ComfyUI_AgnesAI_Nodes
Agnes.ai nodes and example workflows for text-to-image and text-to-video generation in ComfyUI. (Description by CC)
Nodes (5)
The Key That Opens the Rest of the Agnes Nodes
Animate Any Image by Handing Its URL to Agnes
Your Local Image, Uploaded, Because the Cloud Can't See Your Hard Drive
Agnes in a ComfyUI Node
Text to Video Without the 20GB Model Download
ComfyUI_AgnesAI_Nodes
Agnes.ai nodes and example workflows for ComfyUI.
Agnes.ai 的 ComfyUI 节点与示例工作流仓库。
Overview / 概览
This repository includes:
本仓库包含:
Agnes_ConfigAgnes_Text2ImageAgnes_Text2VideoAgnes_ImageToURLAgnes_Image2Video
It also ships example workflows for:
同时附带以下示例工作流:
- text to image / 文生图
- text to video with in-app preview / 文生视频并在 ComfyUI 内预览
- local image to video with automatic upload and in-app preview / 本地图生视频并自动上传图片、在 ComfyUI 内预览
Features / 功能
- Agnes image generation with
agnes-image-2.1-flash
使用agnes-image-2.1-flash进行 Agnes 文生图 - Agnes video generation with
agnes-video-v2.0
使用agnes-video-v2.0进行 Agnes 生视频 - ComfyUI-friendly video outputs: frame batch + audio + source URL
适配 ComfyUI 的视频输出:帧序列 + 音频 + 源视频 URL - Preview-ready video workflows via
VHS_VideoCombine
通过VHS_VideoCombine直接在 ComfyUI 里预览视频 - Local image to public URL conversion for Agnes image-to-video
支持把 ComfyUI 本地图片转成公网 URL,再交给 Agnes 图生视频
Installation / 安装
cd ~/ComfyUI/custom_nodes
git clone https://github.com/Watchcats211/ComfyUI_AgnesAI_Nodes.git
Restart ComfyUI after cloning.
克隆完成后,重启 ComfyUI。
Configuration / 配置
Use Agnes_Config in your workflow.
在工作流中使用 Agnes_Config 节点。
api_key: your Agnes key
api_key:你的 Agnes API Keybase_url: defaults tohttps://apihub.agnes-ai.com/v1
base_url:默认是https://apihub.agnes-ai.com/v1
If you leave api_key blank, the node will try AGNES_API_KEY from the environment.
如果把 api_key 留空,节点会尝试从环境变量 AGNES_API_KEY 读取。
Included Workflows / 内置工作流
workflows/agnes_text2image_workflow.json
- Minimal text-to-image example
最小文生图示例
workflows/agnes_text2video_workflow.json
- Agnes text-to-video
Agnes 文生视频 - Saves the original Agnes mp4 into ComfyUI
output/
会把 Agnes 原始 mp4 保存到 ComfyUIoutput/目录 - Also returns a ComfyUI preview video via
VHS_VideoCombine
同时通过VHS_VideoCombine返回 ComfyUI 内可播放预览
workflows/agnes_image2video_workflow.json
- Load a local image in ComfyUI
在 ComfyUI 中载入本地图片 - Upload it through
Agnes_ImageToURL
通过Agnes_ImageToURL自动上传 - Send the public URL to Agnes image-to-video
把公网图片 URL 交给 Agnes 图生视频接口 - Return a preview video in ComfyUI
在 ComfyUI 内返回可播放预览
Example Media / 示例素材
examples/videos/text_to_video_original.mp4
- Original Agnes text-to-video output saved into ComfyUI
output/
Agnes 文生视频原始输出,来自 ComfyUIoutput/
examples/videos/text_to_video_comfy_preview.mp4
- The corresponding ComfyUI in-app preview video produced through
VHS_VideoCombine
对应的 ComfyUI 内预览视频,由VHS_VideoCombine生成
examples/videos/image_to_video_original.mp4
- Original Agnes image-to-video output saved into ComfyUI
output/
Agnes 图生视频原始输出,来自 ComfyUIoutput/
examples/videos/image_to_video_comfy_preview.mp4
- The corresponding ComfyUI in-app preview video produced through
VHS_VideoCombine
对应的 ComfyUI 内预览视频,由VHS_VideoCombine生成
Image-to-Video Notes / 图生视频说明
Agnes image-to-video expects a public image URL.
Agnes 图生视频接口要求输入公网可访问的图片 URL。
This repo includes Agnes_ImageToURL, which uploads a ComfyUI IMAGE to storage.to and returns a URL that Agnes can read.
本仓库提供了 Agnes_ImageToURL,会把 ComfyUI 的 IMAGE 上传到 storage.to,再返回 Agnes 可读取的 URL。
The sample image at examples/agnes_i2v_source.png is only for reference. After importing the workflow, pick your own image in ComfyUI LoadImage.
examples/agnes_i2v_source.png 只是示例图。导入工作流后,你可以在 ComfyUI 的 LoadImage 节点里换成自己的图片。
Outputs / 输出
Agnes_Text2Image
IMAGESTRINGurl
Agnes_Text2Video
IMAGEframes / 帧序列AUDIOSTRINGurl
Agnes_Image2Video
IMAGEframes / 帧序列AUDIOSTRINGurl
Dependencies / 依赖
This node relies on packages typically already present in a working ComfyUI install:
这些依赖通常在正常可用的 ComfyUI 环境里已经存在:
torchnumpyPillowopencv-python
Optional fallback for audio extraction:
音频提取的可选兜底依赖:
imageio-ffmpeg
For video preview workflows you also need:
如果要使用视频预览工作流,还需要:
ComfyUI-VideoHelperSuite
Security / Release Workflow / 安全与发布流程
Before publishing to GitHub, run:
发布到 GitHub 之前,建议先运行:
python3 scripts/sanitize_release.py --write .
python3 scripts/sanitize_release.py --check .
What it does:
这个脚本会做:
- clears Agnes API keys from workflow JSON
清理工作流 JSON 中的 Agnes API key - clears preview metadata that contains local paths
清理包含本机路径的预览元数据 - normalizes
LoadImageplaceholders
统一LoadImage的占位文件名 - scans the repo for likely secrets before release
在发布前扫描仓库中的疑似敏感信息
Recommended Release Checklist / 推荐发布检查清单
- Run the sanitize script in
--writemode.
先运行--write模式做自动清理。 - Run the sanitize script in
--checkmode.
再运行--check模式做检查。 - Open the workflow JSONs and confirm
api_keyis blank.
打开工作流 JSON,确认api_key是空的。 - Confirm no
sk-...key remains anywhere in the repo.
确认仓库里不再残留任何sk-...密钥。 - Then commit and push.
最后再提交并推送。