Although there are already quite a few felt models available, I find that they are either too cute or do not have a strong enough felt texture. The colors are also not to my liking. Therefore, I aim to train a felt model with a fresh color palette that strikes a balance between cartoonish and realistic.

Both text-to-image and image-to-image generation have shown excellent results, which can be further enhanced when used in conjunction with Canny edge detection, providing greater control over the output.

虽然目前已有不少毛毡模型，但我觉得要么过于可爱，要么毛毡感不强。颜色也不是我喜欢的 。故训练一个颜色清新，介于卡通与写实之间的毛毡模型。

文生图和图生图都有着不错的效果，可以搭配canny使用，获得更强的控制力。