Beth from The Legend of Zelda: Twilight Princess for PonyXL. Trigger phrase is 'beth, tpb3th_clothing'. The trigger phrase 'video game screenshot' did not converge as I wanted during training but may still be helpful. See provided example images for usage.
This model was trained in an experimental way due to limited available training images. A higher volume of mediocre/low quality images was used in place of more ideal dataset such as a smaller amount of high quality images. Screenshots from the video game were used to fill in gaps in the data and regularization images were used in an attempt to prevent the model from converging on the low poly in-game 3d style. I think this experiment went reasonably well. I'd love any feedback.