app.py

import spaces
import random
import torch
import cv2
import gradio as gr
import numpy as np
from huggingface_hub import snapshot_download
from transformers import pipeline
from diffusers.utils import load_image
from kolors.pipelines.pipeline_controlnet_xl_kolors_img2img import StableDiffusionXLControlNetImg2ImgPipeline
from kolors.models.modeling_chatglm import ChatGLMModel
from kolors.models.tokenization_chatglm import ChatGLMTokenizer
from kolors.models.controlnet import ControlNetModel
from diffusers import AutoencoderKL
from kolors.models.unet_2d_condition import UNet2DConditionModel
from diffusers import EulerDiscreteScheduler
from PIL import Image, ImageDraw, ImageFont
import os

device = "cuda"
ckpt_dir = snapshot_download(repo_id="Kwai-Kolors/Kolors")
ckpt_dir_canny = snapshot_download(repo_id="Kwai-Kolors/Kolors-ControlNet-Canny")

# Add translation pipeline
translator = pipeline("translation", model="Helsinki-NLP/opus-mt-ko-en")

text_encoder = ChatGLMModel.from_pretrained(f'{ckpt_dir}/text_encoder', torch_dtype=torch.float16).half().to(device)
tokenizer = ChatGLMTokenizer.from_pretrained(f'{ckpt_dir}/text_encoder')
vae = AutoencoderKL.from_pretrained(f"{ckpt_dir}/vae", revision=None).half().to(device)
scheduler = EulerDiscreteScheduler.from_pretrained(f"{ckpt_dir}/scheduler")
unet = UNet2DConditionModel.from_pretrained(f"{ckpt_dir}/unet", revision=None).half().to(device)
controlnet_canny = ControlNetModel.from_pretrained(f"{ckpt_dir_canny}", revision=None).half().to(device)

pipe_canny = StableDiffusionXLControlNetImg2ImgPipeline(
    vae=vae,
    controlnet=controlnet_canny,
    text_encoder=text_encoder,
    tokenizer=tokenizer,
    unet=unet,
    scheduler=scheduler,
    force_zeros_for_empty_prompt=False
)

@spaces.GPU
def translate_korean_to_english(text):
    if any(ord(char) >= 0xAC00 and ord(char) <= 0xD7A3 for char in text):  # Check if Korean characters are present
        translated = translator(text, max_length=512)[0]['translation_text']
        return translated
    return text

def HWC3(x):
    assert x.dtype == np.uint8
    if x.ndim == 2:
        x = x[:, :, None]
    assert x.ndim == 3
    H, W, C = x.shape
    assert C == 1 or C == 3 or C == 4
    if C == 3:
        return x
    if C == 1:
        return np.concatenate([x, x, x], axis=2)
    if C == 4:
        color = x[:, :, 0:3].astype(np.float32)
        alpha = x[:, :, 3:4].astype(np.float32) / 255.0
        y = color * alpha + 255.0 * (1.0 - alpha)
        y = y.clip(0, 255).astype(np.uint8)
        return y

@spaces.GPU
def process_canny_condition(image, canny_threods=[100,200]):
    np_image = np.array(image)
    np_image = cv2.Canny(np_image, canny_threods[0], canny_threods[1])
    np_image = np_image[:, :, None]
    np_image = np.concatenate([np_image, np_image, np_image], axis=2)
    np_image = HWC3(np_image)
    return Image.fromarray(np_image)

MAX_SEED = np.iinfo(np.int32).max
MAX_IMAGE_SIZE = 1024

def resize_image(image, resolution):
    w, h = image.size
    ratio = resolution / max(w, h)
    new_w = int(w * ratio)
    new_h = int(h * ratio)
    return image.resize((new_w, new_h), Image.LANCZOS)

def text_to_image(text, size=72, position="middle-center"):
    width, height = 1024, 576
    image = Image.new("RGB", (width, height), "white")
    draw = ImageDraw.Draw(image)

    font_files = ["Arial_Unicode.ttf"]
    font = None
    for font_file in font_files:
        font_path = os.path.join(os.path.dirname(__file__), font_file)
        if os.path.exists(font_path):
            try:
                font = ImageFont.truetype(font_path, size=size)
                print(f"Using font: {font_file}")
                break
            except IOError:
                print(f"Error loading font: {font_file}")
    if font is None:
        print("No suitable font found. Using default font.")
        font = ImageFont.load_default()

