Render Frame Sequences

A collection of scripts we use to take a large batch of frame sequences and render them into lossless video files.

License

These scripts are provided under an MIT license. They are not supported by Thinko. Use at your own risk.

Setup

• All frames should be in directories like ./input/<ID>, where "ID" represents a shot number.
• Sequences will be automatically resorted to be sequential, starting from 0, matching the format <ID>_<5-zero padded sequence>.png
• Sequences will be output at ./output/<ID>.mov

If you get a permissions error running the scripts, try making them executable:

chmod u+x *.sh

Process

• Populate the ./input directory with numbered sub-directories containing shots to render
• cd to this folder and run ./loop.sh

Notes

Frame sequences are renamed in place, so for maximum safety you should run the ./loop.sh script on a separate copy.

If an output file already exists at ./output/<ID>.mov, that frame sequence will be skipped. This allows you to re-run the script easily. If you want to rerender a sequence, just delete the output and run ./loop.sh again.

If you exit the script while ffmpeg is running, the incomplete render will not be cleaned up. And, given the previous point, will not be re-rendered. Be sure to clean up manually if this happens.

Frame sequences will only be resorted if both of the following criteria are not met:

• input/<ID>_<00000>.png is present
• input/<ID>_<zero-base frame count>.png is the final file (e.g. if there are 50 frames, the final file is <ID>_00049.png)