Linaqruf/pastel-style-xl-lora

Pastel Style XL LoRA

---

Overview

Pastel Style XL LoRA is a specialized LoRA (Low-Rank Adaptation) adapter, expertly designed to work in conjunction with Animagine XL 2.0. This model specifically focuses on enhancing and imparting a pastel-style aesthetic to anime-themed images. It integrates smoothly with the Stable Diffusion framework, offering a unique capability to produce images with soft, pastel-like qualities without the need for specific keywords or tags.

---

Model Details

• Developed by: Linaqruf
• Model type: LoRA adapter for Stable Diffusion XL
• Model Description: Pastel Style XL LoRA is a compact yet potent model aimed at augmenting the output of larger models, particularly Animagine XL 2.0. It's adept at generating and modifying high-quality anime-themed images, giving them a distinctive pastel aesthetic. This model is an excellent choice for those seeking to add a gentle, pastel touch to their anime creations.
• License: CreativeML Open RAIL++-M License
• Finetuned from model: Animagine XL 2.0

---

🧨 Diffusers Installation

Ensure the installation of the latest diffusers library, along with other essential packages:

pip install diffusers --upgrade
pip install transformers accelerate safetensors

The following Python script demonstrates how to utilize the LoRA with Animagine XL 2.0. The default scheduler is EulerAncestralDiscreteScheduler, but it can be explicitly defined for clarity.

import torch
from diffusers import (
    StableDiffusionXLPipeline, 
    EulerAncestralDiscreteScheduler,
    AutoencoderKL
)

# Initialize LoRA model and weights
lora_model_id = "Linaqruf/pastel-style-xl-lora"
lora_filename = "pastel-style-xl-v2.safetensors"

# Load VAE component
vae = AutoencoderKL.from_pretrained(
    "madebyollin/sdxl-vae-fp16-fix", 
    torch_dtype=torch.float16
)

# Configure the pipeline
pipe = StableDiffusionXLPipeline.from_pretrained(
    "Linaqruf/animagine-xl-2.0", 
    vae=vae,
    torch_dtype=torch.float16, 
    use_safetensors=True, 
    variant="fp16"
)
pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config)
pipe.to('cuda')

# Load and fuse LoRA weights
pipe.load_lora_weights(lora_model_id, weight_name=lora_filename)
pipe.fuse_lora(lora_scale=0.6)

# Define prompts and generate image
prompt = "face focus, cute, masterpiece, best quality, 1girl, green hair, sweater, looking at viewer, upper body, beanie, outdoors, night, turtleneck"
negative_prompt = "lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry"

image = pipe(
    prompt, 
    negative_prompt=negative_prompt, 
    width=1024,
    height=1024,
    guidance_scale=12,
    num_inference_steps=50
).images[0]

# Unfuse LoRA before saving the image
pipe.unfuse_lora()
image.save("anime_girl.png")