    lines = text.split('\n')
    max_line_width = 0
    total_height = 0
    line_heights = []
    for line in lines:
        left, top, right, bottom = draw.textbbox((0, 0), line, font=font)
        line_width = right - left
        line_height = bottom - top
        line_heights.append(line_height)
        max_line_width = max(max_line_width, line_width)
        total_height += line_height

    position_mapping = {
        "top-left": (10, 10),
        "top-left-center": (width // 4 - max_line_width // 2, 10),
        "top-center": ((width - max_line_width) / 2, 10),
        "top-right-center": (3 * width // 4 - max_line_width // 2, 10),
        "top-right": (width - max_line_width - 10, 10),
        "upper-left": (10, height // 4 - total_height // 2),
        "upper-left-center": (width // 4 - max_line_width // 2, height // 4 - total_height // 2),
        "upper-center": ((width - max_line_width) / 2, height // 4 - total_height // 2),
        "upper-right-center": (3 * width // 4 - max_line_width // 2, height // 4 - total_height // 2),
        "upper-right": (width - max_line_width - 10, height // 4 - total_height // 2),
        "middle-left": (10, (height - total_height) / 2),
        "middle-left-center": (width // 4 - max_line_width // 2, (height - total_height) / 2),
        "middle-center": ((width - max_line_width) / 2, (height - total_height) / 2),
        "middle-right-center": (3 * width // 4 - max_line_width // 2, (height - total_height) / 2),
        "middle-right": (width - max_line_width - 10, (height - total_height) / 2),
        "lower-left": (10, 3 * height // 4 - total_height // 2),
        "lower-left-center": (width // 4 - max_line_width // 2, 3 * height // 4 - total_height // 2),
        "lower-center": ((width - max_line_width) / 2, 3 * height // 4 - total_height // 2),
        "lower-right-center": (3 * width // 4 - max_line_width // 2, 3 * height // 4 - total_height // 2),
        "lower-right": (width - max_line_width - 10, 3 * height // 4 - total_height // 2),
        "bottom-left": (10, height - total_height - 10),
        "bottom-left-center": (width // 4 - max_line_width // 2, height - total_height - 10),
        "bottom-center": ((width - max_line_width) / 2, height - total_height - 10),
        "bottom-right-center": (3 * width // 4 - max_line_width // 2, height - total_height - 10),
        "bottom-right": (width - max_line_width - 10, height - total_height - 10),
    }

    x, y = position_mapping.get(position, ((width - max_line_width) / 2, (height - total_height) / 2))
    for i, line in enumerate(lines):
        draw.text((x, y), line, fill="black", font=font)
        y += line_heights[i]

    return image

@spaces.GPU
def infer_canny(prompt, text_for_image, text_position, font_size,
          negative_prompt = "nsfw, facial shadows, low resolution, jpeg artifacts, blurry, bad quality, dark face, neon lights", 
          seed = 397886929, 
          randomize_seed = False,
          guidance_scale = 8.0, 
          num_inference_steps = 50,
          controlnet_conditioning_scale = 0.8,
          control_guidance_end = 0.9,
          strength = 1.0
        ):

    prompt = translate_korean_to_english(prompt)
    negative_prompt = translate_korean_to_english(negative_prompt)

    if randomize_seed:
        seed = random.randint(0, MAX_SEED)
    generator = torch.Generator().manual_seed(seed)

    # Generate text image
    init_image = text_to_image(text_for_image, size=font_size, position=text_position)
    init_image = resize_image(init_image, MAX_IMAGE_SIZE)

    pipe = pipe_canny.to("cuda")
    condi_img = process_canny_condition(init_image)
 
    image = pipe(
        prompt=prompt,
        image=init_image,
        controlnet_conditioning_scale=controlnet_conditioning_scale,
        control_guidance_end=control_guidance_end, 
        strength=strength, 
        control_image=condi_img,
        negative_prompt=negative_prompt, 
        num_inference_steps=num_inference_steps, 
        guidance_scale=guidance_scale,
        num_images_per_prompt=1,
        generator=generator,
    ).images[0]
    return image, seed  # CANNY 이미지 반환 제거

def update_button_states(selected_position):
    return [
        gr.update(variant="primary") if pos == selected_position else gr.update(variant="secondary")
        for pos in position_list
    ]

