Victorian Gothic Horror - V4.0 API Documentation

Version 4.0 was trained on an expanded selection of horror imagery, using a more efficient training technique that worked well. The result is a more robust style that includes the framing, tone, and medium represented in the training dataset. The LoRA no longer needs to be prompted with specific keywords like sepia or photograph, and it will now apply an appropriate style fitting the prompt. I've found it works when you let the model choose by not specifying anything.