app.py

from diffusers import ControlNetModel, StableDiffusionXLControlNetPipeline, AutoencoderKL
from diffusers.utils import load_image
from PIL import Image
import torch
import numpy as np
import cv2
import gradio as gr
from torchvision import transforms 
# from huggingface_hub import login
# login()

controlnet_conditioning_scale = 1.0 

controlnet = ControlNetModel.from_pretrained(
    "briaai/ControlNet-Canny",
    torch_dtype=torch.float16
)

pipe = StableDiffusionXLControlNetPipeline.from_pretrained(
    "briaai/BRIA-2.0",
    controlnet=controlnet,
    torch_dtype=torch.float16,
)
pipe.enable_model_cpu_offload()

low_threshold = 100
high_threshold = 200

def resize_image(image):
    current_size = image.size
    if current_size[0] > current_size[1]:
        center_cropped_image = transforms.functional.center_crop(image, (current_size[1], current_size[1]))
    else:
        center_cropped_image = transforms.functional.center_crop(image, (current_size[0], current_size[0]))
    resized_image = transforms.functional.resize(center_cropped_image, (1024, 1024))
    return resized_image

def get_canny_filter(image):
    
    if not isinstance(image, np.ndarray):
        image = np.array(image) 
        
    image = cv2.Canny(image, low_threshold, high_threshold)
    image = image[:, :, None]
    image = np.concatenate([image, image, image], axis=2)
    canny_image = Image.fromarray(image)
    return canny_image

def process(input_image, prompt):
    # resize input_image to 1024x1024
    input_image = resize_image(input_image)
    
    canny_image = get_canny_filter(input_image)
    images = pipe(
        prompt,image=canny_image, controlnet_conditioning_scale=controlnet_conditioning_scale,
        ).images

    return [canny_image,images[0]]
    
block = gr.Blocks().queue()

with block:
    gr.Markdown("## BRIA 2.0 ControlNet Canny")
    gr.HTML('''
     <p style="margin-bottom: 10px; font-size: 94%">
                This is a demo for BRIA 2.0 ControlNet Canny, a fully legal and safe T2I model. 
              </p>
              ''')
    with gr.Row():
        with gr.Column():
            input_image = gr.Image(sources=None, type="numpy") # None for upload, ctrl+v and webcam
            prompt = gr.Textbox(label="Prompt")
            run_button = gr.Button(value="Run")
            
            
        with gr.Column():
            result_gallery = gr.Gallery(label='Output', show_label=False, elem_id="gallery", columns=[2], height='auto')
    ips = [input_image, prompt]
    run_button.click(fn=process, inputs=ips, outputs=[result_gallery])

block.launch(debug = True)