Unofficial community derivative. Not an OpenMOSS release. This page is a mirror/share only.An unofficial, all-in-one ComfyUI checkpoint for anime-oriented image generation based on MOSS-Image. It combines an INT8 ConvRot, fused-QKV diffusion backbone with the native Qwen3.5-9B text encoder and FLUX.2 VAE into a single file. The diffusion backbone incorporates the community miaoka turbo1 derivative and an NSFW LoRA at 0.7 strength. It is biased toward anime illustration and adult-content capability compared with the official SFT model.License: Apache-2.0 Base: OpenMOSS-Team/MOSS-Image-SFT Author of this merge: Langzaigg (Hugging Face) This upload: Mirror/share only. Not my checkpoint.<hr /><h2 id="architecture-and-components">Architecture &amp; Components</h2><ul><li>Diffusion backbone: INT8 ConvRot with fused-QKV layout</li><li>Text encoder: Native Qwen3.5-9B (included)</li><li>VAE: FLUX.2 VAE (included)</li><li>Merged components:<ul><li>Community miaoka turbo1 derivative (in the diffusion backbone)</li><li>NSFW LoRA at 0.7 strength</li></ul></li><li>Bias: Anime illustration + adult content capability</li></ul>No separate text encoder, VAE, LoRA, or QKV-remapping patch is required.<hr /><h2 id="file">File</h2>FileSizeFormat<code>moss_image_anime_turbo.safetensors</code>14.44 GiB Safetensors checkpointPlace the file in:<pre><code>ComfyUI/models/checkpoints/
</code></pre>Load with CheckpointLoaderSimple or a similar node.<hr /><h2 id="recommended-settings">Recommended Settings</h2>This is a turbo model. Do not treat it like a standard 20–50 step diffusion model.Steps 4Sampler<code>res_multistep</code>Scheduler<code>simple</code>CFG1.0Negative promptEmptyDenoising1.0Resolution~1 MP (width and height divisible by 16) Showcase uses CFG 1.1 WITH negative prompts and 2 MP resolution, so you see this works as well.Fallback: If <code>res_multistep</code> produces anomalies, use <code>euler</code> + <code>simple</code> at 4 steps.<hr /><h2 id="usage">Usage</h2><ol><li>Download <code>moss_image_anime_turbo.safetensors</code>.</li><li>Place it in <code>ComfyUI/models/checkpoints/</code>.</li><li>Load via CheckpointLoaderSimple.</li><li>Use the settings above.</li><li>Optional: Works with <a target="_blank" rel="ugc" href="https://github.com/Langzaigg/ComfyUI-AnimaFlow">ComfyUI-AnimaFlow</a> for structured tags + English descriptions and HTTP/MCP workflows.</li></ol><h3 id="input-support">Input Support</h3><ul><li>Natural-language captions</li><li>MOSS Layout JSON (positioned elements, panels, text)</li><li>Reference images (semantic guidance only; does not preserve identity, clothing, or logos exactly)</li></ul><hr /><h2 id="evaluation-notes-(from-original-hf-card)">Evaluation Notes (from original HF card)</h2><ul><li>95/95 successful generations in controlled tests (19 scenes, same prompts/seeds/resolutions).</li><li>Tested with Euler (4/6 steps) and <code>res_multistep</code> + <code>simple</code> (4 steps).</li><li>Updated checkpoint showed improved backgrounds, clothing, materials, and complex actions compared with an earlier version.</li><li>Median generation time (updated checkpoint, 4 steps): ~27–28 seconds depending on sampler.</li></ul><hr /><h2 id="limitations">Limitations</h2>May fail at:<ul><li>Complex anatomy</li><li>Object interactions</li><li>Exact layout adherence</li><li>Negative-concept suppression</li></ul>Small or multi-paragraph text may be malformed. Prefer short literal strings. Reference images provide semantic guidance only.<hr /><h2 id="safety">Safety</h2>This checkpoint contains an NSFW LoRA and can generate adult content.<ul><li>Do not generate sexual content involving minors.</li><li>Avoid non-consensual intimate imagery or harmful content.</li><li>Use only where lawful.</li></ul>For safer outputs, explicitly specify clothing, pose, and scene in the prompt.<hr /><h2 id="credits">Credits</h2><ul><li>Base model: <a target="_blank" rel="ugc" href="https://huggingface.co/OpenMOSS-Team/MOSS-Image-SFT">OpenMOSS-Team/MOSS-Image-SFT</a> (Apache-2.0)</li><li>INT8 ConvRot source: <a target="_blank" rel="ugc" href="https://huggingface.co/toxicdog/MOSS-Image-SFT-Turbo-INT8-ConvRot">toxicdog/MOSS-Image-SFT-Turbo-INT8-ConvRot</a></li><li>Community components: miaoka turbo1 derivative and NSFW LoRA (merged)</li><li>This merge: Langzaigg – <a target="_blank" rel="ugc" href="https://huggingface.co/Langzaigg/moss-image-anime-turbo">Hugging Face</a></li><li>This CivitAI page: Mirror/share only. All credit to the original authors and contributors listed above.</li></ul><hr /><h2 id="license">License</h2>Apache-2.0 for this model and upstream components. Verify individual licenses of any further derivatives before redistribution or production use.This is an unofficial merge. Verify provenance and comply with upstream terms.

[Model] MOSS-Image Anime Turbo – Unofficial All-in-One ComfyUI Checkpoint

initial release, mirror from HuggingFace

Files with SHA256: 9b535b3ace628cfe87075b7516bc9bb437791a0ff34f1043d87ab991de0dc969