inference.minimax.h3-image.base.txt2img.v0

The inference.minimax.h3-image.base.txt2img.v0 job generates an image from a text prompt with MiniMax H3 Image. Output is 2048x2048 by default and can be any size up to 20 megapixels.

{
    "type": "inference.minimax.h3-image.base.txt2img.v0",
    "config": {
        "prompt": "A rustic farm-to-table feast on a weathered wooden table, overhead food photograph, warm side light"
    }
}

Prompting

The prompt is used exactly as written (up to 4096 tokens). The model responds best to long, complete English descriptions of the whole image, so describe the subject, setting, composition, lighting and style in full rather than giving a short request. Text in quotes renders well.

Size

Set width and height together (default 2048x2048). Any integer size works with each side 468 to 7745 px, an area of 655,360 to 20,000,000 pixels and an aspect ratio between 1:3 and 3:1. The model gives its best results at 4 to 20 megapixels; sizes around 16 MP, such as 4096x4096 or 5472x3072, have the most detail.

{
    "type": "inference.minimax.h3-image.base.txt2img.v0",
    "config": {
        "prompt": "A rustic farm-to-table feast on a weathered wooden table, overhead food photograph, warm side light",
        "width": 5472,
        "height": 3072,
        "seed": 42
    }
}

The same seed reproduces the same image. Typical generation time is about 1-2 s up to 4 MP and about 6 s at 16 MP.

Schema

{
  "type": "object",
  "required": [
    "type",
    "config"
  ],
  "additionalProperties": false,
  "properties": {
    "type": {
      "type": "string",
      "enum": [
        "inference.minimax.h3-image.base.txt2img.v0"
      ]
    },
    "config": {
      "type": "object",
      "required": [
        "prompt"
      ],
      "additionalProperties": false,
      "properties": {
        "prompt": {
          "type": "string",
          "minLength": 1,
          "maxLength": 32000,
          "examples": [
            "A rustic farm-to-table feast on a weathered wooden table, overhead food photograph, warm side light"
          ],
          "description": "The prompt to generate an image from, used as written (up to 4096 tokens). Long, complete descriptions of the whole image work best."
        },
        "width": {
          "type": "integer",
          "minimum": 468,
          "maximum": 7745,
          "default": 2048,
          "description": "Output image width in pixels. Give width and height together: the area must be 655,360 to 20,000,000 pixels with an aspect ratio between 1:3 and 3:1. Defaults to 2048x2048 when both are omitted."
        },
        "height": {
          "type": "integer",
          "minimum": 468,
          "maximum": 7745,
          "default": 2048,
          "description": "Output image height in pixels. Give width and height together: the area must be 655,360 to 20,000,000 pixels with an aspect ratio between 1:3 and 3:1. Defaults to 2048x2048 when both are omitted."
        },
        "seed": {
          "type": "integer",
          "minimum": 0,
          "maximum": 2147483647,
          "description": "Random seed for reproducible results."
        }
      },
      "dependentRequired": {
        "width": [
          "height"
        ],
        "height": [
          "width"
        ]
      }
    }
  }
}