app.py

import gradio as gr
import numpy as np
import torch
from diffusers.utils import load_image
from diffusers import StableDiffusionControlNetPipeline, ControlNetModel
from peft import PeftModel, LoraConfig
from controlnet_aux import HEDdetector
from PIL import Image
import cv2 as cv
import os
from functools import lru_cache
from contextlib import contextmanager

MAX_SEED = np.iinfo(np.int32).max
MAX_IMAGE_SIZE = 1024
IP_ADAPTER = 'h94/IP-Adapter'
IP_ADAPTER_WEIGHT_NAME = "ip-adapter-plus_sd15.bin"

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model_id_default = "stable-diffusion-v1-5/stable-diffusion-v1-5"
torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32

class PipelineManager:
    def __init__(self):
        self.pipe = None
        self.current_model = None
        self.controlnet_cache = {}
        self.hed = None
        
    @lru_cache(maxsize=2)
    def get_controlnet(self, model_name: str) -> ControlNetModel:
        if model_name not in self.controlnet_cache:
            self.controlnet_cache[model_name] = ControlNetModel.from_pretrained(
                model_name, 
                cache_dir="./models_cache",
                torch_dtype=torch_dtype
            ).to(device)
        return self.controlnet_cache[model_name]
    
    def get_hed_detector(self):
        if self.hed is None:
            self.hed = HEDdetector.from_pretrained('lllyasviel/Annotators')
        return self.hed

    def initialize_pipeline(self, model_id, controlnet_model):
        controlnet = self.get_controlnet(controlnet_model)
        if not self.pipe or model_id != self.current_model:
            self.pipe = self.create_pipeline(model_id, controlnet)
            self.current_model = model_id
        return self.pipe

    def create_pipeline(self, model_id, controlnet):
        pipe = StableDiffusionControlNetPipeline.from_pretrained(
            model_id,
            torch_dtype=torch_dtype,
            controlnet=controlnet,
            cache_dir="./models_cache"
        ).to(device)
        
        if os.path.exists('./lora_logos'):
            pipe = self.load_lora_adapters(pipe)
            
        return pipe

    def load_lora_adapters(self, pipe):
        unet_dir = os.path.join('./lora_logos', "unet")
        text_encoder_dir = os.path.join('./lora_logos', "text_encoder")
        
        pipe.unet = PeftModel.from_pretrained(pipe.unet, unet_dir, adapter_name="default")
        if os.path.exists(text_encoder_dir):
            pipe.text_encoder = PeftModel.from_pretrained(pipe.text_encoder, text_encoder_dir)
            
        return pipe.to(device)

@contextmanager
def torch_inference_mode():
    with torch.inference_mode(), torch.autocast(device.type):
        yield

def process_embeddings(prompt, negative_prompt, tokenizer, text_encoder):
    def process_text(text):
        tokens = tokenizer(text, return_tensors="pt", truncation=False).input_ids
        chunks = [tokens[:, i:i+77].to(device) for i in range(0, tokens.size(1), 77)]
        return torch.cat([text_encoder(chunk)[0] for chunk in chunks], dim=1)
    
    prompt_emb = process_text(prompt)
    negative_emb = process_text(negative_prompt)
    max_len = max(prompt_emb.size(1), negative_emb.size(1))
    
    return (
        torch.nn.functional.pad(prompt_emb, (0, 0, 0, max_len - prompt_emb.size(1))),
        torch.nn.functional.pad(negative_emb, (0, 0, 0, max_len - negative_emb.size(1)))
    )

def process_control_image(image_path: str, processor: str, hed_detector) -> Image:
    image = load_image(image_path).convert('RGB')
    
    if processor == 'edge_detection':
        edges = cv.Canny(np.array(image), 80, 160)
        return Image.fromarray(np.repeat(edges[:, :, None], 3, axis=2))
    
    if processor == 'scribble':
        scribble = hed_detector(image)
        processed = cv.medianBlur(np.array(scribble), 3)
        return Image.fromarray(cv.convertScaleAbs(processed, alpha=1.5))

pipeline_mgr = PipelineManager()
controlnet_models = {
    "edge_detection": "lllyasviel/sd-controlnet-canny",
    "scribble": "lllyasviel/sd-controlnet-scribble"
}

def infer(
    prompt, 
    negative_prompt, 
    width=512, 
    height=512, 
    num_inference_steps=20, 
    model_id='stable-diffusion-v1-5/stable-diffusion-v1-5', 
    seed=42, 
    guidance_scale=7.0, 
    lora_scale=0.5,
    cn_enable=False,
    cn_strength=0.0,
    cn_mode='edge_detection',
    cn_image=None,
    ip_enable=False,
    ip_scale=0.5,
    ip_image=None,
    progress=gr.Progress(track_tqdm=True)
    ):

    generator = torch.Generator(device).manual_seed(seed)
    
    with torch_inference_mode():
        pipe = pipeline_mgr.initialize_pipeline(
            model_id, 
            controlnet_models.get(cn_mode, controlnet_models['edge_detection'])
        )
        
        if cn_enable and not cn_image:
            raise gr.Error("ControlNet enabled but no image provided!")

