DreamShaper XL v2 Turbo: 6 steps, guidance 2, DPM++ multistep (pass the seed)

Lykon/dreamshaper-xl-v2-turbo is a turbo (few-step) fine-tune of stabilityai/stable-diffusion-xl-base-1.0 in the DreamShaper family; licence OpenRAIL++. Diffusers recipe from the card: pip install diffusers transformers accelerate; AutoPipelineForText2Image.from_pretrained("lykon/dreamshaper-xl-v2-turbo", torch_dtype=torch.float16, variant="fp16"); set pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config); move to CUDA; generate with num_inference_steps=6 and guidance_scale=2. That is far fewer steps and much lower guidance than standard SDXL. Fix to the example: it creates generator = torch.manual_seed(0) but calls pipe(prompt, num_inference_steps=6, guidance_scale=2) without it, so results are not seeded; pass generator=generator to reproduce an image. The SD1.5 DreamShaper 7 and 8 records cover the non-turbo models. Based on the Lykon/dreamshaper-xl-v2-turbo card pinned to revision f70ad5bce436c640d23fc02cd0a6ccc787280b6e and observed on 2026-10-27. Nothing was run here; the card gives no resolution or negative-prompt guidance. 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/Lykon/dreamshaper-xl-v2-turbo/raw/f70ad5bce436c640d23fc02cd0a6ccc787280b6e/README.md",
      "note": "Retained source snapshot observed 2026-10-27; TEXT.txt; UTF-8 bytes 377:499, 1134:1236, 1399:1535, 1432:1535, 19:274; exact excerpt SHA256 5a605382931e3a0d60be56222d62bd915bbdb622f7c1c05dd2d66f7fb95e9541,f083bdcdf96b846a10c0f2f1f985a7e488e97eee9fea2d6ce9c1c8c0e1b4dc3d,a2ec85785de47875517d013ea3ea444485bb9646e87346877fdd221c642263b5,41bb59943e6e"
    }
  ]
}