This is a Controlnet focused workflow for SDXL, designed to be visually compact and simplified for ease of use.This workflow utilizes CLIP G and CLIP L, which most people don't even realize is the actual way you're supposed to use SDXL. This is a dual positive prompt system. Technically you'd want to prompt each clip differently, but most people just use the same prompt for both CLIP models. CLIP L is generally better at handling keywords and subject details, while CLIP G is better at natural language and composition. Prompting differently between them can potentially get better results.This workflow utilizes Kohya Deep Shrink HiRes Fix, an infinite LoRA loader, and is geared toward DMD2 acceleration and LCM sampling settings out of the box. If you are not using a checkpoint that has DMD2 built in, you will have to use the DMD2 LoRA. HiRes Fix should ideally be used at a block number of 4-32. The downscale factor has a range of 1-9, but can generally be left at default. HiRes Fix can be used at 1024x1024, but generally you want to use it when you generate at resolutions higher than that.I find using HiRes Fix at 1500x1500 to be a good bump in image quality while maintaining coherence. I'd experiment with enabling/disabling HiRes Fix or removing the CLIP models node and plugging clip directly into the checkpoint loader, or just prompting the CLIP G field only if you're not getting the results you want.Negative prompt is hidden by default, as CFG 1 ignores negative prompt anyways. If you modify the workflow to utilize negative prompt for non-DMD2 generation or DMD2 generation with a CFG slightly above 1, it will be a dual negative prompt system._________Required models:CLIP L: <a target="_blank" rel="ugc" href="https://huggingface.co/comfyanonymous/flux_text_encoders/resolve/main/clip_l.safetensors?download=true">https://huggingface.co/comfyanonymous/flux_text_encoders/resolve/main/clip_l.safetensors?download=true</a>CLIP G: <a target="_blank" rel="ugc" href="https://huggingface.co/second-state/stable-diffusion-3.5-large-GGUF/resolve/main/clip_g.safetensors?download=true">https://huggingface.co/second-state/stable-diffusion-3.5-large-GGUF/resolve/main/clip_g.safetensors?download=true</a>DMD2 LoRAs:FP32: <a target="_blank" rel="ugc" href="https://huggingface.co/tianweiy/DMD2/resolve/main/dmd2_sdxl_4step_lora.safetensors?download=true">https://huggingface.co/tianweiy/DMD2/resolve/main/dmd2_sdxl_4step_lora.safetensors?download=true</a>FP16: <a target="_blank" rel="ugc" href="https://huggingface.co/tianweiy/DMD2/resolve/main/dmd2_sdxl_4step_lora_fp16.safetensors?download=true">https://huggingface.co/tianweiy/DMD2/resolve/main/dmd2_sdxl_4step_lora_fp16.safetensors?download=true</a>SDXL ControlNet models:<a target="_blank" rel="ugc" href="https://civarchive.com/models/136070/controlnetxl-cnxl">https://civarchive.com/models/136070/controlnetxl-cnxl</a>Recommended ControlNet models:ControlNet Union: <a target="_blank" rel="ugc" href="https://civarchive.com/models/136070?modelVersionId=655749">https://civarchive.com/models/136070?modelVersionId=655749</a>ControlNet Union ProMax: <a target="_blank" rel="ugc" href="https://civarchive.com/models/136070?modelVersionId=655836">https://civarchive.com/models/136070?modelVersionId=655836</a>

SDXL Controlnet Compact

• Converted to subgraphs for a simpler layout• Custom labeled options for ease of use• Settings toggles

Files with SHA256: 5b24fc4a00fe59844044b0599ce9909745d558ee1cd5c78d5b2d4f7e31a4329b