ComfyUI-PainterLTXV2

ComfyUI-PainterLTX2V

此节点由绘画小子制作

Advanced LTXV Sampling & Latent Preparation for ComfyUI

2026-1-19进行了更新。现在支持：文生视频+图生视频+尾帧视频+首尾帧视频+潜空间放大。

不连图片就是文生视频，可单独连首帧或者尾帧，或者首尾帧

PainterLTX2VPlus

<img width="890" height="455" alt="image" src="https://github.com/user-attachments/assets/e5009b8c-f73b-4bfe-9fd6-5eb78f9e2c93" />

全能型 LTXV 视频潜变量生成节点 | All-in-One LTXV Latent Generator

---

带放大功能工作流

<img width="2083" height="1008" alt="image" src="https://github.com/user-attachments/assets/6cb90d03-9cd8-4782-bfa6-38c369eaf09c" />

不带放大功能工作流

<img width="1760" height="1012" alt="image" src="https://github.com/user-attachments/assets/8d9fdbdc-5d45-4204-a9ce-f226278de9bf" />

功能特点 | Features

支持文生视频、图生视频、首尾帧控制与潜在空间放大。
Supports text-to-video, image-to-video, first/last frame control, and latent upscaling.

整合 LTXVLatentUpsampler 与首末帧控制功能于一体。
Combines LTXVLatentUpsampler and first/last frame control in one node.

可接入音频 VAE 生成音视频同步内容。
Supports audio VAE for synchronized audio-video generation.

输出口自动分离视频与音频潜变量，适配双采样器工作流。
Automatically splits video and audio latents for dual-sampler workflows.

返回设定的宽高数值，便于下游节点直接调用。
Returns configured width and height for downstream nodes.

---

输入说明 | Inputs

video_vae: 视频 VAE 模型 | Video VAE model
width/height: 视频分辨率 | Video resolution
length: 视频帧数 | Number of frames
frame_rate: 帧率 | Frame rate
batch_size: 批处理大小 | Batch size
audio_vae (可选): 音频 VAE | Audio VAE (optional)
start_image (可选): 首帧图像 | First frame image (optional)
end_image (可选): 尾帧图像 | Last frame image (optional)
latent (可选): 输入潜变量 | Input latent (optional)
upscale_model (可选): 放大模型 | Upscale model (optional)

---

输出说明 | Outputs

latent: 主潜变量 | Main latent
video_latent: 视频潜变量 | Video latent
audio_latent: 音频潜变量 | Audio latent
width: 设定宽度 | Configured width
height: 设定高度 | Configured height

---

使用场景 | Use Cases

仅连始帧 = 图生视频基础模式
Start-frame only = Basic image-to-video mode

首尾帧同连 = 视频插值与过渡
Both frames = Video interpolation and transition

接入 latent = 潜变量续写与重采样
With latent = Latent continuation and resampling

叠加放大模型 = 高清化预处理
Plus upscale model = High-resolution preprocessing

---

一句话介绍 | One-Liner

Generate LTXV latents with frame control and upscaling for advanced video synthesis.

Nodes | 节点功能

PainterSamplerLTXV

<img width="1876" height="899" alt="image" src="https://github.com/user-attachments/assets/6b2c1111-3911-40a2-ae46-54ec8a29bc42" />

[English]

• Full KSampler parameters with LTXV native audio-video latent separation
• External sigmas input support (overrides scheduler when connected)
• Flexible step control for partial denoising workflows
• Optimized for dynamic preservation and color fidelity

[中文]

• 完整的 KSampler 参数支持，原生 LTXV 音视频潜空间分离输出
• 外部 sigmas 输入支持（连接时覆盖调度器）
• 灵活的步骤控制，支持局部去噪工作流
• 针对动态效果保留和色彩保真优化

Installation | 安装方法

[English]

1. Clone this repository into ComfyUI/custom_nodes/
2. Restart ComfyUI
3. Nodes will appear in sampling and latent/video/ltxv categories

cd ComfyUI/custom_nodes/
git clone https://github.com/princepainter/ComfyUI-PainterLTXV2.git

[中文]

1. 将本仓库克隆到 ComfyUI/custom_nodes/ 目录
2. 重启 ComfyUI
3. 节点将出现在 sampling 和 latent/video/ltxv 分类中

cd ComfyUI/custom_nodes/
git clone https://github.com/princepainter/ComfyUI-PainterLTXV2.git

Usage | 使用说明

[English]
Connect PainterLTXVtoVideo to prepare latents, then feed into PainterSamplerLTXV for generation. Use external sigmas for custom sampling schedules.

[中文]
将 PainterLTXVtoVideo 连接到 PainterSamplerLTXV 进行生成。可使用外部 sigmas 实现自定义采样调度。

Requirements | 系统要求

[English]

• ComfyUI with comfy_api support
• LTXV model files
• Optional: Audio VAE for sound generation

[中文]

• 支持 comfy_api 的 ComfyUI 环境
• LTXV 模型文件
• 可选：Audio VAE 用于声音生成

Note | 注意事项

[English] Designed for professional workflows requiring pixel-perfect color fidelity and dynamic preservation. All parameters exposed for fine-grained control.

[中文] 专为需要像素级色彩保真和动态保留的专业工作流设计，所有参数开放用于精细控制。

看到这里了，请给我点颗星星，谢谢！ If you find this helpful, please star the repository. Thank you!