⚡ ComfyUI-Binyuan-Sampler (V3.0 Flagship)

冰源全能采样器 - 终极一体化工作流解决方案

Total Integration: A single node that replaces 5-8 standard nodes. Manages Model, CLIP, VAE, LoRA, Latent, and Conditioning outputs in one place.
Smart Resolution Helper: Built-in presets from 512px up to 2K. Includes aspect ratio labels (16:9, 9:16, etc.) and real-time UI synchronization.
Universal Architecture Fix: Custom-built logic to prevent resolution doubling or artifacts when using Flux, Lumina2, and ERNIE-Image models.
Infinite Dynamic LoRA: Add/Remove LoRA slots with a single click. Features independent [Enable/Disable] toggles and precision keyboard weight input.
Visual Feedback: Real-time colored progress monitoring in the console during model/LoRA loading.

Basic T2I: Connect MODEL, CLIP, and VAE from your loader. Select a size from the Size Helper and run.
Dual-Sampling (I2I): Connect the Latent, Model, VAE, and Conditioning outputs of your first Binyuan Sampler directly to the inputs of a second one. Set Denoise to 0.4-0.6 for high-res refinement.
External Ref: Connect any image to image_ref_1 for instant Image-to-Image workflows without extra VAE Encode nodes.

Open your terminal in ComfyUI/custom_nodes.
Run: git clone https://huggingface.co/yjitiu/ComfyUI-Binyuan-Sampler
Restart ComfyUI.
Note: If upgrading, delete the old node on your canvas and re-add it to refresh the UI.

基础文生图：连接加载器的 模型、CLIP 和 VAE。从尺寸助手选择比例后即可出图。
双采样接力：支持无限串联。将第一个节点的 Latent、模型、VAE 和 条件 输出直接连到第二个节点的对应输入，设置 重绘强度 为 0.4-0.6 即可实现高清精修。
外部参考：直接将图像节点连入 参考图_1，节点会自动执行 VAE 编码进入图生图模式。

进入 ComfyUI 的 custom_nodes 文件夹。
在地址栏输入 cmd 打开终端，运行： git clone https://huggingface.co/yjitiu/ComfyUI-Binyuan-Sampler
重启 ComfyUI 即可在“Binyuan”分类下找到。

Created with ❤️ by Binyuan. Welcome to give feedback in the Community channel! 欢迎在社区频道反馈建议！

Downloads last month: -; Downloads are not tracked for this model. How to track