Subtitle Gif Generator

NodeJS command-line utility to generate short looping clips with subtitles from your favorite videos. This is a fun tool for creating looping scenes from your favorite, most quotable shows and movies!

Installation

You have two options on how to install these scripts:

1. Clone the project and run the scripts directly

   git clone https://github.com/josephschmitt/sub-gif-gen.git
   cd sub-gif-gen
   npm install

2. Install the package into another NodeJS project

   npm install @joe-sh/sub-gif-gen --save-dev

Regardless of how you install the project, you'll have to make sure to use a version of the ffmpeg binaries with freetype enabled so that we can render the subtitles without the need for any other dependencies. On a Mac, you can do this by building ffmpeg from source with this option using homebrew:

brew install ffmpeg --with-libass --with-fontconfig --with-freetype

Or using this repo's Brewfile:

brew bundle

Usage

Running from the cloned project:

./scripts/processVideos.js -d path/to/videos -- path/to/gif/output

If you installed as an npm dependency, you can either run in a package.json script:

{
  "dependencies": {
    "sub-gif-gen": "^1.0.0"
  },
  "scripts": {
    "process": "gen-gifs -d path/to/videos -- path/to/gif/output"
  }
}

Or run the script directly in your shell:

./node_modues/.bin/.gen-gifs -d path/to/videos -- path/to/gif/output"

CLI Flags

Required

• -d, --dir: Directory containing your video files. Currently limited to only reading from external .srt files for subtitles. The .srt file must be named the same as the input source, but with the filename set to .srt.

Optional Parameters

• -l, --lang: Language code if your srt subtitle files are named {filename}.{lang}.srt. Defaults to 'en'.
• -o, --offset: Amount of time (in seconds) to offset the gif. By default this is zero, which means the gif is cropped to the exact timecode of the subtitle. You can use this value to extend the time of the clip by adding some time before and after the timecode from the subtitle file.
• -x, --extensions: Allowed extensions for the input file. Will use these to filter the input directory for videos. Defaults to .mkv,.mp4,.m4v,.mov.
• -r, --formats: The output formats for the looping clips. Defaults to gif. Can supply a comma-separated list of formats. Supports gif,mp4,webm.

Optional Flags

• -f, --flatten: If set, will output all the output files to the given output directory. Otherwise it will group files into folders by the input filename. Defaults to false.
• -k, --skipExisting: If set, will skip processing if it finds a gif file already exists in the output directory.
• -s, --sanitize: If set, will sanitize output filenames from their original names to be url-safe using the url-safe-string package.

Use the -- flag to denote the end of the options and then pass the directory to output your gifs. The gifs will be output to a directory of the same name as the input file. The gifs are named the same as the input source file, plus the start-time in milliseconds of the clip.

Environment Flags

To provide a custom location for your ffmpeg binary, set the FFMPEG_BIN environment flag:

env FFMPEG_BIN='path/to/ffmpeg' ./scripts/processVideos.js -d path/to/videos -- path/to/gif/output

The default uses the ffmpeg exported in your PATH.

If something goes wrong, you can set a LOGLEVEL flag to see more verbose output:

env LOGLEVEL=verbose ./scripts/processVideos.js -d path/to/videos -- path/to/gif/output

Index Generation

In addition to the gifs, the script produces a JSON file for each video of structured information about the subtitles and what gifs they belong to. You can use this information along with the scripts/createIndex.js script to produce a single structured data file of all your gifs and their subs. This could be used to generate a data structure compatible with cloud-based search indexers, such as AWS's CloudSearch.

The indexer takes in an art template and a glob of JSON files to produce the index from. The glob should point to the .json index files produced by the scripts/processVideos.js script, and you should use this data to produce a file that can be ingested by your cloud-based search indexer. You can find an example template compatible with AWS's CloudSearch in the templates/ directory.

./scripts/createIndex.js --template templates/aws.cloudsearch.art --indexes **.json -- index.json

You'd then be able to upload index.json to AWS's CloudSearch to create a searchable gif index.