<h1 id="what-is-this-workflow-qxuw5elnx">What is this workflow?</h1>This is my personal ComfyUI workflow built upon what I've learned and used in SD1.5, expanded to cover all possible use cases for myself. On a 4GB VRAM + 16GB RAM system, with everything active, it can still run and provide strong results (if you're willing to wait a while).This workflow includes so, so many things that it's better to view it for yourself. The following is a small subset of supported features of this workflow:<ul><li>v-prediction support + RescaleCFG (disabled by default)</li><li>Controllable CLIP Skip</li><li>ELLA + ollama prompt upscale (SD1.5 exclusive feature)</li><li>Scalable ControlNet group (disabled by default)</li><li>Scalable IP-Adapter group (disabled by default)</li><li>Dynamic Thresholding (disabled by default)</li><li>2-pass txt2img</li><li>Perlin noise latent blend</li><li>Watermark removal using CLIPSeg + Lama cleaner</li><li>Scalable ADetailer group</li><li>Notifications and sounds</li><li>Preview chooser for batch images</li><li>Full Civitai metadata support</li><li>Prompt and LoRA scheduling</li><li>Multi-checkpoint setup (still in testing)</li><li>Wildcard (and wildcard file) support</li><li>...</li></ul>Despite being built for SD1.5 first and foremost, this workflow can also use SDXL models or any all-in-one checkpoint models easily that is supported by <a target="_blank" rel="ugc" href="https://github.com/receyuki/comfyui-prompt-reader-node?tab=readme-ov-file">prompt-reader-node</a>. In these cases, ELLA would be completely useless, so do not enable it.<h1 id="how-to-use-this-workflow-3b21iklfr">How to use this workflow?</h1><ul><li>Step 0: Get <a target="_blank" rel="ugc" href="https://github.com/comfyanonymous/ComfyUI">ComfyUI</a> and <a target="_blank" rel="ugc" href="https://github.com/ltdrdata/ComfyUI-Manager">ComfyUI-Manager</a></li><li>Step 1: Download the workflow file</li><li>Step 2: Import it into ComfyUI</li><li>Step 3: Download all missing nodes</li><li>Step 4: Grab all missing models (more details below)</li><li>Step 5: Have fun generating!</li></ul><h2 id="model-requirements-2awz1djng">Model requirements</h2><h3 id="basic-o1ylo1joy">Basic</h3><ul><li>A checkpoint file (SD1.5, SDXL,...)</li><li>A VAE file (optional if baked VAE)</li><li><a target="_blank" rel="ugc" href="https://openmodeldb.info/models/4x-AnimeSharp">4x-AnimeSharp</a> upscale model (or any other 4x upscale model of choice)</li><li><a target="_blank" rel="ugc" href="https://github.com/Sirosky/Upscale-Hub/releases/tag/AniScale2">2x-AniScale2 </a>upscale model (or any other combination of upscale model for the final upscale step)</li></ul><h3 id="controlnet-aq1g91zdx">ControlNet</h3>All models are available inside ComfyUI-Manager under Model Manager. If you're using PonyXL or IllustriousXL, adapt your own solutions to it.<ul><li>Depth ControlNet (using Depth-Anything preprocessor)</li><li>Lineart ControlNet (using AnyLine Lineart preprocessor)</li><li>OpenPose ControlNet (using DWPose preprocessor)</li><li>Depends on what other ControlNet model you add</li></ul><h3 id="other-features-9zvdfyz7m">Other features</h3><ul><li>IP-Adapter: an IP-Adapter model and its related CLIP-G (CLIP Vision) model. They should all be available under Model Manager</li><li>ELLA: See <a target="_blank" rel="ugc" href="https://github.com/TencentQQGYLab/ComfyUI-ELLA?tab=readme-ov-file#orange_book-models">https://github.com/TencentQQGYLab/ComfyUI-ELLA?tab=readme-ov-file#orange_book-models</a></li><li>ADetailer/FaceDetailer: Most models should be available under Model Manager, except Anzhc's face YOLO, which can be acquired at <a target="_blank" rel="ugc" href="https://huggingface.co/Anzhc/Anzhcs_YOLOs">https://huggingface.co/Anzhc/Anzhcs_YOLOs</a>. Place the downloaded file inside ultralytics/segm</li><li>Ollama: See <a target="_blank" rel="ugc" href="https://ollama.com/download">https://ollama.com/download</a> and follow their instructions. After installing ollama on your system, grab <a target="_blank" rel="ugc" href="https://ollama.com/huihui_ai/llama3.2-abliterate:3b">https://ollama.com/huihui_ai/llama3.2-abliterate:3b</a></li></ul>

[ComfyUI Workflow] All-in-one Text-to-Image Workflow (ControlNet, IP-Adapter, ADetailer, ELLA,...)

This is very much overdue at this point, but here's the updated, hopefully not broken (and insanely overkill) version update.Changelog: V2.0 - Restart<ul><li>Switched PowerNoiseSuite's Perlin noise sampler to Restart sampler + Latent Noise Injection<ul><li>WARNING: Restart sampling can and will stress your system out, so be careful</li></ul></li><li>Switched from comfyui-prompt-reader-nodes back to ComfyUI Image Saver for better LoRA support and better caching<ul><li>As part of the transition to Image Saver, the 2nd checkpoint will also be included within the metadata, and LoRA loading will be handled by loading through prompts. To best use this system, enable LoRA autocomplete in ComfyUI-Custom-Scripts by going into Settings -&gt; pysssss -&gt; Autocomplete -&gt; Loras enabled</li></ul></li><li>Merged IP-Adapter and ControlNet to one Image References section and condensed certain features (mostly IP-Adapter)</li><li>Added WD1.4 Tagger to make use of images more effectively</li><li>Split certain sections to allow for more refined control in each section</li><li>Added more advanced model-changing settings (CFGZeroStar, RenormCFG, and MaHiRo Guidance,...) [probably mutually exclusive, but could work together]</li><li>Added documentations for usability</li></ul>(As of 24/12/2025: Slightly updated to fix certain issues like Schedule Selector being unavailable and converted some of the merged nodes into subgraphs)

Files with SHA256: 11afe4510d3b5391400ba8c9617a713e17d212183ca6e74a830f215c9dcd9699