This LoRA is designed to generate Ruijerd Supardia ( rjsd ) from Mushoku Tensei: Jobless Reincarnation, with his three looks (green hair, blue hair, or bald with his headband)
Available for Krea2.
Recommended LoRA Strength Krea2: 0.8 – 1.0
Trigger words: rjsd,
Base Character: KREA2: rjsd, A mature adult man with short spiky green hair, yellow eyes, a red gem set in the middle of his forehead and a long diagonal scar across his face, fair skin.
Other looks: KREA2 Blue hair: replace "short spiky green hair" with "short spiky blue hair"
Bald: replace the hair and the gem with "a bald head wearing a black headband with a plain smooth metal forehead plate engraved with a simple spiral pattern"
Original Outfit (Optional): Describe these pieces if you want to generate his signature outfit:
Piece by piece: KREA2
Chest: shirtless under a sleeveless harness vest of dark brown leather straps with beige panels decorated with brown Greek-key spiral patterns
Abdomen: wide tan leather straps with round metal disc buckles across his abdomen
Arms: a black armband with a red, blue and white zigzag pattern on the upper arm, dark brown cloth wraps on the forearms
Belt: a wide woven belt with dark red, white and blue stripes and arrow motifs
Legs: baggy olive-green trousers, dark brown leg wraps over his shins
Footwear: simple brown sandals
Tips: KREA2: write your prompt as natural sentences, starting with the trigger word, like "rjsd, A mature adult man with short spiky green hair..." and then "He is wearing ...". Quality tags are not needed. The LoRA was trained on anime screenshots, so you can add "An anime screenshot" at the start.
For the bald look, always describe the headband plate as above. Otherwise Krea2 may write the trigger word on it.
Krea2 itself (also without the LoRA) can produce a mottled, paper-like texture on flat bright colors, especially over plain black backgrounds for anime.
Please check the gallery images to see the exact prompts and settings I used to get an idea of how it behaves. I'd love to see your generations, so feel free to share your results in the comments!
Training Base Model This LoRA was trained using the following base model:
krea2_raw_bf16
Tested in Forge Neo UI.