MiniMax H3 Image for Generation and Editing
MiniMax H3 Image generates images from a prompt or edits one input image. Choose the aspect ratio, 1K to 4K resolution, up to 10 images and JPEG, PNG or WebP output.
## Overview
MiniMax H3 Image is MiniMax's image generation and editing model. Describe an image and it creates it, or give it one image and describe the change you want. It is built to follow detailed instructions and keep the parts of an image you ask it to keep.
## What you can build
- Edits such as background swaps, relighting, seasonal changes and object or material changes
- Concept art, scene studies and marketing visuals from a written brief
- Product shots and variations that keep the same subject
- Several versions of one prompt to compare directions
## Inputs
- A prompt describing the image to create, or the change to make and what must stay the same (up to 4,096 characters; longer text is cut).
- An optional input image to edit (JPEG, PNG or WebP). Without one, the model generates from the prompt.
- An aspect ratio: auto, 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 5:4, 4:5, 21:9 or 9:21. With an input image, auto keeps its ratio; without one, auto uses 16:9.
- A resolution: 1K, 2K or 4K (about 1024×1024, 2048×2048 or 4096×4096 pixels, shaped to the aspect ratio).
- The number of images to generate, from 1 to 10.
- The output format: JPEG, PNG or WebP.
## Outputs
- One to ten images in the chosen format, at the chosen aspect ratio and resolution.
- With an input image, each output is an edited version of it.
## Recommended settings
- Say clearly what must stay the same in an edit, for example "Keep the man and the red kayak the same."
- Keep the aspect ratio on auto for edits so the output is framed like your image.
- Use 1K or 2K while exploring, then 4K for the final image.
## Limitations
- One input image per run. For several input images or a custom width and height, use H3 Image Advanced.
- No mask-based editing: you can't limit a change to a painted area.
- No transparent backgrounds.
- The prompt may be rewritten into a more detailed description before generation, so small wording details can change.
- Small text, logos and fine details can drift; check the results.
## Safety & compliance
- MiniMax runs content safety checks and may block a request or an image.
- Only upload images you have the right to use, and get consent before editing photos of identifiable people.
Example prompts
Great starting points for h3 image.
API quick start
Run h3 image with a single API call.
{
"prompt": "Turn this into an autumn scene: the trees…",
"resolution": "1K",
"samples": 1,
"inputImage": "https://your-cdn.com/input.png"
}curl -X POST "https://api.wiro.ai/v1/Run/minimax/h3-image" \
-H "Content-Type: application/json" \
-H "x-api-key: YOUR_WIRO_API_KEY" \
--data-binary @- <<'JSON'
{
"prompt": "Turn this into an autumn scene: the trees…",
"resolution": "1K",
"samples": 1,
"inputImage": "https://your-cdn.com/input.png"
}
JSONOther models you might like
Explore similar image generation models.
