diff --git a/rdbench-viz/rdbench-viz.py b/rdbench-viz/rdbench-viz.py
new file mode 100644
index 0000000..cafaeb6
--- /dev/null
+++ b/rdbench-viz/rdbench-viz.py
@@ -0,0 +1,88 @@
+import argparse
+from pathlib import Path
+import struct
+import math
+import numpy as np
+from PIL import Image, ImageDraw
+from tqdm import tqdm
+# import matplotlib.pyplot as plt
+import matplotlib.cm as cm
+from matplotlib.colors import LinearSegmentedColormap
+
+
+def read_dat(file: Path):
+    ar = np.array(struct.unpack(f'{file.stat().st_size // 8}d', file.read_bytes()))
+    return ar.reshape((int(math.sqrt(ar.shape[0])), -1))
+
+def create_frame_image(data, cmap = "Blues", frame_number = None):
+    """フレームデータを画像に変換し、カラーマップを適用"""
+    # Normalize data for visualization
+    normalized_data = (data - data.min()) / (data.max() - data.min())
+    
+    # Apply colormap using matplotlib
+    if cmap == 'PastelReds':  # Custom pastel-like Reds colormap
+        colors = [(1, 1, 1), (1, 0.8, 0.8), (1, 0, 0)]  # white -> pink -> red
+        colormap = LinearSegmentedColormap.from_list("PastelReds", colors, N=256)
+    else:
+        colormap = cm.get_cmap(cmap)
+
+    colored_data = colormap(normalized_data)  # This returns an (M, N, 4) array with RGBA values
+    
+    # Convert the RGBA colormap array to RGB (dropping the alpha channel)
+    rgb_data = (colored_data[:, :, :3] * 255).astype(np.uint8)
+    
+    # Convert numpy array to RGB image
+    img = Image.fromarray(rgb_data, 'RGB')
+    
+    # Draw the frame number on the image
+    # draw = ImageDraw.Draw(img)
+    # draw.text((10, 10), f'Frame {frame_number}', fill=(255, 255, 255))  # Use white color for text in RGB mode
+    return img
+
+def generate_gif(input_dir, output_file, initial_duration, frame_duration, cmap, verbose):
+    input_files = sorted(list(Path(input_dir).glob('*.bin')))
+    
+    if not input_files:
+        print(f"No binary files found in directory: {input_dir}")
+        return
+    
+    # Initialize the list to store the images (frames)
+    images = []
+    
+    if verbose:
+        print(f"Processing {len(input_files)} files...")
+    
+    # Process each file
+    for i, file in enumerate(tqdm(input_files, desc="Processing files", disable=not verbose)):
+        data = read_dat(file)
+        img = create_frame_image(data, cmap, i)
+        images.append(img)
+    
+    # Save the images as a GIF
+    if verbose:
+        print(f"Saving GIF to {output_file}...")
+    
+    images[0].save(
+        output_file, save_all=True, append_images=images[1:], 
+        duration=[initial_duration] + [frame_duration] * (len(images) - 1),
+        loop=0
+    )
+
+    if verbose:
+        print(f"GIF saved successfully: {output_file}")
+
+def main():
+    parser = argparse.ArgumentParser(description="Create an animated GIF from binary data files.")
+    parser.add_argument("input", type=str, help="Input directory containing .bin files")
+    parser.add_argument("output", type=str, help="Output GIF file")
+    parser.add_argument("--initial_duration", type=int, default=1000, help="Duration of the initial frame in milliseconds")
+    parser.add_argument("--frame_duration", type=int, default=100, help="Duration of each frame in milliseconds")
+    parser.add_argument("--cmap", type=str, default="Blues", help="Colormap to use for the frames (default: Blues)")
+    parser.add_argument("--verbose", "-v", action="store_true", help="Enable verbose output with progress")
+
+    args = parser.parse_args()
+
+    generate_gif(args.input, args.output, args.initial_duration, args.frame_duration, args.cmap, args.verbose)
+
+if __name__ == "__main__":
+    main()