Style: Drawn / Anime | Character-Focused | Strong Posing

Hard Mix LoRA is an experimental build designed for anime-style character generation. It produces a clean drawn look, handles faces and bodies with consistency, and gives reliable posing and positioning across prompts. Different than the checkpoint, trained with similar data but more focused and on a smaller data set.

Im actively working on the next refined version with sharper eyes and improved facial detailing — this experimental release is a preview of what’s to come.

• ✅ Strong drawn/anime illustration style

• ✅ Faces and bodies render with good accuracy

• ✅ Posing & positioning work naturally

• ✅ Pairs well with Illustrious LoRAs & Illustrious Base

• ✅ Best results between 0.4 – 1.0 strength (tested at 0.6 & 0.8)

⚡ My other models, along with ongoing updates and exclusive builds below. V6 is here!: https://linktr.ee/XRart.