这是在我以前的工作流插件上进行改进的升级版, 注意：这是一个采样器，同时也是一个简单的工作流，一个简单的采样器可以串联N个，这样可以达到多次采样，同时还具备外接其它插件。 别外还有一个版本没有发布，它内置了放大功能。 使用方法很简单，选好一切参数，输入提示词，外接保存图片节点，点运行就可以得到图片，不再像普通的其它工作流有繁所的节点，使画布变得简洁模型选择：有分离式和整包选，整包适合以前的老模型，比如SD SDXL。分离试支持：flux /krea2/z image 等，clip有二个，不管你用那个模型，第一个clip必须得选 ，第二用在双clip，如果没有双clip不用选，其它的参数跟你一往一样选即可This is an upgraded version of my previous workflow plugin.Note: This acts as both a sampler and a simple workflow. You can chain multiple simple samplers together for multi-step sampling, and it also supports connecting to other external plugins. (By the way, there’s another unreleased version that comes with a built-in upscale feature!)How to use: It’s super simple. Just set your parameters, enter your prompt, connect an image-saving node, and hit Run to get your image. No more cluttered canvases with endless nodes like in typical workflows!Model Selection: You can choose between Unified and Separated modes:<ul><li>Unified mode: Best for older models like SD and SDXL.</li><li>Separated mode: Supports Flux, Krea 2, Z Image, etc.</li></ul>About CLIP: There are two CLIP slots. No matter which model you use, the first CLIP is mandatory. The second one is only needed if your model requires dual CLIPs; otherwise, just leave it as is.For all other parameters, just set them up exactly as you normally would!

binyuan_sampler_plugin_v6.0

如果不明白，可以去<a target="_blank" rel="ugc" href="https://github.com/yjitiu/ComfyUI-Binyuan-Sampler-EN">yjitiu/ComfyUI-Binyuan-Sampler-EN：ComfyUI自定义节点——双语（英/中文）一体化Ultimate Sampler V6.5 EN。加载整套检查点或分割堆栈（扩散+双CLIP+VAE）、Flux/SD3/Wan/Qwen-Image/Krea2/Z-Image/LTXV、LoRA列表、fp8/nvfp4。可以安装在中文版旁边。</a>应大伙的要求，我又重新写了一个中英文双语版，这样方便使用英文的人使用。使用时注意以下几点：第一：解压，整个文件夹丢进 ComfyUI/custom_nodes/，重启 ComfyUI第二：在comfyui画布里右键 → Binyuan → 🛡️ Binyuan Sampler V6.5 EN｜中英双语版特别留意：如果出图时报 memory_efficient_attention_forward / capability (12, 0)，那是显卡（RTX 50 系）+ xformers 的环境问题 —— 官方 KSampler 也会一样报，最好的办法是启动加 --use-pytorch-cross-attention --disable-xformers 即可解决。As requested by everyone, I’ve rewritten a bilingual (Chinese-English) version to make it easier for English users to utilize. Please note the following points when using:<ol><li>Extract the files and place the entire folder into <code>ComfyUI/custom_nodes/</code>, then restart ComfyUI.</li><li>Right-click in the ComfyUI canvas → Binyuan → 🛡️ Binyuan Sampler V6.5 EN｜Bilingual Version</li></ol>Special Note: If you encounter a <code>memory_efficient_attention_forward / capability (12, 0)</code> error during image generation, this is an environment issue related to your graphics card (RTX 50 series) and xformers. The official KSampler will report the same error. The best solution is to add <code>--use-pytorch-cross-attention --disable-xformers</code> to your startup arguments to resolve it.

Files with SHA256: 090b1b39ca9876fb9bda08fb275c725f7db5f0771e4622a615d6249444989a9e