finegrain-image-enhancer-endpoint

Finegrain Image Enhancer - Inference Endpoint

This repository contains the Inference Endpoint version of the Finegrain Image Enhancer.

Usage

Send a POST request to the endpoint with:

{
  "image": "base64_encoded_image_data",
  "prompt": "masterpiece, best quality, highres",
  "upscale_factor": 2.0,
  "seed": 42
}

Returns:

{
  "enhanced_image": "base64_encoded_enhanced_image",
  "original_size": [width, height],
  "enhanced_size": [width, height]
}

Parameters

• image (required): Base64 encoded input image
• prompt (optional): Enhancement prompt (default: "masterpiece, best quality, highres")
• negative_prompt (optional): Negative prompt (default: "worst quality, low quality, normal quality")
• seed (optional): Random seed (default: 42)
• upscale_factor (optional): Upscale factor 1-4 (default: 2.0)
• controlnet_scale (optional): ControlNet scale 0-1.5 (default: 0.6)
• controlnet_decay (optional): ControlNet decay 0.5-1 (default: 1.0)
• condition_scale (optional): Condition scale 2-20 (default: 6)
• tile_width (optional): Tile width 64-200 (default: 112)
• tile_height (optional): Tile height 64-200 (default: 144)
• denoise_strength (optional): Denoise strength 0-1 (default: 0.35)
• num_inference_steps (optional): Number of steps 1-30 (default: 18)
• solver (optional): Solver type "DDIM" or "DPMSolver" (default: "DDIM")