<h3 id="comfyui-workflow-for-image-generation">ComfyUI Workflow for Image Generation</h3>(it also allows you to create a video of the generated image using Wan2.x Models)<ul><li>Needed Custom Node:<ul><li>Git-Hub: <a target="_blank" rel="ugc" href="https://github.com/r-vage/ComfyUI-RvTools_v2">RvTools_v2 (Updated: 2.4.45)</a></li><li>Comfyui-Registry: <a target="_blank" rel="ugc" href="https://registry.comfy.org/nodes/comfyui-rvtools_v2">RvTools_v2</a>if you install the node with the manager choose nightly because i don't upload every change to comfyui registry. if you choose latest you'll install the latest version stored on comfy reg. nightly is the current version from git hub.</li></ul></li><li>if you get type mismatch errors, install the frontend version 1.26.0<ul><li>python_embeded\python.exe -m pip install comfyui-frontend-package==1.26.0</li></ul></li><li>if you want to use pulid or reactor you need to install numpy 1.26.4<ul><li>python_embeded\python.exe -m pip install numpy==1.26.4</li></ul></li><li><a target="_blank" rel="ugc" href="https://civarchive.com/articles/15787/listing-links-resources">Resource Listing</a></li><li>when using torch 2.7.x+cu128 use triton 3.3.1 (3.4 will cause issues)<ul><li>python_embeded\python.exe -m pip install -U "triton-windows==3.3.1.post19"</li></ul></li></ul>When you look at the image showing the workflow, you can see that it is divided into groups. Each group offers useful tools for influencing the image to be generated. Almost every group has a ctx (context menu) for input and one at the end for output of this group. If there are groups in the workflow that you do not need or cannot use, select the entire group (hold Ctrl + draw a rectangle with the left mouse button over the area) with all nodes in it and the two ctx and delete them. The same applies if you want to duplicate a group e.g. a detailer. Select the group+ctx, copy it, and paste it into a free space. Connect the two ctx with the ones where the group is to be inserted.the workflow has 2 stop funtions:<ul><li>one is in the initial render group, it allows you, when using a random seed, to stop after each render and choose if you want to quene for a new image or set the seed to fixed and process this image.</li><li>the 2nd is before the wan groups. if this one is enabled and auto queue is also enabled the wan groups are ignored. means you can render multible images without creating a video of each of them. if auto queue is not enabled the workflow stops after the image is processed and you can adjust the prompt for the video in the wan: prompt group, disable the stop option after that and start a new queue (fixed seed!) to create a video of the generated image.</li></ul>

iGEN II for FLUX | SD | HiDream | Qwen | I2V (STD | GGUF)

<ul><li>replaced the LLM Node in "tags to natural language" using searge llm now, works way better.</li><li>install instructions are in the workflow (updated)</li><li>updated the model description click on "show more"</li></ul>

Files with SHA256: 7bf0cc14f65301997e1afe1e9c95e9b8de37aea02ea4fa6edc24d820a02b1fc8