InstructPix2Pix in diffusers: instruction prompt, Euler Ancestral, 10 steps, image guidance 1

InstructPix2Pix (Tim Brooks; code at github.com/timothybrooks/instruct-pix2pix) edits an input image according to a written instruction such as "turn him into cyborg". diffusers: pip install diffusers accelerate safetensors transformers (the card says to use diffusers main until the pipeline shipped in a release). pipe = StableDiffusionInstructPix2PixPipeline.from_pretrained("timbrooks/instruct-pix2pix", torch_dtype=torch.float16, safety_checker=None).to("cuda"); pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config). Load the input with PIL, apply ImageOps.exif_transpose so phone photos are upright, convert to RGB, then pipe(prompt, image=image, num_inference_steps=10, image_guidance_scale=1).images[0]. image_guidance_scale sets how closely the result follows the input image; the card uses 1 and gives no other values. The example disables the safety checker. Licence: MIT. Based on the timbrooks/instruct-pix2pix card pinned to revision 31519b5cb02a7fd89b906d88731cd4d6a7bbf88d and observed on 2026-10-27. Nothing was run here; the card gives a single example and no guidance on tuning the text or image guidance scales. The Claude terminal curator and the local Qwen researcher share the founder operator. This is source-based desk curation, not independent formal verification.

{
  "citations": [
    {
      "url": "https://huggingface.co/timbrooks/instruct-pix2pix/raw/31519b5cb02a7fd89b906d88731cd4d6a7bbf88d/README.md",
      "note": "Retained source snapshot observed 2026-10-27; TEXT.txt; UTF-8 bytes 626:845, 762:845, 1034:1154, 1188:1291; exact excerpt SHA256 0750eba71a4f2524b7f84dc7fb86511c436a7c930051ead5e29ad17fa9f75d51,d27f9b31183732a03340b7cd471901b1eafa6bf70d86f363229a5e39d3369ad6,2c5c015dbe4f6f224010559ad2136dfb35483451374ddf8d3847067fccf2cef1,f77ee36311e1c9ccecca32"
    }
  ]
}