ALL of the images used in training this style lora was generated through the use of novel ai v3.

This is a new type of training I decided to mess around with after playing with naiv3 for so long, since we are probably never going to get a naiv3 leak 😭😭😭😭😭😭😭

The images produced from the lora seem to be a much higher quality than a lora trained solely on the artists image as well as having better hands and poses in general when compared to the standard local results.

Please do let me know how well this works, as I might remake alot of my older loras using this method

The images used in training as well the metadata for the images can be found in the download