboarder can land and roll forward; a potter can shape one rim while keeping both hands on the clay. Say what the subject does first and where the movement settles. Avoid asking several people to perform separate actions at once when one action would communicate the idea. Check hands, contact points, and object count around the most demanding moment.
H3 Max includes audio generation. Describe the sound you want alongside the visible action: wheels rolling over pavement, rain against a window, or quiet room ambience around a spoken line. Keep dialogue brief enough for the selected duration and identify who speaks. Listen to the complete result before using it. Requesting a sound or line does not guarantee exact pronunciation, timing, lip movement, or a clean final mix.
Choose the workflow around the material you already have and the result you need. These are shot ideas to try, rather than guarantees about a particular output.
Start with a clean product image and request a measured camera move or changing reflection. Preserve the silhouette and important packaging details in the brief. Review labels carefully; add final typography in an editor when exact lettering matters.
Build a single action with an immediate focal point and a clear finish. A turn, landing, or walk through the frame can serve as a short insert. Leave complex choreography and multiple scene changes for separate clips.
Use a character image for a glance, a controlled hand movement, or a brief spoken moment. Keep faces and hands visible in the source. Inspect the performance throughout, especially when a hand touches a prop or a person speaks.
Describe a location from text or animate an existing environment image. Give the scene a restrained camera path and specific ambient sound. Foreground objects, visible depth, and a simple route can make the intended movement easier to communicate.
Generated video can change small details between frames. Faces, fingers, labels, reflections, and overlapping objects deserve close inspection. Complex physical interactions and tightly synchronized dialogue may need several revisions. Higher resolution makes a clip larger; it does not by itself correct an action, restore an identity, or guarantee accurate text.
H3 Max offers 480p, 768p, and 1080p with audio. The original H3 has different resolution choices, including upscaled outputs. Check the version and settings shown in the generator before continuing. Review the full downloaded clip: a short preview cannot establish that every frame or part of the soundtrack matches your brief.
Start with a scene description or one image, choose the version and output settings, then check the estimate and generation availability in the workspace.
Choose your starting point, keep the action focused, and review the selected version and estimate before continuing.
For text to video, describe the subject, setting, lighting, and composition before the action. For image to video, choose one image with a readable subject and uncluttered surroundings. The image supplies the opening appearance, so focus the prompt on movement. Use an image you have permission to animate and check the crop before selecting it. This workflow accepts a starting image, not a separate ending frame.
Put the main action first, then add a camera direction and the sound that supports it. For example: “The cyclist rides past the shop, camera tracking alongside, with tire noise and distant street ambience.” Add only the details that matter to the result. If a face, object, or sign must stay consistent, name it clearly. Avoid several scene changes competing for a few seconds.
Select H3 Max, then choose 5–15 seconds and 480p, 768p, or 1080p. Text to video also lets you choose the frame shape for the intended placement. A short draft can help you assess an idea before committing to a longer clip. Review the estimate whenever you change duration or resolution; higher settings cost more. Preparing the draft itself does not use generation credits.
Watch at normal speed, then pause around fast motion, hand movements, and the final frame. Check identity, product shape, object count, readable text, and camera continuity. Listen separately for unwanted sounds, unclear speech, or timing problems. If the shot misses the brief, simplify the action or change one instruction at a time. Assemble several approved clips in an editor when your idea needs a longer sequence.
Plans for image and video
Annual plans are billed upfront. The same plan credits refresh every 30 days and expire at the end of that window; unused subscription credits do not roll over.
Explore the workspace and prepare a first request before choosing a paid plan.
For occasional image and video experiments with a predictable monthly ceiling.
Subscriptions renew automatically at the selected monthly or yearly rate until canceled. Cancel future renewals in your account.
For creators who iterate every week and need room for several directions.
Subscriptions renew automatically at the selected monthly or yearly rate until canceled. Cancel future renewals in your account.
For sustained, high-volume creation and access to the highest available output quality.
Subscriptions renew automatically at the selected monthly or yearly rate until canceled. Cancel future renewals in your account.
Prices are in USD. Subscriptions renew automatically; credit packs are one-time purchases.
Compare plansYes. Choose text to video and describe the scene, action, camera, and sound. Image to video uses one starting image instead. Choose that workflow when you already have a composition or subject appearance you want to animate. A separate ending image is not an option in this generator.
Audio is included with Max. You can describe ambience, sound effects, or a short line of dialogue in the prompt. Listen to the completed clip and review speech clarity, volume and synchronization before using it. A visually convincing result still needs a separate sound check.
No. The product orbit, skateboard landing, and pottery scene were made with the original MiniMax H3 at 768p. They illustrate three motion briefs and are labeled accordingly. They should not be used to compare Max quality or predict results at other settings.
Select the version, duration, and resolution in the generator to see the estimate for your draft. Review it again after changing settings. Draft preparation uses no generation credits. If the action button shows Preview only, you can prepare the draft but cannot start generation from that preview.
Start with a scene description or one image, choose the version and output settings, then check the estimate and generation availability in the workspace.
