Joey - v1.0 Showcase
 wearing hat, _lora_SC_Joey-07_0.8_, bokeh, motion blur.jpeg)
 and swim trunks , _lora_SC_Joey-07_1_, looking at viewer, smi.jpeg)
This was an interesting piece of work to create. The training data consisted of ~80 images. Most I pulled from Joey's work at Sean Cody.
He only had a few solo pics so the rest I pulled from photoshoots from his duos which meant inpainting out the other models. I also used some pics from his insta, but they weren't that high quality so I ended up removing a lot of those from the final training dataset.
I used both anime and photo captions for the training data and used dataset-tag-editor-standalone to prune the tags. Chatgpt helped me to create a small Python script to combine the anime and photo tags since the colab I use to caption overwrites tags.
I'm happy with the results, but I've also been looking at these pics so much that I'm not sure I'm unbiased. Hopefully, others enjoy Joey and can test out how well this LoRA behaves.