Did you write it wrong
#1
by
demo001s
- opened
Thank you for pointing out!
哈哈
我发现这个模型在人物方面还可以，但是试了几张室内的图，不太好用。
wanghaofan
changed discussion status to
closed
wanghaofan
changed discussion status to
open
Yes, it should work better on human generation. This is because of the limited diversity of training data, we will continue train on more data.
there it is.
https://huggingface.co/datasets/kadirnar/fluxdev_controlnet_16k
Wow, I have just had a look at these images, just for curiosity. Are both the images and the text automatically generated? I only read 4 of them, and they are all wrong one way or another. I am not sure I'd use these for training.
这个是别的人， 我看了一些，基本上没有描述错的吧。不知道你说的哪些。