Generate a 2K video from prompt-led image, video, and audio references with the MiniMax H3 API on Muapi.
About this model
MiniMax H3 Reference to Video uses a prompt plus image or video references to guide subjects, motion, visual style, and scene continuity. You can also include reference audio when supplying a visual reference.
Cost analysis
Configuration schema
| Parameter | Type | Description | Default |
|---|---|---|---|
| Prompt | string | Describe the video scene and how the references should be used. | Use the references to create a cinematic product reveal with a slow camera orbit. |
| Reference Images | array | Reference image URLs. Provide at least one image or video reference. | undefined |
| Reference Videos | array | Reference video URLs. Provide at least one image or video reference. | undefined |
| Reference Audio | array | Optional reference audio URLs. Audio cannot be used alone. | undefined |
| Aspect Ratio | Enum (6 options) | - | 16:9 |
| Resolution | Enum (1 options) | - | 2k |
| Duration | Enum (11 options) | Output duration in seconds. | 5 |
Describe the video scene and how the references should be used.
Use the references to create a cinematic product reveal with a slow camera orbit.Reference image URLs. Provide at least one image or video reference.
undefinedReference video URLs. Provide at least one image or video reference.
undefinedOptional reference audio URLs. Audio cannot be used alone.
undefined-
16:9-
2kOutput duration in seconds.
5Developer documentation
Provide a prompt and at least one reference_images or reference_videos URL. Add optional reference_audios, then poll the returned request ID or provide a webhook URL.
Frequently asked
Use one or more image or video URLs, with optional audio URLs to guide the generation.
No. MiniMax H3 Reference to Video requires at least one image or video reference.
The first 5 reference images and all reference audio are free. Each reference image after the first 5 costs $0.055. Reference videos cost $0.1825 per second of their combined duration, up to 3 videos and 15 seconds total per request.