AI 模型资源
精选 ComfyUI 最佳AI模型合集。来自 Civitai、Hugging Face 等平台的验证下载链接。
按分类浏览
Checkpoints
Curated collection of the most popular Stable Diffusion checkpoint models. From photorealistic to anime, find the perfect base model for your AI art workflow.
Flux
Explore the FLUX model family by Black Forest Labs — from the blazing-fast Schnell to the high-quality Dev, Kontext, Fill, and Redux variants.
LoRA
Lightweight fine-tuned LoRA adapters for adding specific styles, characters, and concepts to your generations.
ControlNet
ControlNet models for precise control over image composition using depth maps, poses, edges, and more.
VAE
Variational Auto-Encoder models that improve color accuracy and detail in generated images.
Upscale
Super-resolution models to upscale and enhance your AI-generated images to print-ready quality.
Featured Models
Juggernaut XL
One of the most popular SDXL checkpoints. Excels at photorealism, digital painting, and character generation with improved hands and poses.
DreamShaper 8
A versatile general-purpose model capable of art, anime, and photorealism. The "swiss knife" of SD 1.5 models with 57K+ favorites.
DreamShaper XL
SDXL version of DreamShaper — a general purpose model for art, photography, and anime. Features Lightning variant for 3-6 step generation.
RealVisXL V5.0
A top-tier photorealism-focused SDXL checkpoint. Produces stunning realistic images of people, landscapes, and objects with Lightning speed.
Pony Diffusion V6 XL
Massively popular SDXL finetune for diverse art styles — anime, western art, and character generation. 74K+ favorites and 130M+ generations.
Stable Diffusion XL 1.0
The official SDXL base model by Stability AI. Foundation for thousands of fine-tunes and the standard for high-resolution AI image generation.
在线试用这些模型
无需安装 — 使用 ComfyUI Web 在浏览器中直接运行AI模型。
免费试用 AI 应用常见问题
如何在 ComfyUI 中使用这些模型?▾
下载模型文件并放置到 ComfyUI 的相应文件夹中(例如:检查点模型放入 models/checkpoints/)。然后在工作流的加载器节点中选择该模型。或者直接在 ComfyUI Web 上在线试用 — 无需安装。
SD 1.5 和 SDXL 模型有什么区别?▾
SDXL 模型生成更高分辨率的图像(默认 1024x1024),相比 SD 1.5(默认 512x512)拥有更好的细节和构图。但 SDXL 需要更多显存(至少 6-8GB)。SD 1.5 模型拥有更丰富的 LoRA 和扩展生态系统。
Flux 模型是什么?有何不同?▾
FLUX 模型由 Black Forest Labs 开发,是拥有 120 亿参数的新一代图像生成模型。它们提供卓越的提示词跟随能力、文字渲染和图像质量。FLUX.1 [schnell] 仅需 1-4 步即可生成图像,而 [dev] 提供最高画质。
这些模型可以免费使用吗?▾
此处列出的大多数模型均可免费下载。但许可证条款各不相同 — 部分允许商用(如采用 Apache 2.0 的 FLUX.1 Schnell),其他则仅限非商业用途。商业使用前请务必查看模型页面上的具体许可证。
没有 GPU 可以运行这些模型吗?▾
本地运行模型需要至少 4-8GB 显存的独立显卡。或者,您可以使用 ComfyUI Web 在云端运行这些模型 — 无需 GPU,打开浏览器即可开始生成。