PainterSampler for ComfyUI

ComfyUI 图生视频双模型串联采样器.本节点由抖音博主：绘画小子 制作

---

一个节点，一样画质。
One node, same quality.

**用单个 PainterSampler 替换两个 KSamplerAdvanced —— 显存、内存、生成效果完全一致。但是不需要再连那么多线，可以让你的工作流看起来更简洁 **
Replace two KSamplerAdvanced with a single PainterSampler — same VRAM, same RAM, same frames.But there's no need to create so many node connections – this will make your ComfyUI workflow look much cleaner. <img src="20251117111814.jpg" alt="图1" width="900">

和comfyui官方采样器效果对比Comparison with the Ksampler of the official ComfyUI

<table> <tr> <td><img src="image (2)/01.gif" alt="图1" width="300"></td> <td><img src="image (2)/02.gif" alt="图2" width="300"></td> <td><img src="image (2)/03.gif" alt="图3" width="300"></td> </tr> </table> ---

功能 | Features

• 一步完成 Wan2.1 / Wan2.2 图生视频
  Wan2.1 / Wan2.2 image-to-video in ONE step

• 高噪 + 低噪模型背靠背运行，无需额外采样器
  High-noise + low-noise models run back-to-back, no extra sampler needed

• 像素级对齐官方工作流
  Pixel-perfect match to official workflow

• 零开销，逐帧二进制相同
  Zero overhead, bit-identical frames

---

安装 | Install

1. 克隆到 custom_nodes 文件夹
   Clone into custom_nodes:

cd ComfyUI/custom_nodes
git clone https://github.com/yourname/ComfyUI-PainterSampler.git

2. 重启 ComfyUI，节点出现在 sampling / painter
   Restart ComfyUI, node appears in sampling / painter

---

用法（4 步 Wan2.2 示例）| Usage (4-step Wan2.2)

| 参数 | 中文 | English | |---|---|---| | high_model | 高噪模型（可接 LoRA）| high-noise model (LoRA if needed) | | low_model | 低噪模型（可接 LoRA）| low-noise model (LoRA if needed) | | steps | 4 | 4 | | start_at_step | 0 | 0 | | switch_at_step | 2 | 2 | | end_at_step | 4 | 4 | | return_leftover_noise | disable | disable |

连接方式与 KSamplerAdvanced 完全一致，只需一个节点。
Wire exactly like KSamplerAdvanced — only one node.

---

许可证 | License

MIT – 随意使用。
MIT – use it however you want.