Model is good, generate face most accurate compared to other model, can make that shadow over eyes, only downside is that the model don't understand how dress and wings interact, also pixel art sometimes generate text and chibi models. If you want I can train another lora but i think this is fine for nowSuggested tag: osalula, 1girl, solo, solid white eyes, shadow over eyes, blue hair, multi eye, yellow beak on forehead, dark skin, feathers on forehead, blue feathers, blue winged arms, black humanoid hands, black bird legs, blue dress, symbol on dress

Alula Oneshot

This version is better at face, better at clothes, better at literally everything.Recommended pixel art or no lineart because dataset has those

Files with SHA256: eed0c9eae937df206c7451327cb7e590e3b9785372db5d77f118939e8badd689