Credits: https://civitai.com/user/iLab
This is a Flux.1 Dev Lora model that can greatly enhance the aesthetic quality of the image. For example, the surface of animals generated by the original Flux base model can look like plastic toys, and this model can enhance the realism of their appearance. The training set images do not contain portraits, but for generating portraits of real people this model can enhance the aesthetic lines and compositions of the picture, and also have the same effect on anime characters.
Using weights from 0.3-1.0 can be tried. Example images are generated using Flux.0 Dev FP8+lora, images include embedded full comfyui workflow, grid map weights are 0, 0.6, 0.8, 1 respectively.
Please use this model within the protocol of the original Flux Dev model.
Thank you for downloading and using it, if you have any nice looking diagrams please post them to here, much appreciated.
In addition, a good friend of mine, Surenjike, has made an enhanced aesthetic workflow based on this model, for details on how to use it and download the workflow, please refer to the link: https://civitai.com/articles/9372/we-have-broken-through-the-aesthetic-limitations-of-flux-devflux-dev
这是一个能极大程度提升画面美学质感的Flux.1 Dev Lora模型，比如Flux原底模生成的动物表面会显得很像塑料玩具，加上这个模型可以增强其外观真实度。训练集图像不包含人像，但是对于生成真实人物肖像本模型可以加强画面美学线条及构图，同时对动漫角色也有同样效果。
使用权重从0.3-1.0都可以尝试。例图都是使用Flux.0 Dev FP8+lora生成，图片包含内嵌完整comfyui工作流，网格图权重分别为0，0.6，0.8，1。
请在Flux Dev原模型的协议内使用本模型。
感谢您的下载和使用，如果有好看的图请给我返图，非常感谢。
另外我的一位好朋友素人极客在这模型基础上做了一个加强的美学工作流，具体使用方法和工作流下载可以参考链接：https://civitai.com/articles/9372/we-have-broken-through-the-aesthetic-limitations-of-flux-devflux-dev