ComfyUI-ImageReward

★ 32

模型加载图像评估自动化评分提示对齐

为 ComfyUI 提供 ImageReward 模型支持：ImageRewardLoader 加载模型，ImageRewardScore 根据提示对图像打分并输出分数，便于自动化筛选与优化。

💡 在生成流程中根据提示自动对候选图像进行打分筛选。

🍴 6 Forks💻 Python🔄 2025-02-24

🔗 GitHub 原文

📦

网盘下载

复制链接后前往夸克网盘下载

https://pan.quark.cn/s/86a6deb1b5f6

📦 requirements.txt

image-reward==1.4
clip
@
git+https://github.com/openai/CLIP.git

📄 README

ComfyUI-ImageReward

Score images with ImageReward.

Installation

Within your ComfyUI directory, and within your Python venv:

cd custom_nodes
git clone https://github.com/ZaneA/ComfyUI-ImageReward
pip install -r ComfyUI-ImageReward/requirements.txt

Usage

Add an ImageRewardLoader node, this has the default model name prefilled, and is passed directly to the ImageReward loader. Note, the model will be downloaded on first run.

Add an ImageRewardScore node, connect the model, your image, and your prompt (either enter this directly, or right click the node and convert prompt to an input first). Connect the SCORE_FLOAT or SCORE_STRING output to an appropriate node.