SDXL-ControlNet model for line art conditioning
Top 59.0% on SourcePulse
MistoLine is an SDXL-ControlNet model designed for versatile and robust line art conditioning, enabling high-quality image generation from various line art inputs. It targets users working with Stable Diffusion XL who require consistent and accurate line art control across different preprocessing methods, eliminating the need to switch between multiple ControlNet models.
How It Works
MistoLine employs a novel "Anyline" preprocessing algorithm and is retrained on the stabilityai/stable-diffusion-xl-base-1.0
Unet. This approach allows it to generalize across diverse line art conditions, including hand-drawn sketches and model-generated outlines, outperforming existing ControlNet models in detail restoration, prompt alignment, and stability.
Quick Start & Requirements
.safetensors
files (e.g., mistoLine_rank256.safetensors
) and integrate into ComfyUI or AUTOMATIC1111-WebUI workflows.dpmpp_2m_sde
sampler, karras
scheduler, 0.93 denoise, 1.0 controlnet strength, and 0.0 start percent.Highlighted Details
mistoLine_rank256.safetensors
is noted as performing better than mistoLine_fp16.safetensors
.Maintenance & Community
Licensing & Compatibility
Limitations & Caveats
1 year ago
1 day