From 9248df314b176a2ea8078c6433fed5ec80905733 Mon Sep 17 00:00:00 2001 From: coderfromthenorth93 <213232275+coderfromthenorth93@users.noreply.github.com> Date: Tue, 25 Nov 2025 13:47:59 -0500 Subject: [PATCH 1/7] Add/update changelog/index.mdx --- changelog/index.mdx | 29 ++++++++++++++++++++++++++++- 1 file changed, 28 insertions(+), 1 deletion(-) diff --git a/changelog/index.mdx b/changelog/index.mdx index e0beacc7..bd1ab5cb 100644 --- a/changelog/index.mdx +++ b/changelog/index.mdx @@ -4,6 +4,34 @@ description: "Track ComfyUI's latest features, improvements, and bug fixes. For icon: "clock-rotate-left" --- + + +**Major Model Support** +- **Flux 2 Integration**: Added comprehensive support for Flux 2 models including dedicated API nodes for Flux 2 Pro, expanding high-quality image generation workflows +- **Enhanced HunyuanVideo Workflows**: Updated workflow templates with new HunyuanVideo and Nano Banana Pro variants, plus improved display names for better node organization + +**API & Performance Improvements** +- **Enhanced API Security**: The `--disable-api-nodes` flag now enforces Content Security Policy headers to ensure secure offline frontend operation +- **Fixed Non-Async API Access**: Resolved critical issues with public API access that affected certain workflow integrations +- **Improved Gemini Model Support**: Fixed edge cases in API node responses for better reliability when working with Gemini models + +**Video Processing Enhancements** +- Added `get_frame_count` and `get_frame_rate` methods to video processing classes, enabling more precise video workflow control and analysis + +**Text Encoder & Quantization Updates** +- **Optimized Text Encoder Quantization**: Cleaned up and resolved issues with quantized text encoders for improved performance and stability +- **Hardware-Aware Quantization**: Added intelligent detection to prevent unsupported FP8 matrix operations on incompatible hardware +- **Enhanced Quantization Support**: Improved support for pinning quantized tensors with better memory management + +**Block Information Support** +- Added BlockInfo support for Chroma, Qwen-Image, and HunyuanVideo models, providing better model introspection and debugging capabilities + +**Frontend & Dependencies** +- Updated frontend to version 1.30.6 with improved user interface elements +- Bumped Transformers library version for enhanced model compatibility and performance + + + **Model Compatibility & Enhancements** @@ -68,7 +96,6 @@ icon: "clock-rotate-left" - Unified RoPE function implementation across models - **Performance & Memory Optimizations** From ddac81216e62765c8147b69a5b25693d9801e049 Mon Sep 17 00:00:00 2001 From: coderfromthenorth93 <213232275+coderfromthenorth93@users.noreply.github.com> Date: Tue, 25 Nov 2025 13:48:00 -0500 Subject: [PATCH 2/7] Add/update zh-CN/changelog/index.mdx --- zh-CN/changelog/index.mdx | 29 ++++++++++++++++++++++++++++- 1 file changed, 28 insertions(+), 1 deletion(-) diff --git a/zh-CN/changelog/index.mdx b/zh-CN/changelog/index.mdx index 3d4285f9..f1b81ba4 100644 --- a/zh-CN/changelog/index.mdx +++ b/zh-CN/changelog/index.mdx @@ -1,9 +1,36 @@ --- title: "更新日志" -description: "跟踪 ComfyUI 的最新功能、改进和错误修复。