资源合集：闲话AI官网 talk2ai.cn

Realistic Photo to Manga Style（ https://openart.ai/workflows/monkey_well-worn_31/realistic-photo-to-manga-style/Cl3xyeUX0EVQXY4bVnPN ）的一个修改版本，将肖像大师替换成了图像输入节点，以方便用户使用自己的图片来生成漫画图像。

漫画风格使用了自训练的Lora模型：残照金 SDXL - Hoko Manga Style，可以在civitai下载（  https://civitai.com/models/647408/zanshou-kin-sdxl-hoko-manga-style  ）

生成图像的宽和高默认使用手动输入，因为模型在单边低于768的情况下生成效果不佳，也可以直接获取原始图片的宽和高来进行生成。

This is a modified version of the "Realistic Photo to Manga Style" workflow (https://openart.ai/workflows/monkey_well-worn_31/realistic-photo-to-manga-style/Cl3xyeUX0EVQXY4bVnPN). The comfyui-portrait-master node has been replaced with Load Image node to allow users to easily use their own images for manga-style conversion.

The manga style uses a custom-trained Lora model: Zanshou-kin(残照金) SDXL - Hoko Manga Style, which can be downloaded from Civitai (https://civitai.com/models/647408/zanshou-kin-sdxl-hoko-manga-style).

The width and height for generating images are set to manual input by default, as the model performs poorly when either dimension is below 768 pixels. Alternatively, you can directly use the original image's width and height for generation.