ERNIE Image Model

Compatible with both base and turbo models. Trained on Ostris AI Toolkit. Used same dataset as previous Loras. This was the first one I trained, so it's experimental. Let me know how it does.

Flux Model

Use weight of 0.90 for best results (atleast that's the weight that gets me the best results).

Images trained on a subset of images of official promo art, trailers and gameplay screenshots.

SD1.5 Model

Use weight of around 0.67 for best results.

LoRA trained on SD 1.5.

Images trained on a subset of images of official promo art, trailers and gameplay screenshots. Unfortunately, not much variety. This causes the necklace appearance in almost all generated images.

Images generated using GalaxyTimeMachine Photo model. However, also works well with DreamShaper and chilloutmix.