svjack/Genshin_Impact_Scaramouche_Storyline_Video_Captioned_Tiny

import os
import cv2
import numpy as np
from moviepy.editor import VideoFileClip
from tqdm import tqdm
import shutil

def change_resolution_and_save(input_path, output_path, target_width=1024, target_height=768, max_duration=4):
    """Process images and videos to target resolution and split videos into segments."""
    os.makedirs(output_path, exist_ok=True)

    for root, dirs, files in os.walk(input_path):
        for file in tqdm(files, desc="Processing files"):
            file_path = os.path.join(root, file)
            relative_path = os.path.relpath(file_path, input_path)
            output_dir = os.path.dirname(os.path.join(output_path, relative_path))

            # Process images
            if file.lower().endswith(('.png', '.jpg', '.jpeg')):
                try:
                    img = cv2.imread(file_path)
                    h, w = img.shape[:2]
                    scale = min(target_width / w, target_height / h)
                    new_w = int(w * scale)
                    new_h = int(h * scale)
                    resized_img = cv2.resize(img, (new_w, new_h), interpolation=cv2.INTER_AREA)
                    background = np.zeros((target_height, target_width, 3), dtype=np.uint8)
                    x_offset = (target_width - new_w) // 2
                    y_offset = (target_height - new_h) // 2
                    background[y_offset:y_offset+new_h, x_offset:x_offset+new_w] = resized_img
                    output_file_path = os.path.join(output_path, relative_path)
                    os.makedirs(os.path.dirname(output_file_path), exist_ok=True)
                    cv2.imwrite(output_file_path, background)

                    # Copy corresponding txt file
                    base_name = os.path.splitext(file)[0]
                    txt_source = os.path.join(root, f"{base_name}.txt")
                    if os.path.exists(txt_source):
                        txt_target = os.path.join(output_dir, f"{base_name}.txt")
                        shutil.copy2(txt_source, txt_target)
                except Exception as e:
                    print(f"Failed to process image {file_path}: {e}")

            # Process videos
            elif file.lower().endswith('.mp4'):
                try:
                    clip = VideoFileClip(file_path)
                    total_duration = clip.duration
                    base_name = os.path.splitext(file)[0]

                    if total_duration <= max_duration:
                        # Process the entire video
                        output_filename = f"{base_name}.mp4"
                        output_file_path = os.path.join(output_dir, output_filename)
                        os.makedirs(os.path.dirname(output_file_path), exist_ok=True)

                        def process_frame(frame):
                            img = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
                            h, w = img.shape[:2]
                            scale = min(target_width / w, target_height / h)
                            new_w = int(w * scale)
                            new_h = int(h * scale)
                            resized_img = cv2.resize(img, (new_w, new_h), interpolation=cv2.INTER_AREA)
                            background = np.zeros((target_height, target_width, 3), dtype=np.uint8)
                            x_offset = (target_width - new_w) // 2
                            y_offset = (target_height - new_h) // 2
                            background[y_offset:y_offset+new_h, x_offset:x_offset+new_w] = resized_img
                            return cv2.cvtColor(background, cv2.COLOR_BGR2RGB)

                        processed_clip = clip.fl_image(process_frame)
                        fps = processed_clip.fps if processed_clip.fps else 24
                        processed_clip.write_videofile(
                            output_file_path,
                            codec='libx264',
                            fps=fps,
                            preset='slow',
                            threads=4,
                            audio=False
                        )
                        processed_clip.close()

                        # Copy corresponding txt file
                        txt_source = os.path.join(root, f"{base_name}.txt")
                        if os.path.exists(txt_source):
                            txt_target = os.path.join(output_dir, f"{base_name}.txt")
                            #shutil.copy2(txt_source, txt_target)
                            with open(txt_source, "r") as f:
                                txt = "In the style of Scaramouche , " + f.read().strip()
                            with open(txt_target, "w") as o:
                                o.write(txt)
                    else:
                        # Split and process the video
                        num_segments = int(total_duration // max_duration)
                        for i in range(num_segments):
                            start_time = i * max_duration
                            end_time = min((i+1) * max_duration, total_duration)
                            sub_clip = clip.subclip(start_time, end_time)

                            output_filename = f"{base_name}_{i}.mp4"
                            output_file_path = os.path.join(output_dir, output_filename)
                            os.makedirs(os.path.dirname(output_file_path), exist_ok=True)

                            def process_frame(frame):
                                img = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
                                h, w = img.shape[:2]
                                scale = min(target_width / w, target_height / h)
                                new_w = int(w * scale)
                                new_h = int(h * scale)
                                resized_img = cv2.resize(img, (new_w, new_h), interpolation=cv2.INTER_AREA)
                                background = np.zeros((target_height, target_width, 3), dtype=np.uint8)
                                x_offset = (target_width - new_w) // 2
                                y_offset = (target_height - new_h) // 2
                                background[y_offset:y_offset+new_h, x_offset:x_offset+new_w] = resized_img
                                return cv2.cvtColor(background, cv2.COLOR_BGR2RGB)

                            processed_clip = sub_clip.fl_image(process_frame)
                            fps = processed_clip.fps if processed_clip.fps else 24
                            processed_clip.write_videofile(
                                output_file_path,
                                codec='libx264',
                                fps=fps,
                                preset='slow',
                                threads=4,
                                audio=False
                            )
                            processed_clip.close()

                            # Copy corresponding txt file
                            txt_source = os.path.join(root, f"{base_name}.txt")
                            if os.path.exists(txt_source):
                                txt_target = os.path.join(output_dir, f"{base_name}_{i}.txt")
                                '''
                                shutil.copy2(txt_source, txt_target)
                                '''
                                with open(txt_source, "r") as f:
                                    txt = "In the style of Scaramouche , " + f.read().strip()
                                with open(txt_target, "w") as o:
                                    o.write(txt)

                    clip.close()
                except Exception as e:
                    print(f"Failed to process video {file_path}: {e}")

# Example usage
change_resolution_and_save(
    input_path="Genshin_Impact_Scaramouche_Storyline_Video_Captioned_Tiny",
    output_path="Genshin_Impact_Scaramouche_Storyline_Video_Captioned_Tiny_960x544x6",
    target_width=960,
    target_height=544,
    max_duration=6
)