如需详细发布说明，请查看 [Github releases](https://github.com/comfyanonymous/ComfyUI/releases) 页面。" +description: "跟踪 ComfyUI 的最新功能、改进和错误修复。详细发布说明请查看 [Github releases](https://github.com/comfyanonymous/ComfyUI/releases) 页面。" icon: "clock-rotate-left" --- + + +**重大模型支持** +- **Flux 2 集成**：新增对 Flux 2 模型的全面支持，包括专用的 Flux 2 Pro API 节点，扩展了高质量图像生成工作流程 +- **增强的 HunyuanVideo 工作流**：更新了工作流模板，包含新的 HunyuanVideo 和 Nano Banana Pro 变体，并改进了显示名称以便更好地组织节点 + +**API 和性能改进** +- **增强的 API 安全性**：`--disable-api-nodes` 标志现在强制执行内容安全策略标头，确保离线前端操作的安全性 +- **修复非异步 API 访问**：解决了影响某些工作流集成的公共 API 访问关键问题 +- **改进的 Gemini 模型支持**：修复了 API 节点响应中的边缘情况，提高了使用 Gemini 模型时的可靠性 + +**视频处理增强** +- 为视频处理类添加了 `get_frame_count` 和 `get_frame_rate` 方法，实现更精确的视频工作流控制和分析 + +**文本编码器和量化更新** +- **优化的文本编码器量化**：清理并解决了量化文本编码器的问题，提高了性能和稳定性 +- **硬件感知量化**：添加了智能检测功能，防止在不兼容硬件上进行不支持的 FP8 矩阵操作 +- **增强的量化支持**：改进了对固定量化张量的支持，具有更好的内存管理 + +**块信息支持** +- 为 Chroma、Qwen-Image 和 HunyuanVideo 模型添加了 BlockInfo 支持，提供更好的模型自省和调试功能 + +**前端和依赖项** +- 将前端更新至 1.30.6 版本，改进了用户界面元素 +- 升级了 Transformers 库版本，以增强模型兼容性和性能 + + **模型兼容性与增强功能** From c00c75411ca342e9b034cfdeba85bdcdd3844904 Mon Sep 17 00:00:00 2001 From: coderfromthenorth93 <213232275+coderfromthenorth93@users.noreply.github.com> Date: Tue, 25 Nov 2025 19:06:16 -0500 Subject: [PATCH 3/7] Add/update changelog/index.mdx --- changelog/index.mdx | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/changelog/index.mdx b/changelog/index.mdx index e0beacc7..3de2b8cc 100644 --- a/changelog/index.mdx +++ b/changelog/index.mdx @@ -4,6 +4,15 @@ description: "Track ComfyUI's latest features, improvements, and bug fixes. For icon: "clock-rotate-left" --- + + +**Stability & Performance Fixes** +- **Critical Crash Fix**: Resolved system stability issues that could interrupt workflow execution +- **Flux 2 VRAM Optimization**: Reduced memory usage for Flux 2 text encoder, allowing for more efficient processing of larger models and batch operations +- **Workflow Template Updates**: Updated to v0.7.20, providing users with improved starter templates and example workflows for better onboarding experience + + + **Model Compatibility & Enhancements** @@ -68,7 +77,6 @@ icon: "clock-rotate-left" - Unified RoPE function implementation across models - **Performance & Memory Optimizations** From 3d1e03752db03a3efb50df1c4a66fe281e95cc9c Mon Sep 17 00:00:00 2001 From: coderfromthenorth93 <213232275+coderfromthenorth93@users.noreply.github.com> Date: Tue, 25 Nov 2025 19:06:17 -0500 Subject: [PATCH 4/7] Add/update zh-CN/changelog/index.mdx --- zh-CN/changelog/index.mdx | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/zh-CN/changelog/index.mdx b/zh-CN/changelog/index.mdx index 3d4285f9..e1c66616 100644 --- a/zh-CN/changelog/index.mdx +++ b/zh-CN/changelog/index.mdx @@ -1,9 +1,17 @@ --- title: "更新日志" -description: "跟踪 ComfyUI 的最新功能、改进和错误修复。如需详细发布说明，请查看 [Github releases](https://github.com/comfyanonymous/ComfyUI/releases) 页面。" +description: "追踪 ComfyUI 的最新功能、改进和错误修复。