Ghibli_lora

Ghibli Style LoRA for FLUX.1 Kontext Model

This repository provides the Ghibli style LoRA adapter for the FLUX.1 Kontext Model. This LoRA is part of a collection of 20+ style LoRAs trained on high-quality paired data generated by GPT-4o from the OmniConsistency dataset.

Contributor: Tian YE & Song FEI, HKUST Guangzhou.

Style Showcase

Here are some examples of images generated using this style LoRA:

Inference Example

from diffusers import FluxKontextPipeline
from diffusers.utils import load_image
import torch

# Load the base pipeline
pipeline = FluxKontextPipeline.from_pretrained(
    "black-forest-labs/FLUX.1-Kontext-dev", 
    torch_dtype=torch.bfloat16
).to('cuda')

# Load the LoRA adapter for the Ghibli style directly from the Hub
pipeline.load_lora_weights("Kontext-Style/Ghibli_lora", weight_name="Ghibli_lora_weights.safetensors", adapter_name="lora")
pipeline.set_adapters(["lora"], adapter_weights=[1])

# Load a source image (you can use any image)
image = load_image("https://huggingface.co/datasets/black-forest-labs/kontext-bench/resolve/main/test/images/0003.jpg").resize((1024, 1024))

# Prepare the prompt
# The style_name is used in the prompt and for the output filename.
style_name = "Ghibli"
prompt = f"Turn this image into the Ghibli style."

# Run inference
result_image = pipeline(
    image=image, 
    prompt=prompt, 
    height=1024, 
    width=1024, 
    num_inference_steps=24
).images[0]

# Save the result
output_filename = f"{style_name.replace(' ', '_')}.png"
result_image.save(output_filename)

print(f"Image saved as {output_filename}")

Feel free to open an issue or contact us for feedback or collaboration!