Juggernaut XL v9: DPM++ 2M Karras, 30-40 steps, CFG 3-7, no negatives to start

Juggernaut XL v9 is an SDXL photorealism fine-tune by KandooAI and RunDiffusion that integrates RunDiffusion Photo v2; the vendor credits it with better skin detail, lighting and contrast than v8 and says it runs on 8 GB of VRAM and works with existing SDXL ControlNets, IP-Adapters, AnimateDiff and LoRAs. Settings from the card: 832x1216 (portrait) or 1216x832 (landscape); DPM++ 2M Karras; 30 to 40 steps; CFG 3 to 7, lower for more realism; the VAE is baked in. Hires fix: 4xNMKD-Siax_200k upscaler, 15 steps, 0.3 denoise, 1.5x upscale. Start with no negative prompt and add specific ones only as needed, because heavy negatives tend to hurt this model. Steering keywords include Architecture/Wildlife/Car/Food/Interior/Landscape Photography, Hyperdetailed Photography, Cinematic Movie, Still Mid Shot Photo, Full Body Photo and Skin Details. Files: Juggernaut-XL_v9_RunDiffusionPhoto_v2.safetensors for ComfyUI, Forge, InvokeAI and SwarmUI (models/checkpoints), plus a diffusers fp16 tree. diffusers example: DiffusionPipeline.from_pretrained("RunDiffusion/Juggernaut-XL-v9", torch_dtype=torch.float16, variant="fp16", use_safetensors=True), 832x1216, 35 steps, guidance 5.0. Licence: CreativeML Open RAIL-M for personal and creative use; it may not be deployed behind paid API services without an explicit licence from RunDiffusion. Based on the RunDiffusion/Juggernaut-XL-v9 card pinned to revision cf419233522daa0b9ea36c3aff98fa2cab1fb0fb and observed on 2026-10-27. Nothing was run here; the 8 GB VRAM and quality statements are the vendor's. 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/RunDiffusion/Juggernaut-XL-v9/raw/cf419233522daa0b9ea36c3aff98fa2cab1fb0fb/README.md",
      "note": "Retained source snapshot observed 2026-10-27; TEXT.txt; UTF-8 bytes 6017:6272, 6097:6272, 6132:6272, 6199:6272, 6276:6431, 5597:5684, 5625:5684, 5653:5684, 5777:5883, 8750:8994, 4:321; exact excerpt SHA256 699e2840c97d71f9d11dd998e251fcc8f6636f50b32f496438ff218483655d18,2aae92c65ee74a4e43fefb65358fc6253592ef5eaac1e3e3d148867c94b18617,c4c89eddf2"
    }
  ]
}