查看详细发布说明，请访问 [Github 发布页面](https://github.com/comfyanonymous/ComfyUI/releases)。" icon: "clock-rotate-left" --- + + +**稳定性和性能修复** +- **关键崩溃修复**：解决了可能中断工作流执行的系统稳定性问题 +- **Flux 2 显存优化**：减少了 Flux 2 文本编码器的内存使用量，允许更高效地处理大型模型和批处理操作 +- **工作流模板更新**：更新至 v0.7.20，为用户提供改进的入门模板和示例工作流，提升用户初始体验 + + **模型兼容性与增强功能** From 21526b5d10bfcdc10dbbba2726617bd1f93824e1 Mon Sep 17 00:00:00 2001 From: coderfromthenorth93 <213232275+coderfromthenorth93@users.noreply.github.com> Date: Wed, 26 Nov 2025 03:55:48 -0500 Subject: [PATCH 5/7] Add/update changelog/index.mdx --- changelog/index.mdx | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/changelog/index.mdx b/changelog/index.mdx index e0beacc7..4886a92d 100644 --- a/changelog/index.mdx +++ b/changelog/index.mdx @@ -4,6 +4,19 @@ description: "Track ComfyUI's latest features, improvements, and bug fixes. For icon: "clock-rotate-left" --- + + +**Z Image Model Integration** +- **New Z Image Model Support**: Added Z Image model implementation with adjustments for optimized performance in image processing workflows. This expands ComfyUI's model compatibility for specialized image generation and manipulation tasks. + +**LoRA & Mixed Precision Improvements** +- **Fixed LoRA Functionality**: Resolved critical issue preventing LoRAs from working properly with mixed FP8 precision models, ensuring consistent fine-tuning capabilities across different precision configurations. + +**Flux2 Memory Optimization** +- **Enhanced Reference Image Memory Estimation**: Fixed memory estimation for Flux2 reference images, improving VRAM efficiency and reducing out-of-memory errors during reference-guided generation workflows. + + + **Model Compatibility & Enhancements** @@ -68,7 +81,6 @@ icon: "clock-rotate-left" - Unified RoPE function implementation across models - **Performance & Memory Optimizations** From c58ff6dd9e3cbfe008e70ad943b5c682abe12042 Mon Sep 17 00:00:00 2001 From: coderfromthenorth93 <213232275+coderfromthenorth93@users.noreply.github.com> Date: Wed, 26 Nov 2025 03:55:49 -0500 Subject: [PATCH 6/7] Add/update zh-CN/changelog/index.mdx --- zh-CN/changelog/index.mdx | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/zh-CN/changelog/index.mdx b/zh-CN/changelog/index.mdx index 3d4285f9..fa46aadb 100644 --- a/zh-CN/changelog/index.mdx +++ b/zh-CN/changelog/index.mdx @@ -1,9 +1,21 @@ --- title: "更新日志" -description: "跟踪 ComfyUI 的最新功能、改进和错误修复。如需详细发布说明，请查看 [Github releases](https://github.com/comfyanonymous/ComfyUI/releases) 页面。" +description: "跟踪 ComfyUI 的最新功能、改进和错误修复。详细发布说明请查看 [Github 发布页面](https://github.com/comfyanonymous/ComfyUI/releases)。" icon: "clock-rotate-left" --- + + +**Z Image 模型集成** +- **新增 Z Image 模型支持**：添加了 Z Image 模型实现，优化了图像处理工作流的性能。这扩展了 ComfyUI 在专业图像生成和处理任务方面的模型兼容性。 + +**LoRA 和混合精度改进** +- **修复 LoRA 功能**：解决了阻止 LoRA 在混合 FP8 精度模型下正常工作的关键问题，确保在不同精度配置下微调功能的一致性。 + +**Flux2 内存优化** +- **增强参考图像内存估算**：修复了 Flux2 参考图像的内存估算问题，提高了 VRAM 效率并减少了在参考引导生成工作流中出现的内存不足错误。 + + **模型兼容性与增强功能** From c02eee438187d42ea15df49c1f82495539f5ac37 Mon Sep 17 00:00:00 2001 From: ComfyUI Wiki Date: Wed, 26 Nov 2025 18:24:05 +0800 Subject: [PATCH 7/7] Update Changelog --- changelog/index.mdx | 51 +++++++++++++++++-------------------- zh-CN/changelog/index.mdx | 53 ++++++++++++++++++--------------------- 2 files changed, 48 insertions(+), 56 deletions(-) diff --git a/changelog/index.mdx b/changelog/index.mdx index d87ee9bd..0c771241 100644 --- a/changelog/index.mdx +++ b/changelog/index.mdx @@ -6,51 +6,46 @@ icon: "clock-rotate-left" -**Z Image Model Integration** -- **New Z Image Model Support**: Added Z Image model implementation with adjustments for optimized performance in image processing workflows. This expands ComfyUI's model compatibility for specialized image generation and manipulation tasks. - -**LoRA & Mixed Precision Improvements** -- **Fixed LoRA Functionality**: Resolved critical issue preventing LoRAs from working properly with mixed FP8 precision models, ensuring consistent fine-tuning capabilities across different precision configurations. +**New Model Support** +- Added Z Image model with optimized performance for image processing workflows -**Flux2 Memory Optimization** -- **Enhanced Reference Image Memory Estimation**: Fixed memory estimation for Flux2 reference images, improving VRAM efficiency and reducing out-of-memory errors during reference-guided generation workflows. +**Bug Fixes** +- Fixed LoRA functionality with mixed FP8 precision models +- Enhanced Flux2 reference image memory estimation to reduce VRAM errors -**Stability & Performance Fixes** -- **Critical Crash Fix**: Resolved system stability issues that could interrupt workflow execution -- **Flux 2 VRAM Optimization**: Reduced memory usage for Flux 2 text encoder, allowing for more efficient processing of larger models and batch operations -- **Workflow Template Updates**: Updated to v0.7.20, providing users with improved starter templates and example workflows for better onboarding experience +**Bug Fixes** +- Fixed critical system crashes that could interrupt workflow execution +- Optimized Flux 2 text encoder VRAM usage for better performance -**Major Model Support** -- **Flux 2 Integration**: Added comprehensive support for Flux 2 models including dedicated API nodes for Flux 2 Pro, expanding high-quality image generation workflows -- **Enhanced HunyuanVideo Workflows**: Updated workflow templates with new HunyuanVideo and Nano Banana Pro variants, plus improved display names for better node organization +**New Model Support** +- Added comprehensive Flux 2 model support including Flux 2 Pro API nodes +- Enhanced HunyuanVideo workflows with new variants and improved display names -**API & Performance Improvements** -- **Enhanced API Security**: The `--disable-api-nodes` flag now enforces Content Security Policy headers to ensure secure offline frontend operation -- **Fixed Non-Async API Access**: Resolved critical issues with public API access that affected certain workflow integrations -- **Improved Gemini Model Support**: Fixed edge cases in API node responses for better reliability when working with Gemini models +**API & Security Improvements** +- Improved API security with Content Security Policy headers +- Fixed public API access issues and Gemini model support -**Video Processing Enhancements** -- Added `get_frame_count` and `get_frame_rate` methods to video processing classes, enabling more precise video workflow control and analysis +**Video Processing** +- Added `get_frame_count` and `get_frame_rate` methods for