UPDATE: Added SDXL version trained on same dataset. Much more versatile than the 1.5 version.(Gallery images are placeholder previews generated during training so I can use the model for generation and make better ones.)Testing Civitai's FLUX training capabilities by updating <a target="_blank" rel="ugc" href="https://civarchive.com/models/183131/reina-tanakas-graduation-dress">https://civarchive.com/models/183131/reina-tanakas-graduation-dress</a>. I used the same 36 screenshots, but with new captions in natural language. As before, the intent of the LoRA is to reproduce the outfit while allowing freedom to prompt for someone else wearing it. If no specific instructions on the appearance are given in the prompt, the subject will look like Reina Tanaka.

Reina's Graduation Dress

Files with SHA256: e121ad1dbd26bccea7780066f75e8049edf1edb11e1f501e85a781ad390565ba