stable-diffusion-xl-lightning Beta

b

stable-diffusion-xl-lightning Beta

Text-to-Image • bytedance

@cf/bytedance/stable-diffusion-xl-lightning

SDXL-Lightning is a lightning-fast text-to-image generation model. It can generate high-quality 1024px images in a few steps.

Usage

Workers - TypeScript

export interface Env {
  AI: Ai;
}

export default {
  async fetch(request, env): Promise<Response> {

    const inputs = {
      prompt: "cyberpunk cat",
    };

    const stream = await env.AI.run(
      "@cf/bytedance/stable-diffusion-xl-lightning",
      inputs
    );

    return new Response(response, {
      headers: {
        "content-type": "image/jpg",
      },
    });
  },
} satisfies ExportedHandler<Env>;

curl

curl https://api.cloudflare.com/client/v4/accounts/$CLOUDFLARE_ACCOUNT_ID/ai/run/@cf/bytedance/stable-diffusion-xl-lightning  \
  -X POST  \
  -H "Authorization: Bearer $CLOUDFLARE_API_TOKEN"  \
  -d '{ "prompt": "cyberpunk cat" }'

Parameters

Input

• prompt string min 1

  A text description of the image you want to generate

• negative_prompt string

  Text describing elements to avoid in the generated image

• height integer min 256 max 2048

  The height of the generated image in pixels

• width integer min 256 max 2048

  The width of the generated image in pixels

• image array

  For use with img2img tasks. An array of integers that represent the image data constrained to 8-bit unsigned integer values

  • items number

    A value between 0 and 255

• image_b64 string

  For use with img2img tasks. A base64-encoded string of the input image

• mask array

  An array representing An array of integers that represent mask image data for inpainting constrained to 8-bit unsigned integer values

  • items number

    A value between 0 and 255

• num_steps integer default 20 max 20

  The number of diffusion steps; higher values can improve quality but take longer

• strength number default 1

  A value between 0 and 1 indicating how strongly to apply the transformation during img2img tasks; lower values make the output closer to the input image

• guidance number default 7.5

  Controls how closely the generated image should adhere to the prompt; higher values make the image more aligned with the prompt

• seed integer

  Random seed for reproducibility of the image generation

Output

API Schemas

The following schemas are based on JSON Schema

Input

{
    "type": "object",
    "properties": {
        "prompt": {
            "type": "string",
            "minLength": 1,
            "description": "A text description of the image you want to generate"
        },
        "negative_prompt": {
            "type": "string",
            "description": "Text describing elements to avoid in the generated image"
        },
        "height": {
            "type": "integer",
            "minimum": 256,
            "maximum": 2048,
            "description": "The height of the generated image in pixels"
        },
        "width": {
            "type": "integer",
            "minimum": 256,
            "maximum": 2048,
            "description": "The width of the generated image in pixels"
        },
        "image": {
            "type": "array",
            "description": "For use with img2img tasks. An array of integers that represent the image data constrained to 8-bit unsigned integer values",
            "items": {
                "type": "number",
                "description": "A value between 0 and 255"
            }
        },
        "image_b64": {
            "type": "string",
            "description": "For use with img2img tasks. A base64-encoded string of the input image"
        },
        "mask": {
            "type": "array",
            "description": "An array representing An array of integers that represent mask image data for inpainting constrained to 8-bit unsigned integer values",
            "items": {
                "type": "number",
                "description": "A value between 0 and 255"
            }
        },
        "num_steps": {
            "type": "integer",
            "default": 20,
            "maximum": 20,
            "description": "The number of diffusion steps; higher values can improve quality but take longer"
        },
        "strength": {
            "type": "number",
            "default": 1,
            "description": "A value between 0 and 1 indicating how strongly to apply the transformation during img2img tasks; lower values make the output closer to the input image"
        },
        "guidance": {
            "type": "number",
            "default": 7.5,
            "description": "Controls how closely the generated image should adhere to the prompt; higher values make the image more aligned with the prompt"
        },
        "seed": {
            "type": "integer",
            "description": "Random seed for reproducibility of the image generation"
        }
    },
    "required": [
        "prompt"
    ]
}

Output

{
    "type": "string",
    "contentType": "image/png",
    "format": "binary",
    "description": "The generated image in PNG format"
}