For Version 2, I processed the source assets—fixing issues like black borders around text and upscaling/redrawing blurry material into high definition. This effectively resolves the shortcomings of the initial, rough LoRA.

You can try using the trigger word illuxxxtrandy_style; while the model works fine without it, including it is a good idea to avoid any unforeseen issues.

All the images I generated used the "One Obsession" model; if you want your results to closely match mine, I recommend using that model as well.