Tag
Recommends the locally deployable face swap LoRA "BFS-Best-Face-Swap," supporting face, head, and body replacement while preserving the original hair, skin texture, and lighting. Provides ComfyUI workflows, Draw Things presets, and Python Diffusers deployment options.
Claude Opus 5.5 via Claude Code autonomously produced a full 3D music video for Senbonzakura, using Blender MCP for modeling/rendering and ComfyUI (Anima) for character generation under a single high-level prompt. The project demonstrates end-to-end agentic creative production with minimal human intervention.
Sam Wasserman 透露他们为 ComfyUI 构建了一个开源版 Higgsfield 平台，包含 61+ 应用（视频、音乐、音频、音效、代理、提示辅助、深度映射等）的完整专业工作流套件，而非单一功能。
Licon VBVR 是一个针对 MiniMax-H3 的图像转视频 LoRA，基于重建的 VBVR 数据训练，可减少物体复制、裁剪与穿插问题，并提升角色动作逻辑与空间一致性。
Comfy-Org 发布 Fibo-2.0 的 ComfyUI 重打包模型文件，包含 bf16 与 int8 量化版本，并附带 Qwen3-VL 4B 文本编码器与 FLUX2 VAE，供非商业用途使用。
GameSummoner 发布一周后更新，大幅简化本地部署，现在只需一行安装脚本即可运行，支持 llama-swap 切换模型、用 ComfyUI 工作流生成游戏，并可与 Claude Opus 5.5、DeepSeek V4.1、Gemini 4 等前沿模型对比效果。
Kromoss is a full fine-tune of MOSS-Image-Base at 1024px, shipping with two 8-step turbo LoRAs (DMD and DMD2-distilled) and drop-in support for diffusers and ComfyUI, including an int8 quantized variant.
Comfy-Org 发布了 MOSS-Image 模型的 ComfyUI 重打包文件，包含 base 与 SFT 版本的 bf16/int8 权重，以及配套的 Qwen3.5 9B 文本编码器和 Flux2 VAE。
Hugging Face 上发布 lodestones/MOSS-Image-SFT-8-steps-dmd，一个将 MOSS-Image-SFT（50 步、CFG 3）蒸馏为 8 步、无 CFG 的工作进展检查点，采用 data-space DMD 方法，可作为 LoRA 用于 ComfyUI。
Black Forest Labs 发布 FLUX 3 Image Klein 的 ComfyUI 打包版本，这是一个 5.6B 参数的开放权重模型，统一支持文生图、单参考图编辑和多参考图合成，并支持提示词上采样器与多种量化格式。
作者分享了 Qwen-image2.1 "去审查版" 的本地部署教程，称在仅 8G 显存的 RTX 5060 上即可运行，并整理了 5 个常见坑点，提醒大家适度使用。
An introduction to AtomicChat's "decensored" text encoder for Qwen-Image-2.1-Turbo, released on HuggingFace in GGUF quantized form, plus a complete tutorial for running it locally with ComfyUI on 8GB VRAM (RTX 5060). Includes the full file download checklist, node configurations, 5 real-world gotchas with their fixes, and prompt examples.
OpenMOSS 发布了 MOSS-Image-Base 文生图基础模型，支持自然语言提示、标签序列和 JSON 布局控制文本位置与物体排列，并配套 SFT 版本、ComfyUI 插件和提示增强工具。
DGX Monarch是個人開發的開源專案，讓使用者能在雙DGX Spark叢集上透過ComfyUI進行圖像/影片生成，達到近2倍的渲染速度提升，支援BF16/FP8/NVFP4/INT8等多種精度。作者同時分享了使用AI輔助開發此專案所耗費的各平台訂閱成本。
Kandinsky Lab releases Kandinsky 6.0 Pro (29B) and Lite (3B) video generation models with a built-in video upscaler, already supported in ComfyUI and Diffusers.
ComfyUI v0.39.0 has been released, bringing updates to the modular, node-graph AI creation engine for images, video, 3D, and audio that emphasizes optimized local inference on consumer hardware.
Scumble is a free, open-source desktop editor for AI inpainting that turns text prompts into color-matched layers, running on users' own ComfyUI instances or API keys for models like FLUX 3, GPT Image, and Nano Banana.
The Qwen Image 2.1 Consistency LoRA is a tool that maintains image consistency during edits, preventing drift and unwanted repaints in AI image generation workflows.
Akatz Labs released an experimental MiniMax H3 character-swap LoRA (1,000 training steps) that replaces a selected character in a source video using an image reference while preserving the scene. The authors note background preservation is decent, but motion timing, facial expressions, and hard cuts remain unreliable.
The user experimented with running Qwen 27B and Qwen Image 2.1 on local GPUs to generate 3D models and renders for printing, using tools like CADQuery and ComfyUI, and plans future projects for creating models from photos.