better video workflow control -**Text Encoder & Quantization Updates** -- **Optimized Text Encoder Quantization**: Cleaned up and resolved issues with quantized text encoders for improved performance and stability -- **Hardware-Aware Quantization**: Added intelligent detection to prevent unsupported FP8 matrix operations on incompatible hardware -- **Enhanced Quantization Support**: Improved support for pinning quantized tensors with better memory management +**Performance Optimizations** +- Optimized text encoder quantization with hardware-aware detection +- Enhanced quantization support with better memory management -**Block Information Support** -- Added BlockInfo support for Chroma, Qwen-Image, and HunyuanVideo models, providing better model introspection and debugging capabilities +**Developer Tools** +- Added BlockInfo support for Chroma, Qwen-Image, and HunyuanVideo models **Frontend & Dependencies** -- Updated frontend to version 1.30.6 with improved user interface elements -- Bumped Transformers library version for enhanced model compatibility and performance +- Updated frontend to v1.30.6 +- Bumped Transformers version diff --git a/zh-CN/changelog/index.mdx b/zh-CN/changelog/index.mdx index 5a347df8..3bcddfc7 100644 --- a/zh-CN/changelog/index.mdx +++ b/zh-CN/changelog/index.mdx @@ -6,51 +6,48 @@ icon: "clock-rotate-left" -**Z Image 模型集成** -- **新增 Z Image 模型支持**：添加了 Z Image 模型实现，优化了图像处理工作流的性能。这扩展了 ComfyUI 在专业图像生成和处理任务方面的模型兼容性。 - -**LoRA 和混合精度改进** -- **修复 LoRA 功能**：解决了阻止 LoRA 在混合 FP8 精度模型下正常工作的关键问题，确保在不同精度配置下微调功能的一致性。 +**新模型支持** +- 添加 Z Image 模型支持，优化图像处理工作流性能 -**Flux2 内存优化** -- **增强参考图像内存估算**：修复了 Flux2 参考图像的内存估算问题，提高了 VRAM 效率并减少了在参考引导生成工作流中出现的内存不足错误。 +**错误修复** +- 修复 LoRA 在混合 FP8 精度模型下的功能问题 +- 增强 Flux2 参考图像内存估算，减少显存错误 -**稳定性和性能修复** -- **关键崩溃修复**：解决了可能中断工作流执行的系统稳定性问题 -- **Flux 2 显存优化**：减少了 Flux 2 文本编码器的内存使用量，允许更高效地处理大型模型和批处理操作 -- **工作流模板更新**：更新至 v0.7.20，为用户提供改进的入门模板和示例工作流，提升用户初始体验 +**错误修复** +- 修复可能中断工作流执行的关键系统崩溃问题 +- 优化 Flux 2 文本编码器显存使用，提升性能 + +**模板更新** +- 更新工作流模板至 v0.7.20，提供更好的入门示例 -**重大模型支持** -- **Flux 2 集成**：新增对 Flux 2 模型的全面支持，包括专用的 Flux 2 Pro API 节点，扩展了高质量图像生成工作流程 -- **增强的 HunyuanVideo 工作流**：更新了工作流模板，包含新的 HunyuanVideo 和 Nano Banana Pro 变体，并改进了显示名称以便更好地组织节点 +**新模型支持** +- 新增 Flux 2 模型全面支持，包括 Flux 2 Pro API 节点 +- 增强 HunyuanVideo 工作流，添加新变体和改进显示名称 -**API 和性能改进** -- **增强的 API 安全性**：`--disable-api-nodes` 标志现在强制执行内容安全策略标头，确保离线前端操作的安全性 -- **修复非异步 API 访问**：解决了影响某些工作流集成的公共 API 访问关键问题 -- **改进的 Gemini 模型支持**：修复了 API 节点响应中的边缘情况，提高了使用 Gemini 模型时的可靠性 +**API 和安全改进** +- 改进 API 安全性，添加内容安全策略标头 +- 修复公共 API 访问问题和 Gemini 模型支持 -**视频处理增强** -- 为视频处理类添加了 `get_frame_count` 和 `get_frame_rate` 方法，实现更精确的视频工作流控制和分析 +**视频处理** +- 添加视频处理方法：`get_frame_count` 和 `get_frame_rate`，实现更好的工作流控制 -**文本编码器和量化更新** -- **优化的文本编码器量化**：清理并解决了量化文本编码器的问题，提高了性能和稳定性 -- **硬件感知量化**：添加了智能检测功能，防止在不兼容硬件上进行不支持的 FP8 矩阵操作 -- **增强的量化支持**：改进了对固定量化张量的支持，具有更好的内存管理 +**性能优化** +- 优化文本编码器量化，支持硬件感知检测 +- 增强量化支持，改善内存管理 -**块信息支持** -- 为 Chroma、Qwen-Image 和 HunyuanVideo 模型添加了 BlockInfo 支持，提供更好的模型自省和调试功能 +**开发者工具** +- 为 Chroma、Qwen-Image 和 HunyuanVideo 模型添加 BlockInfo 支持 **前端和依赖项** -- 将前端更新至 1.30.6 版本，改进了用户界面元素 -- 升级了 Transformers 库版本，以增强模型兼容性和性能 +- 更新前端至 v1.30.6 并升级 Transformers 库