anime-cel-shaded

Classic Anime Cel-Shaded SDXL LoRA

What it does

This LoRA model captures the essence of classic anime cel-shaded illustrations, featuring clean ink line art with crisp shadow shapes, vibrant colors, and expressive characters reminiscent of high-quality TV animation.

How to use

Trigger the effect by including '4n1mec3l' in your prompt. A suggested strength setting between 0.6 and 1.0 works well; start at around 0.85 for balanced results. Use a sampler that supports LoRA, such as Euler or Heun, with standard settings of 25-50 steps.

Best for

• Illustrating scenes from classic anime series
• Depicting expressive characters in vibrant color schemes
• Creating detailed backgrounds and environments
• Enhancing the visual impact of character designs
• Rendering dynamic action sequences

Tips

• Experiment with different strengths to achieve varying degrees of cel-shading intensity.
• Combine this LoRA with textural or lighting enhancements for more complex illustrations.

Example Prompts

4n1mec3l portrait of a teenage girl with long hair in profile, looking pensively out the window, soft morning light, vibrant colors

4n1mec3l ancient city skyline at dusk, dramatic sky, warm golden lights on buildings, wide shot emphasizing architecture and mood

4n1mec3l close-up of a cat sitting on a windowsill, looking out into raindrops falling outside, blue and grey color palette, soft focus

4n1mec3l ninja jumping over rooftops at night, dynamic pose, stars twinkling in the background, dark shadows and bright highlights

4n1mec3l cozy coffee shop interior with a lone customer reading a book, warm ambient lighting, soft colors, intimate setting, peaceful mood

4n1mec3l abstract pattern of interlocking circles and lines forming a mandala, vivid hues, sharp contrasts between light and dark

Trigger word

Activate with 4n1mec3l. 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/anime-cel-shaded")
image = pipe("4n1mec3l, your prompt", num_inference_steps=30).images[0]