# MiniMax H3 Max Image to Video - Core.Today AI API

> MiniMax H3 Max image-to-video via Fal.AI. fal's post-trained variant of MiniMax Hailuo-03 (H3), tuned for stronger prompt adherence and better aesthetics — animates a first-frame image (optional last frame for keyframe control) into 5-15 second clips with native audio at 480P/768P native or 1080P refined output.

- **Provider**: MiniMax
- **Model ID**: minimax/h3-max/image-to-video
- **Category**: Video Generation
- **Credits**: 930 per 5s 768P video — billed per second (480P 116/s, 768P 186/s, 1080P 372/s)
- **Speed**: Medium
- **Quality**: Ultra

## Features
- Animates a first-frame image with stronger prompt adherence
- Optional last frame for first-to-last keyframe control
- Native audio generated alongside the video
- 480P/768P native generation or 1080P refinement, billed per second

## Use Cases
- Bringing product shots and key visuals to life
- Transitions between two designed frames
- Animated social posts from existing artwork
- Previs clips from concept art

## API Endpoint

Base URL: https://api.core.today/v1
Create Prediction: POST /predictions
Get Status: GET /predictions/{job_id}

## Authentication
Header: X-API-Key: YOUR_API_KEY

## Input Parameters

### Required
- **prompt**: string - Text prompt for video generation (up to 50,000 characters).
- **image_url**: string - URL of the image to use as the first frame. The output canvas follows this image.

### Optional
- **end_image_url**: string - Optional URL of the image to use as the last frame, for first-to-last keyframe generation.
- **duration**: integer (default: 5) - The duration of the video in seconds (5-15), billed per second.
- **resolution**: string (default: 768P) - 480P and 768P are native generation modes; 1080P is latent refinement from a native 768P source.
  Options: 480P, 768P, 1080P
- **prompt_expansion_mode**: string (default: balanced) - How much effort to spend rewriting the prompt before generation. 'disabled' skips expansion, 'balanced' returns in about a second, 'quality' spends up to ~30s on a richer prompt.
  Options: disabled, balanced, quality
- **enable_safety_checker**: boolean (default: true) - If set to true, the safety checker will be enabled.
- **seed**: integer - Random seed. A random seed is selected when omitted.

## Examples

### Landscape pull-back
The camera pulls back from a still landscape image at the default 768P.
```json
{
  "model": "minimax/h3-max/image-to-video",
  "input": {
    "prompt": "The camera slowly pulls back from the scene, revealing the full landscape as clouds drift overhead and light shifts across the terrain.",
    "image_url": "https://storage.googleapis.com/falserverless/example_inputs/hailuo23/pro_i2v_in.jpg",
    "duration": 5,
    "resolution": "768P"
  }
}
```

### Quick 480P draft
A cheaper 8-second 480P draft to check motion before the final render.
```json
{
  "model": "minimax/h3-max/image-to-video",
  "input": {
    "prompt": "Clouds race across the sky as sunlight sweeps over the hills, time-lapse feel.",
    "image_url": "https://storage.googleapis.com/falserverless/example_inputs/hailuo23/pro_i2v_in.jpg",
    "duration": 8,
    "resolution": "480P"
  }
}
```

## Response Format
```json
{
  "job_id": "abc123",
  "status": "pending | processing | completed | failed",
  "result": "URL or data (when completed)"
}
```

## Usage Flow
1. POST /predictions with model and input -> receive job_id
2. GET /predictions/{job_id} -> poll until status is completed or failed
3. Result contains output URL(s)

## Tips
- The output canvas follows the first-frame image — crop it to the aspect ratio you want.
- Add end_image_url to control where the motion lands.
- Iterate at 480P (116/s), then render the final at 768P or 1080P.
- Billing is per second of duration (5-15s).

## Documentation
https://fal.ai/models/minimax/h3-max/image-to-video