app.py

import gradio as gr
import cv2
import numpy as np
from PIL import Image, ImageEnhance
from gradio_imageslider import ImageSlider

def apply_filter(image, filter_type, intensity):
    image = np.array(image)
    normalized_intensity = intensity / 100.0

    if filter_type == "Grayscale":
        return convert_to_grayscale(image)
    elif filter_type == "Soft Glow":
        base_intensity = 0.1
        adjusted_intensity = base_intensity + (normalized_intensity * (1 - base_intensity))
        gaussian = cv2.GaussianBlur(image, (15, 15), 0)
        soft_glow = cv2.addWeighted(image, 1 - adjusted_intensity, gaussian, adjusted_intensity, 0)
        return soft_glow
    elif filter_type == "Portrait Enhancer":
        base_intensity = 0.5
        adjusted_intensity = base_intensity + (normalized_intensity * (1 - base_intensity))
        image_pil = Image.fromarray(image)
        enhancer = ImageEnhance.Sharpness(image_pil)
        image_pil = enhancer.enhance(1 + 0.5 * adjusted_intensity)
        enhancer = ImageEnhance.Color(image_pil)
        image_pil = enhancer.enhance(1 + 0.5 * adjusted_intensity)
        enhanced_image = np.array(image_pil)
        return enhanced_image
    elif filter_type == "Warm Tone":
        warm_image = cv2.addWeighted(image, 1.0, np.full(image.shape, (20, 66, 112), dtype=np.uint8), 0.3 * normalized_intensity, 0)
        return warm_image
    elif filter_type == "Cold Tone":
        cold_image = cv2.addWeighted(image, 1.0, np.full(image.shape, (112, 66, 20), dtype=np.uint8), 0.3 * normalized_intensity, 0)
        return cold_image
    elif filter_type == "High-Key":
        high_key = cv2.convertScaleAbs(image, alpha=1.0 + 0.3 * normalized_intensity, beta=20)
        return high_key
    elif filter_type == "Low-Key":
        low_key = cv2.convertScaleAbs(image, alpha=1.0 - 0.1 * normalized_intensity, beta=-10)
        return low_key
    elif filter_type == "Haze":
        haze = cv2.addWeighted(image, 1.0, np.full(image.shape, 255, dtype=np.uint8), 0.3 * normalized_intensity, 0)
        return haze
    else:
        return image

def convert_to_grayscale(image):
    gray_image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    return cv2.cvtColor(gray_image, cv2.COLOR_GRAY2BGR)

def convert_and_save(image, filter_type, intensity):
    image_cv = cv2.cvtColor(np.array(image), cv2.COLOR_RGB2BGR)
    filtered_image = apply_filter(image_cv, filter_type, intensity)
    
    original_image_pil = Image.fromarray(cv2.cvtColor(image_cv, cv2.COLOR_BGR2RGB))
    filtered_image_pil = Image.fromarray(cv2.cvtColor(filtered_image, cv2.COLOR_BGR2RGB))

    output_path = "filtered_image.jpg"
    filtered_image_pil.save(output_path)

    return [original_image_pil, filtered_image_pil], output_path

def get_filter_description(filter_type):
    descriptions = {
        "Grayscale": "이미지를 흑백으로 변환합니다.",
        "Soft Glow": "부드러운 빛을 추가하여 이미지를 은은하게 만듭니다.",
        "Portrait Enhancer": "피부 톤을 균일하게 하고 선명도를 조절하여 인물을 더욱 돋보이게 만듭니다.",
        "Warm Tone": "따뜻한 색조를 추가하여 이미지에 온기를 더합니다.",
        "Cold Tone": "차가운 색조를 추가하여 이미지에 시원함을 더합니다.",
        "High-Key": "밝고 화사한 이미지를 만들어냅니다.",
        "Low-Key": "어두운 톤을 강조하여 분위기 있는 이미지를 만듭니다.",
        "Haze": "부드럽고 흐릿한 효과를 추가하여 몽환적인 이미지를 만듭니다."
    }
    return descriptions.get(filter_type, "")

with gr.Blocks() as iface:
    with gr.Row():
        with gr.Column():
            image_input = gr.Image(type="pil", label="이미지 업로드")
            filter_input = gr.Radio(
                ["Grayscale", "Soft Glow", "Portrait Enhancer", "Warm Tone", "Cold Tone", "High-Key", "Low-Key", "Haze"], 
                label="필터 선택",
                value="Soft Glow"
            )
            intensity_slider = gr.Slider(1, 100, value=50, label="필터 강도")
            description_output = gr.Markdown(get_filter_description("Soft Glow"))

        with gr.Column():
            slider_output = ImageSlider(label="Before and After", type="pil")
            download_link = gr.File(label="Download Filtered Image")

    filter_input.change(fn=get_filter_description, inputs=filter_input, outputs=description_output)

    process_button = gr.Button("필터 적용")
    process_button.click(
        fn=convert_and_save,
        inputs=[image_input, filter_input, intensity_slider],
        outputs=[slider_output, download_link]
    )

    iface.title = "인물 사진에 최적화된 필터"
    iface.description = "인물 사진에 최적화된 다양한 필터를 적용할 수 있습니다."

iface.launch()