ukiyoe-woodblock

SDXL LoRA: Ukiyo-e Woodblock Print Style

What it does

Capture the essence of traditional Japanese ukiyo-e woodblock prints with this LoRA model. It renders images in the distinctive style characterized by flat color planes, bold ink outlines, and a palette dominated by indigo and vermilion.

How to use

To apply the ukiyo-e aesthetic, add 'uk1y03' as your trigger word to your prompt. For best results, set the strength between 0.6 and 1.0, with 0.85 being a good starting point. Use an EMA sampler for smoother transitions.

Best for

• Adding traditional Japanese art style to any scene
• Illustrating historical or fantasy scenes involving Japan
• Depicting iconic figures like samurai warriors or courtesans in period attire
• Enhancing natural landscapes with the elegance of ukiyo-e
• Creating stylized artwork inspired by Edo-period aesthetics

Tips

• Incorporate Japanese motifs and patterns to complement the style.
• Adjust the strength based on the complexity of your scene for optimal results.

Example Prompts

uk1y03 A serene Japanese tea house under cherry blossoms at dusk, evoking peace and tranquility.

uk1y03 Vibrant indigo kimono with intricate gold patterns, elegant portrait of a samurai warrior in profile.

uk1y03 Bold red bridge crossing blue river, architectural scene with pagoda and cherry trees in full bloom.

uk1y03 A carpenter meticulously carving woodblocks for ukiyo-e print, hands painted in black ink outlines.

uk1y03 Moonlit night over Tokyo bay, mysterious and eerie mood with silhouettes of boats and distant lights.

uk1y03 Abstract pattern inspired by ukiyo-e motifs, swirling indigo and vermillion colors on aged paper texture.

Trigger word

Activate with uk1y03. Recommended strength: 0.9.

Showcase

Usage with diffusers

from diffusers import AutoPipelineForText2Image
import torch

pipe = AutoPipelineForText2Image.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16).to("cuda")
pipe.load_lora_weights("strkyyy/ukiyoe-woodblock")
image = pipe("uk1y03, your prompt", num_inference_steps=30).images[0]