position_list = [
    "top-left", "top-left-center", "top-center", "top-right-center", "top-right",
    "upper-left", "upper-left-center", "upper-center", "upper-right-center", "upper-right",
    "middle-left", "middle-left-center", "middle-center", "middle-right-center", "middle-right",
    "lower-left", "lower-left-center", "lower-center", "lower-right-center", "lower-right",
    "bottom-left", "bottom-left-center", "bottom-center", "bottom-right-center", "bottom-right"
]


css = """
footer {
    visibility: hidden;
}
.text-position-grid {
    display: grid;
    grid-template-columns: repeat(5, 1fr);
    gap: 2px;
    margin-bottom: 10px;
    width: 150px;
}
.text-position-grid button {
    aspect-ratio: 1;
    padding: 0;
    border: 1px solid #ccc;
    background-color: #f0f0f0;
    cursor: pointer;
    font-size: 10px;
    transition: all 0.3s ease;
}
.text-position-grid button:hover {
    background-color: #e0e0e0;
}
.text-position-grid button.selected {
    background-color: #007bff;
    color: white;
    transform: scale(1.1);
}
"""

with gr.Blocks(theme="Nymbo/Nymbo_Theme", css=css) as Kolors:
    text_position = gr.State("middle-center")
    with gr.Row():
        with gr.Column(elem_id="col-left"):
            with gr.Row():
                prompt = gr.Textbox(
                    label="Prompt",
                    placeholder="Enter your prompt",
                    lines=2,
                    value="coffee in a cup bokeh --ar 85:128 --v 6.0 --style raw5, 4K, 리얼리티 사진"  # Default value added here
                )
            with gr.Row():
                text_for_image = gr.Textbox(
                    label="Text for Image Generation",
                    placeholder="Enter text to be converted into an image",
                    lines=3,
                    value="대한 萬世 GO"  # Default value added here
                )
            with gr.Row():
                with gr.Column():
                    gr.Markdown("Text Position")
                    with gr.Row(elem_classes="text-position-grid"):
                        position_buttons = [gr.Button("•") for _ in range(25)]
                    
                    for btn, pos in zip(position_buttons, position_list):
                        btn.click(lambda p=pos: p, outputs=text_position)
                        btn.click(update_button_states, inputs=[text_position], outputs=position_buttons)
                with gr.Column():
                    font_size = gr.Slider(
                        label="Text Size",
                        minimum=12,
                        maximum=144,
                        step=1,
                        value=72
                    )
            with gr.Accordion("Advanced Settings", open=False):
                negative_prompt = gr.Textbox(
                    label="Negative prompt",
                    placeholder="Enter a negative prompt",
                    visible=True,
                    value="nsfw, facial shadows, low resolution, jpeg artifacts, blurry, bad quality, dark face, neon lights"
                )
                seed = gr.Slider(
                    label="Seed",
                    minimum=0,
                    maximum=MAX_SEED,
                    step=1,
                    value=0,
                )
                randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
                with gr.Row():
                    guidance_scale = gr.Slider(
                        label="Guidance scale",
                        minimum=0.0,
                        maximum=10.0,
                        step=0.1,
                        value=8.0,
                    )
                    num_inference_steps = gr.Slider(
                        label="Number of inference steps",
                        minimum=10,
                        maximum=50,
                        step=1,
                        value=50,
                    )
                with gr.Row():
                    controlnet_conditioning_scale = gr.Slider(
                        label="Controlnet Conditioning Scale",
                        minimum=0.0,
                        maximum=1.0,
                        step=0.1,
                        value=0.8,
                    )
                    control_guidance_end = gr.Slider(
                        label="Control Guidance End",
                        minimum=0.0,
                        maximum=1.0,
                        step=0.1,
                        value=0.9,
                    )
                with gr.Row():
                    strength = gr.Slider(
                        label="Strength",
                        minimum=0.0,
                        maximum=1.0,
                        step=0.1,
                        value=1.0,
                    )
            with gr.Row():
                canny_button = gr.Button("Start", elem_id="button")


        with gr.Column(elem_id="col-right"):
            result = gr.Image(label="Result", show_label=False)  # Gallery에서 Image로 변경
            seed_used = gr.Number(label="Seed Used")

    canny_button.click(
        fn = infer_canny,
        inputs = [prompt, text_for_image, text_position, font_size, negative_prompt, seed, randomize_seed, guidance_scale, num_inference_steps, controlnet_conditioning_scale, control_guidance_end, strength],
        outputs = [result, seed_used]
    )
        


    # Set initial button states
    Kolors.load(update_button_states, inputs=[text_position], outputs=position_buttons)

Kolors.launch()