Ukiyo-e LoRa Style (Flux.1D)

A respectful homage to classic Japanese woodblock printmaking. This LoRA aims to reproduce crisp linework, a limited indigo/vermilion/sumi palette, washi paper grain, and subtle bokashi gradients from the Edo–Meiji tradition—without modern plastic lighting or heavy 3D shading. It’s built “the old way”: patient curation and careful training to keep compositions calm, balanced, and timeless.

What it’s good at

• Portraits, coastal seascapes, lantern-lit streets, seasonal motifs (maple leaves, waves, mist).

• Elegant kimono patterns, cartouches, publisher seals, and disciplined negative space.

Training summary (Civitai Lora Trainer / FLUX.1D)

• Base model: FLUX.1-dev

• Dataset: ~1000 curated ukiyo-e images, 1:1 @ 1024px

• Epochs: 12 · Batch: 2 · Dim/Alpha: 16/16

• UNet LR: 1e-4 · Text LR: 1e-5

• Bucket: OFF (dataset is square) · Flip aug: OFF · Clip Skip: 1

Usage

• Works without a trigger. Optional helper tag: ukiyoe_style.

• Add 2–3 style hints in the prompt, e.g. ukiyo-e woodblock print, Japanese printmaking, washi paper texture.

• LoRA weight: 0.7–0.9 (push to 1.0 for a stronger print effect).

Recommended inference

• Resolution: 1024×1024 (or tasteful verticals 1024×1536 / 1024×2048)

• Steps: 28–36 · CFG: 3.5–5.0

• Sampler: DPM++ 2M Karras (or a comparable FLUX-friendly sampler)

Negative prompt (short)

modern logos, watermark, photoreal plastic look, harsh bloom, heavy 3D shading, text artifacts, extra fingers, gore

Tip: If the style “overprints,” lower LoRA weight to 0.6–0.7 or reduce CFG. If it feels too light, raise weight to 1.0 and emphasize ukiyo-e woodblock print in the prompt.