<h3 id="comfyui-workflow-(image-generation)">Comfyui Workflow (Image Generation)</h3><hr /><ul><li>required Custom Nodes:<ul><li>ComfyUI_Eclipse: Version 2.4.49, <a target="_blank" rel="ugc" href="https://github.com/r-vage/ComfyUI_Eclipse">GitHub</a>, <a target="_blank" rel="ugc" href="https://registry.comfy.org/publishers/rvage/nodes/ComfyUI_Eclipse">ComfyReg</a></li><li>ComfyUI_SmartLML: Version 2.5.4, <a target="_blank" rel="ugc" href="https://github.com/r-vage/ComfyUI_SmartLML">GitHub</a>, <a target="_blank" rel="ugc" href="https://registry.comfy.org/publishers/rvage/nodes/ComfyUI_SmartLML">ComfyReg</a></li><li><a target="_blank" rel="ugc" href="https://github.com/numz/ComfyUI-SeedVR2_VideoUpscaler">https://github.com/numz/ComfyUI-SeedVR2_VideoUpscaler</a></li><li>Ollama Registry: <a target="_blank" rel="ugc" href="https://registry.ollama.ai/huihui_ai">https://registry.ollama.ai/huihui_ai</a><ul><li>to run qwen3.5 update the ollama image (0.17.7): docker pull ollama/ollama:latest 2&gt;&amp;1</li><li>remove the eclipse-container: docker rm eclipse-ollama 2&gt;&amp;1</li></ul></li></ul></li><li>if you need help installing nunchaku or how to use the nodes check the readme files in the repo <a target="_blank" rel="ugc" href="https://github.com/r-vage/ComfyUI_Eclipse/tree/main/Readme">readme</a><a target="_blank" rel="ugc" href="https://github.com/r-vage/ComfyUI_RvTools/tree/main/Readme"> </a>folder at github.</li><li>llama-cpp-python is needed for the loading_method: GGUF (llama-cpp-python) in the Smart LML Node<ul><li><a target="_blank" rel="ugc" href="https://github.com/r-vage/ComfyUI_Eclipse/releases/download/wheels/llama_cpp_python-0.3.16-cp312-cp312-win_amd64.whl">https://github.com/r-vage/ComfyUI_Eclipse/releases/download/wheels/llama_cpp_python-0.3.16-cp312-cp312-win_amd64.whl</a></li></ul></li></ul><ul><li>nunchaku wheels:<ul><li><a target="_blank" rel="ugc" href="https://github.com/nunchaku-tech/nunchaku/releases">https://github.com/nunchaku-tech/nunchaku/releases</a></li><li><a target="_blank" rel="ugc" href="https://github.com/nunchaku-tech/nunchaku/releases/download/v1.0.2/nunchaku-1.0.2+torch2.9-cp312-cp312-win_amd64.whl">https://github.com/nunchaku-tech/nunchaku/releases/download/v1.0.2/nunchaku-1.0.2+torch2.9-cp312-cp312-win_amd64.whl</a></li></ul></li><li>nunchaku model files:<ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/spooknik/Fluxmania-SVDQ/tree/main">https://huggingface.co/spooknik/Fluxmania-SVDQ/tree/main</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/nunchaku-tech/nunchaku-flux.1-dev/tree/main">https://huggingface.co/nunchaku-tech/nunchaku-flux.1-dev/tree/main</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/nunchaku-tech/nunchaku-qwen-image">https://huggingface.co/nunchaku-tech/nunchaku-qwen-image</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/nunchaku-tech/nunchaku-qwen-image-edit-2509/tree/main">https://huggingface.co/nunchaku-tech/nunchaku-qwen-image-edit-2509/tree/main</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/nunchaku-tech">https://huggingface.co/nunchaku-tech</a></li></ul></li><li>Wildcards<ul><li><a target="_blank" rel="ugc" href="https://civarchive.com/models/20868?modelVersionId=532723">https://civarchive.com/models/20868?modelVersionId=532723</a></li></ul></li></ul><ul><li>Version 3 uses a new layout and replaces context pipes with KJ's Set/Get nodes plus 2 additional Eclipse nodes (GetFirst and GetAllActive).</li></ul><ul><li>Why the layout changed: In the old workflow, changing the model or LoRAs triggered a full re-run including SmartLML's image description, which is slow. The new layout decouples this — SmartLML runs once to describe the image, and that result is reused via Set/Get nodes. Subsequent runs skip the description step when the seed/index is fixed. When shuffling images (-4), SmartLML naturally re-runs since the image changes.</li></ul><ul><li>The basic guider is gone — the CFG guider is used for everything (set CFG to 1.0 for Flux, etc.).</li></ul><ul><li>Some rgthree nodes have been ported to Eclipse (Fast Bypasser/Muter, Fast Groups Bypasser/Muter, Repeater, Image Compare). They are converted to work with the V3 API and Nodes 2.0, using a different approach — no LiteGraph.registerNodeType interception or subclassing.</li></ul><ul><li>GetFirst &amp; GetAllActive: Virtual frontend nodes that extend KJNodes' Set/Get system with priority-based variable resolution.<ul><li>Get First resolves the first active (not muted/bypassed) SetNode from a prioritized list of variables — ideal for fallback chains (e.g., try LoRA model first, fall back to base model).</li></ul><ul><li>Get All Active outputs all active SetNode variables simultaneously, each on its own output slot — perfect for collecting multiple active components.</li></ul><ul><li>Both nodes feature:<ul><li>Type filtering to show only matching SetNode variables</li><li>Automatic rename tracking when SetNodes are renamed</li><li>Reorder context menu (Move to Top / Up / Down / Bottom, Insert Above)</li><li>Green dot indicators showing which variables are currently active</li><li>Optional virtual link visualization</li></ul></li></ul></li><li><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/5df8be6a-2942-46a8-a656-2e986472378d/original=true/5df8be6a-2942-46a8-a656-2e986472378d.jpeg" /></li></ul>

iGEN II for FLUX1/2 | SD | HiDream | Qwen | Z-Image (STD | GGUF | Nunchaku)

<ul><li>fixed logical issues in the order from load image to the final ref image<ul><li>resize before crop was just dumb. now it's load image -&gt; crop -&gt; resize -&gt; rembg</li><li>if you have errors in the auto crop group about bert, update the node layerstyle adv.</li></ul></li><li>the core 3.33 didn't populate the loaded model data (set pipe initial render) because of the generation data node that i've used (it was already fixed in the adv. version)</li><li>because i hate to work with the 3.33 adv layout i've deleted this version too.</li></ul>

Files with SHA256: 3d2fc62d6e5fd91a3ba03c259cca18841ed50a2439b24547ab0e08eed1b5bf7e