        if ip_enable and not ip_image:
            raise gr.Error("IP-Adapter enabled but no image provided!")
        
        prompt_emb, negative_emb = process_embeddings(
            prompt, 
            negative_prompt, 
            pipe.tokenizer, 
            pipe.text_encoder
        )
        
        params = {
            'prompt_embeds': prompt_emb,
            'negative_prompt_embeds': negative_emb,
            'guidance_scale': guidance_scale,
            'num_inference_steps': num_inference_steps,
            'width': width,
            'height': height,
            'generator': generator,
            'cross_attention_kwargs': {"scale": lora_scale},
        }
        
        if cn_enable:
            params['image'] = process_control_image(
                cn_image,
                cn_mode,
                pipeline_mgr.get_hed_detector()
            )
            params['controlnet_conditioning_scale'] = float(cn_strength)
        else:
            params['image'] = torch.zeros((1, 3, 512, 512)).to(device)  # заглушка, чтобы pipeline не падал
            params['controlnet_conditioning_scale'] = 0.0
            
        if ip_enable:
            pipe.load_ip_adapter(IP_ADAPTER, subfolder="models", weight_name=IP_ADAPTER_WEIGHT_NAME)
            params['ip_adapter_image'] = load_image(ip_image).convert('RGB')
            pipe.set_ip_adapter_scale(ip_scale)
            
        pipe.fuse_lora(lora_scale=lora_scale)
        
        return pipe(**params).images[0]

css = """
#col-container {
    margin: 0 auto;
    max-width: 640px;
}
"""

with gr.Blocks(css=css) as demo:
    with gr.Column(elem_id="col-container"):
        gr.Markdown("# ⚽️ Football Logo Generator")
        
        with gr.Row():
            model_id = gr.Textbox(
                label="Model ID",
                max_lines=1,
                placeholder="Enter model id like 'stable-diffusion-v1-5/stable-diffusion-v1-5'",
                value=model_id_default
            )

        prompt = gr.Textbox(
            label="Prompt",
            max_lines=1,
            placeholder="Enter your prompt",
        )

        negative_prompt = gr.Textbox(
            label="Negative prompt",
            max_lines=1,
            placeholder="Enter a negative prompt",
        )

        with gr.Row():
            seed = gr.Number(
                label="Seed",
                minimum=0,
                maximum=MAX_SEED,
                step=1,
                value=42,
            )

        with gr.Row():
            guidance_scale = gr.Slider(
                label="Guidance scale",
                minimum=0.0,
                maximum=10.0,
                step=0.1,
                value=7.0,
            )

        with gr.Row():
            lora_scale = gr.Slider(
                label="LoRA scale",
                minimum=0.0,
                maximum=1.0,
                step=0.1,
                value=0.5,
            )

        with gr.Row():
            num_inference_steps = gr.Slider(
                label="Number of inference steps",
                minimum=1,
                maximum=50,
                step=1,
                value=20,
            )

        # Секция Control Net
        cn_enable = gr.Checkbox(label="Enable ControlNet")    
        with gr.Column(visible=False) as cn_options:
            with gr.Row():
                cn_strength = gr.Slider(0, 2, value=0.8, step=0.1, label="Control strength", interactive=True)
                cn_mode = gr.Dropdown(
                    choices=["edge_detection", "scribble"],
                    value="edge_detection",
                    label="Work regime",
                    interactive=True,
                )
            cn_image = gr.Image(type="filepath", label="Control image")

        cn_enable.change(
            lambda x: gr.update(visible=x),
            inputs=cn_enable,
            outputs=cn_options
        )
        
        # Секция IP-Adapter
        ip_enable = gr.Checkbox(label="Enable IP-Adapter")
        with gr.Column(visible=False) as ip_options:
            ip_scale = gr.Slider(0, 1, value=0.5, step=0.1, label="IP-adapter scale", interactive=True)
            ip_image = gr.Image(type="filepath", label="IP-adapter image", interactive=True)

        ip_enable.change(
            lambda x: gr.update(visible=x),
            inputs=ip_enable,
            outputs=ip_options
        )

        with gr.Accordion("Optional Settings", open=False):
            with gr.Row():
                width = gr.Slider(
                    label="Width",
                    minimum=256,
                    maximum=MAX_IMAGE_SIZE,
                    step=32,
                    value=512,
                )
            
            with gr.Row():
                height = gr.Slider(
                    label="Height",
                    minimum=256,
                    maximum=MAX_IMAGE_SIZE,
                    step=32,
                    value=512,
                )

        run_button = gr.Button("Run", scale=1, variant="primary")
        result = gr.Image(label="Result", show_label=False)
    
    gr.on(
        triggers=[run_button.click, prompt.submit],
        fn=infer,
        inputs=[
            prompt,
            negative_prompt,
            width,
            height,
            num_inference_steps,
            model_id,
            seed,
            guidance_scale,
            lora_scale,
            cn_enable,
            cn_strength,
            cn_mode,
            cn_image,
            ip_enable,
            ip_scale,
            ip_image
        ],
        outputs=[result],
    )

if __name__ == "__main__":
    demo.launch()