DjStompzone/FLUX-1-Dev-LineArt-ControlNet

1import gradio as gr2import torch3from PIL import Image4from diffusers.utils import load_image5from diffusers.pipelines.flux.pipeline_flux_controlnet import FluxControlNetPipeline6from diffusers.models.controlnet_flux import FluxControlNetModel7 8base_model = 'black-forest-labs/FLUX.1-dev'9controlnet_model = 'promeai/FLUX.1-controlnet-lineart-promeai'10controlnet = FluxControlNetModel.from_pretrained(controlnet_model, torch_dtype=torch.bfloat16)11pipe = FluxControlNetPipeline.from_pretrained(base_model, controlnet=controlnet, torch_dtype=torch.bfloat16)12pipe.to("cuda")13 14def generate_image(prompt, control_image, controlnet_conditioning_scale, num_inference_steps, guidance_scale):15    control_image = load_image(control_image) if isinstance(control_image, str) else control_image16 17    result = pipe(18        prompt,19        control_image=control_image,20        controlnet_conditioning_scale=controlnet_conditioning_scale,21        num_inference_steps=num_inference_steps,22        guidance_scale=guidance_scale,23    ).images[0]24    25    return result26 27with gr.Blocks() as demo:28    gr.Markdown("# FLUX ControlNet Pipeline Interface")29    30    with gr.Row():31        with gr.Column():32            prompt = gr.Textbox(label="Prompt", lines=3, placeholder="Enter your prompt here...")33            control_image = gr.Image(source="upload", type="filepath", label="Control Image")34            35            controlnet_conditioning_scale = gr.Slider(0.0, 1.0, value=0.6, label="ControlNet Conditioning Scale")36            num_inference_steps = gr.Slider(1, 100, value=28, step=1, label="Number of Inference Steps")37            guidance_scale = gr.Slider(1.0, 10.0, value=3.5, label="Guidance Scale")38            39            generate_button = gr.Button("Generate Image")40        41        with gr.Column():42            output_image = gr.Image(label="Generated Image")43    generate_button.click(44        generate_image, 45        inputs=[prompt, control_image, controlnet_conditioning_scale, num_inference_steps, guidance_scale],46        outputs=output_image47    )48 49demo.launch()