Build a shot from a written scene
Describe a location, subject, and one clear action. Use MiniMax H3 text-to-video to try an opening shot before committing to a photo, storyboard, or shoot.
- Text to Video
- No image needed
Text to video · First & last frames
MiniMax H3 AI Video Generator
Turn a written scene or a still image into a moving shot with MiniMax H3. Add a second image to shape the ending, choose 4–15 seconds, and generate in 768P or 2K on Lanta AI.
- 4–15 seconds
- 768P / 2K
- Up to 2 images
What you can do with MiniMax H3
Start with the material you have: a written scene, a single image, or a planned opening and ending.
Put an existing image in motion
Animate a product photo, character illustration, or room image. Add a small gesture or camera move to develop the still into a social clip or scene preview.
- Image to Video
- 1 start frame
Give the ending a destination
Add an end frame when a reveal must land on a particular composition. Guide MiniMax H3 from a wide product view toward a closer finish, leaving space for copy added later.
- Start + end frames
- 2 images
Let the action set the length
Choose any whole second from 4 to 15. A brief camera move needs less room than an entrance, turn, and settle; fit the shot to the action before spending more credits.
- 4–15 seconds
- Whole-second control
Frame for the place it will play
Make a vertical opener, a square product clip, or a wide scene. MiniMax H3 offers six ratios in text mode; image mode follows the source image, so crop it first.
- 6 text-to-video ratios
- Image-led framing
Test a draft, then choose 2K
Choose 768P for motion studies and 2K for a larger delivery format. Inspect labels, faces, and fine edges at the final viewing size before adding the shot to your edit.
- 768P
- 2K
See MiniMax H3 in motion
A sample shows how a scene moves. Before planning your own MiniMax H3 video, pay attention to the subject, camera, and final beat.
Three things to watch
- The subject
- Follow one face or object through the shot. Look for shifts in shape, edges, and small details.
- The camera
- Watch whether movement has a clear direction. Use one camera instruction in your first test.
- The finish
- Pause the last frame. Check that the subject has settled and the composition leaves room for your next cut.
MiniMax H3 pricing on Lanta AI
Pick a duration and resolution to see the cost before generating. The same MiniMax H3 rate applies to text-to-video and image-to-video.
- 768P
- 12credits per second
For testing movement and composition.
- 2K
- 20credits per second
For a larger output after refining the brief.
An 8-second draft costs 96 credits at 768P; 2K costs 160 credits. Shorten a test while refining the motion, then allow more time for the final action.
View Lanta AI credit plans| Clip length | 768P | 2K |
|---|---|---|
| 4 seconds | 48 credits | 80 credits |
| 8 seconds | 96 credits | 160 credits |
| 15 seconds | 180 credits | 300 credits |
A new resolution starts a new generation, so compare the new clip before replacing your draft.
How to create a MiniMax H3 video
Start in the generator above. For each MiniMax H3 draft, change one input and inspect the result before refining again.
Choose your starting point
Text to Video / Image to Video
MiniMax H3 is preselected in both modes. Start with text for a new scene, or choose Image to Video when a photo or two planned frames should guide the shot.
Describe what should move
Prompt / Start Frame / End Frame
The first image starts the shot and the second defines its ending. Describe the movement between them; keep the subject, lighting, and setting compatible across both images.
Set the format and budget
Duration / Quality / Video Ratio
Choose 4–15 seconds and 768P or 2K. Text mode offers 1:1, 3:4, 4:3, 9:16, 16:9, and 21:9. In image mode, prepare the framing in your source images. Check the credit total.
Generate, review, refine
Generate / Generation History
Generate your MiniMax H3 clip and follow its status in history. Inspect the subject and last frame. If the action feels rushed, simplify it or give it more time; change one input before regenerating.
Practice brief · Image to Video · 8 seconds · 768P
MiniMax H3 prompting guide: connect the frames
A product reveal is easier to plan when both compositions agree. Keep the bottle, light, and background consistent; let the prompt supply the camera movement.
Storyboard study: a wide opening and a closer finish.
- 0–2 s
Establish
- 2–6 s
Move closer
- 6–8 s
Settle
If the bottle changes shape, reduce the camera travel and bring the two compositions closer together. Add exact brand lettering in your editor after generation.
Prompt to copy
Use the first image as the opening and the second as the ending of an 8-second continuous shot. An amber glass bottle stands on a dark stone pedestal. Hold the wide view briefly, then move the camera slowly closer and recenter the bottle. Keep its shape, cap and label consistent. Let a soft highlight travel across the glass. Ease to a stop in the final composition, leaving the bottle still for the last moment.
Read the official MiniMax H3 prompting guide
Compare other video workflows
Grok Imagine
Explore a workflow with multiple reference images when one start and one end frame are not enough.
Wan 2.6
Compare audio and multi-shot controls for a sequence that needs more than a single camera move.
Seedance 2.0
Explore multimodal references when video or audio inputs need to guide the result.
FAQs of MiniMax H3 AI Video Generator on Lanta AI
What is MiniMax H3?
MiniMax H3 is a video model from MiniMax. On Lanta AI, you can create 4–15-second videos from text, a starting image, or a pair of start and end frames, with 768P and 2K output options.
How many images can I upload to MiniMax H3?
Upload one starting image and, optionally, one ending image. Each image must be no larger than 10 MB, with its longer side no more than 2.5 times its shorter side. Use clear, similarly framed images of the same subject for a continuous shot.
Can I edit a video or upload audio references here?
This MiniMax H3 generator accepts text and up to two images. It does not offer video editing or video and audio reference uploads. For a workflow that takes those reference types, explore Seedance 2.0 in the model links above.
Does MiniMax H3 include audio controls on Lanta?
MiniMax describes native audio as a model capability, but this page has no H3 audio controls or audio-reference input. Check the actual result before relying on a soundtrack or synchronized dialogue, and plan any required audio separately.
Is MiniMax H3 free to use?
Sign in with enough Lanta AI credits to generate. Your total depends on duration and resolution and appears before submission. Check your account balance and current plan for included credits.
Is this MiniMax H3 Max or Turbo?
The selected model is MiniMax H3. Max and Turbo are not separate options here. Plan your clip using the duration, resolution, and credit total shown in this generator.
How long does MiniMax H3 take to generate?
Waiting time varies with the queue and request. Follow the job in your generation history instead of submitting the same request again while it is processing. The selected 4–15 seconds describes the output clip, not the wait.
What if MiniMax H3 changes the subject between frames?
Start by reducing the difference between the two images: match the subject, lighting, camera height, and crop. Ask for one simple movement and inspect an early draft. A higher resolution adds output detail, but it does not guarantee accurate faces, labels, or lettering.
Make your next shot with MiniMax H3
Start with a scene in words or an image you already have. Add an ending frame when the final composition matters, then give the movement a clear direction.
Create with MiniMax H3