Download over 2K before updating the next version.
[if no one uses means my checkpoint is not welcomed, thus I don't need to waste money and time on checkpoint.]
Advanced Version is a V1.0+, There will be more contrast and detail.
Trained on Over 2G of pictures carefully selected by PAseer and manually labelled.
Pursuit of generalisability and soft switching between different styles.
If you don't know what to use as a base model, use this.
Description
V1
FAQ
Comments (5)
Nice Work！
i tried a lots of checkpoints the last week and this one is in my top 5. thank you!!!
Hello! Was the training made on generated images? Or did you used photographs and art scans also? Thanks! :)
generated images need to be fixed by PS (for imagination and unreal concepts---the more will influence the quality, so keep balance), to make the anatomy become better. original photographs and free copyright art scans should at least 2K. Try to avoid using PS to modify these pictures, and the light should be a bit stronger than studio shooting. contrast needs to be clearer, because AI learns a kind of diffusion of f(x), or I would like to say pixel position relationship. Therefore, the higher the quality of single pixels and the clearer the pixel position structure, the better the predictable results will be.
@Aseer thanks!