ANIME CONTROLNET MASTER — OpenPose, Depth, Canny &amp; HED | Complete Control PanelOverviewTake control of your anime image generation with ANIME CONTROLNET MASTER, a structured ComfyUI workflow designed to preserve the composition of a reference image while generating anime characters with greater control over pose, perspective, contours, and fine details.Built around the WAI Illustrious SDXL checkpoint, this workflow combines four independently controllable ControlNets with a two-stage generation and refinement process, AI upscaling, and dedicated face and hand detailing.Whether you’re creating anime illustrations, recreating poses, experimenting with character designs, or refining existing compositions, this workflow gives you a flexible foundation for controlled image generation.✨ Key Features1. Four Independent ControlNetsEach ControlNet has its own enable/bypass control, allowing you to test different combinations without activating every controller simultaneously.<ul><li>OpenPose / DWPose: Extracts body, hand, and facial keypoints to guide the character’s pose and body structure.</li><li>Depth: Helps preserve spatial relationships, perspective, and foreground/background structure.</li><li>Canny: Extracts strong edges and contours to guide the overall shape and composition.</li><li>HED / Soft Edge: Captures softer contours and structural details that may be lost with hard-edge detection.</li></ul>Important: You can enable or disable each ControlNet independently using the Fast Groups Bypasser panel. Compare the results and find the combination that works best for your reference image.2. Visual Control Map PreviewsInspect the generated control maps before committing to your final image.Preview the OpenPose skeleton, Depth map, Canny edges, and HED soft edges to identify problems with the reference processing before adjusting your generation parameters.3. Two-Stage Generation and RefinementThe workflow uses a main generation stage followed by a low-denoise refinement pass.<ul><li>Stage 1 — Main Generation: Establishes the character, composition, and overall image structure.</li><li>Stage 2 — Refinement: Uses a second sampling pass to improve details while helping preserve the existing composition.</li></ul>4. 4x UltraSharp UpscalingIncludes the 4x-UltraSharpV2 upscaler to increase image resolution before the refinement stage.The workflow also includes a tiled VAE decoding and encoding path to support the image-processing pipeline.5. Regional Face and Hand DetailingDedicated detailing groups allow you to refine facial features and hands separately.<ul><li>Face Detailer: Targets the face for localized refinement.</li><li>Hand Detailer: Targets hand regions to help improve anatomy and finger details.</li></ul>The workflow uses detection and segmentation components to isolate target regions, reducing the need to regenerate the entire image for localized corrections.6. Organized Workflow with Bilingual NotesThe node graph includes organized sections, preview nodes, parameter recommendations, and explanations in English and Portuguese.This makes the workflow easier to understand, customize, and troubleshoot.🧰 RequirementsBase checkpoint<ul><li>WAI Illustrious SDXL V17 (waiIllustriousSDXL_v170.safetensors)</li></ul>Required custom nodes<ul><li>ComfyUI-Impact-Pack</li><li>ComfyUI-Impact-Subpack</li><li>comfyui_controlnet_aux</li><li>rgthree-comfy</li></ul>Models and assets<ul><li>OpenPose ControlNet compatible with Illustrious/SDXL</li><li>Depth ControlNet compatible with Illustrious/SDXL</li><li>Canny ControlNet compatible with Illustrious/SDXL</li><li>HED / Soft Edge ControlNet compatible with Illustrious/SDXL</li><li>Depth Anything V2 preprocessor model</li><li>DWPose detector models</li><li>Face and hand detection models</li><li>SAM segmentation model</li><li>4x-UltraSharpV2 upscaler</li></ul>Make sure all required models are installed in the appropriate ComfyUI model folders. Exact folder names may vary depending on your installation and custom-node versions.⚙️ Recommended Starting ParametersThese are suggested starting points, not universal optimal settings.ParameterStarting valueOpenPose strength0.85Depth strength0.40Canny strength0.65HED strength0.50Main sampler steps28Main sampler CFG6.5Main sampler denoise1.00Refinement steps16Refinement CFG5.5Refinement denoise0.25Face refinement denoise0.22Hand refinement denoise0.38Adjust these values according to your checkpoint, reference image, prompt, and desired level of structural preservation.🚀 How to Use<ol><li>Load the workflow into ComfyUI.</li><li>Select your installed Illustrious/SDXL checkpoint and configure the desired character LoRA and prompt.</li><li>Load a single reference image.</li><li>Inspect the four ControlNet preview maps.</li><li>Use the control panel to enable or bypass individual ControlNets.</li><li>Generate the initial image and evaluate the composition.</li><li>Allow the upscale and low-denoise refinement stages to process the image.</li><li>Enable face and hand detailing as needed.</li><li>Compare the final result and adjust the controller strengths for another pass.</li></ol>Pro tip: Start with OpenPose alone, then test Depth, Canny, and HED individually before combining them. This makes it easier to understand which controller improves the result and which may be introducing unwanted changes.🎯 Best Use Cases<ul><li>Anime character generation from reference compositions</li><li>Pose-guided illustration</li><li>Composition and perspective preservation</li><li>Character replacement experiments</li><li>Anime artwork refinement</li><li>Controlled comparisons between different ControlNet combinations</li><li>Face and hand correction</li><li>High-resolution image finishing</li></ul>⚠️ Important NotesThis workflow is a flexible generation and refinement setup, not a guarantee of perfect reproduction. Results depend on the reference image, checkpoint, prompts, ControlNet compatibility, and parameter choices.The included workflow is configured with example character/style LoRAs. Replace them with your own compatible LoRAs when needed, and make sure every model referenced by the workflow is installed.This workflow is intended for users who want more control over their ComfyUI generation process and are willing to experiment with different settings.If you find this workflow useful, leave a review, share your results, and let me know which ControlNet combination works best for you!Happy creating! 🎨

Version 1.0 — Initial ReleaseFirst public release of the ANIME CONTROLNET MASTER workflow for ComfyUI.Features<ul><li>Independent ControlNet activation and bypass controls.</li><li>OpenPose for pose and body positioning control.</li><li>Depth for scene depth and spatial structure.</li><li>Canny for edge and contour guidance.</li><li>HED for soft-edge and structural guidance.</li><li>Face and hand detailing for improved anatomical refinement.</li><li>Two-stage image generation and refinement.</li><li>4x UltraSharp upscaling for enhanced image resolution.</li><li>Organized control panel for easier workflow management.</li></ul>Notes<ul><li>Enable only the ControlNet modules you need for each generation.</li><li>Recommended settings are starting points and may require adjustments depending on the checkpoint, prompt, LoRAs, and input image.</li><li>Custom nodes and compatible models must be installed separately.</li><li>Results may vary depending on your configuration.</li></ul>Thank you for downloading and testing this workflow. Feedback and suggestions for future improvements are welcome!

Files with SHA256: bb8e2f309b20dcb66956c34bb5ea3adb68c810592b245a9d682c5054f9510f18