(I'll spiffy these descriptions up later) Short and sweet, this is a workflow designed to be used with my Anima (Soon to be Krea as well) <a target="_blank" rel="ugc" href="https://civarchive.com/models/2771981/sillytavern-auto-expressions?modelVersionId=3202811">SillyTavern Character Expression Generation Workflow</a>. It creates animated tachi-e AVIF animations that works with silly tavern. I'm currently experimenting, so who knows where this will go in the future. Input = Character with #FFFFFF background. Requirements:<ul><li>Preferably 64-96gb of RAM (Video Length changes this)</li><li>Preverably 16-20gb of VRAM (To have any meaningful video length)</li></ul>Versions:<ul><li>Simple - No git hub cloning required. Just the basics. Gets the job done.</li><li>Advanced - Requires ComfyUI-KYNode (Not in ComfyUI Manager Repos) (Allows you greater control // FFMPEG CLI // Exported frames to touch up)</li></ul><pre><code># Advanced Version Requires:

cd ~/.../ComfyUI/custom_nodes

git clone https://github.com/yorkane/ComfyUI-KYNode.git</code></pre>Consumer Note:I assume you are intelligent and are fluent in ComfyUI, the basics of how LLMs operate, as well as in CLI. (I mean, c'mon, you're a silly tavern user, aren't ya?) You don't have to be a power user but you do have to be competent enough to figure out how to make this work with your setup. Theoretically, it should work out of the box, but this is a very unique workflow designed to do a specific task, rather than have knobs and dials to tailor the workflow to your specs. So, no guarantees. I currently have it set for 9 seconds, 48fps. This is works for 96gb of ram and 20gb of VRAM. Technical Note:I use ComfyUI-LLM-text-processor node set for this workflow.What does that mean? It means you need to install llama.cpp with pip, matching your cuda version. (AMD users and other questions, ask grok or chatgpt for help) Furthermore, linux users, you may have issues with the LLM node's python syntax pointing to the wrong binary for llama.cpp. Two ways to fix this. Either, One, post the llama_<a target="_blank" rel="ugc" href="http://cli.py">cli.py</a> llama_<a target="_blank" rel="ugc" href="http://binary.py">binary.py</a> from: .../ComfyUI/custom_nodes/ComfyUI-LLM-text-processor Into your favorite AI bot and ask it to fix it to work with your distro of linux. Two, alternatively, you can opt for another node set to replace the 'LLM' node, such as the QwenVL nodes, which come self-contained with auto downloaders. I use the ComfyUI-LLM-text-processor node set because it allows me freedom to choose my favorite LLM, which currently is: <a target="_blank" rel="ugc" href="https://huggingface.co/DavidAU/Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored-NM-DAU-NEO-MTP-GGUF/tree/main">DavidAU's Qwen38TTurboFableFusion709 + mmproj-BF16</a>

Files with SHA256: a3c719023a6f664514cf5aa20408ef1c9f867bed4e73a0e5770d5486f8322791