I'll create a Python script that automates the video creation process using FFmpeg, incorporating all the requirements you've specified. This script will use the subprocess module to run FFmpeg commands, os and random modules for file handling, and ffmpeg-python to get the audio duration.

Here's the Python script that accomplishes your task:

This script does the following:

1. It uses the ffmpeg-python library to get the duration of the voiceover audio file.
2. It scans the "images" directory for .png, .jpg, and .jpeg files, and randomly shuffles them.
3. It calculates the number of images needed based on the voiceover duration and the specified image duration (10 seconds).

4. It prepares the FFmpeg command, including:

   • Input for each image, looped for 10 seconds.
   • A complex filter that applies the zoom effect to each image and concatenates them.
   • The voiceover audio input.
   • Output video configuration, including NVIDIA GPU acceleration (h264_nvenc), resolution, and framerate.
5. It executes the FFmpeg command using subprocess.run().

To use this script:

1. Install the required Python library:

   pip install ffmpeg-python
2. Ensure you have FFmpeg installed on your system and accessible from the command line.
3. Place your images in an "images" folder in the same directory as the script.
4. Place your "voiceover.mp3" file in the same directory as the script.

5. Run the script:

   python create_video.py

The script will create a video named "output.mp4" in the same directory, matching the duration of the voiceover, with images changing every 10 seconds and a slow zoom-in effect applied to each image. The video will be encoded using your NVIDIA RTX 3060 GPU for faster processing.

Note: If the voiceover is shorter than the total duration of all images, the video will be trimmed to match the voiceover length. If the voiceover is longer, images will loop to fill the entire duration.

I understand. I'll modify the script to address these two issues. Here's the updated version of the script:

Here are the changes I've made to address the issues:

1. Output Directory Issue:

   • I've added a constant OUTPUT_DIR that points to an "output" folder in the same directory as the script.
   • The script now creates this directory if it doesn't exist using os.makedirs(OUTPUT_DIR, exist_ok=True).
   • The output video is now saved in this "output" folder.

2. Image Selection Issue:

   • Instead of shuffling the image list once at the beginning, the script now randomly selects an image for each 10-second segment.
   • This is done using random.choice(image_files) inside the loop that prepares the FFmpeg inputs.

Additional improvements:

• I've used os.path.dirname(os.path.abspath(__file__)) to get the directory of the script, ensuring that all paths are relative to the script's location.
• All file paths now use os.path.join() for better cross-platform compatibility.

To use this updated script:

1. Ensure you have the required Python library installed:

   pip install ffmpeg-python
2. Make sure FFmpeg is installed on your system and accessible from the command line.
3. Place your images in an "images" folder in the same directory as the script.
4. Place your "voiceover.mp3" file in the same directory as the script.

5. Run the script:

   python create_video.py

The script will create an "output" folder (if it doesn't exist) and save the final video as "output.mp4" inside this folder. Each 10-second segment of the video will now use a randomly selected image from the "images" folder, addressing both issues you mentioned.