A web frontend FFmpeg wrapper for easy, user-friendly generation of timelapses from image sequences.
Output files are 1080p MP4 files intended to work with Adobe Premiere for further editing.
This web frontend is well-suited to run on an on-site server. In my configuration, it runs on a local machine that also hosts a NAS. A network client can then process timelapses directly on the NAS.
This project is used for the HikeArt YouTube Channel.
The easiest way to start using this project is by installing the public Docker image. The image is automatically built from this repository using Travis CI.
If you're not familiar with running a Docker container see: https://docs.docker.com/get-started/
To download and run the timelapse-queue image:
# Point this at a directory containing timelapses you wish to process.
# The path must be absolute.
TIMELAPSE_DIR=/home/jeff/
docker run \
-p 8080:8080 \
--mount type=bind,source=${TIMELAPSE_DIR?},target=/mnt/fsroot \
jheidel/timelapse-queueOnce it's running, navigate to http://localhost:8080.
TODO: provide some guidance for running this as a service.
Timelapses are configured via a web application, written using Polymer 3.
The backend is a monolithic Go binary which is responsible for serving the web frontend as well as processing the timelapses.
The timelapse server reads images sequentially and applies any resizing & cropping & image stacking effects. The images are then piped to a FFmpeg binary run as a subprocess which writes out the video file. The server manages queueing of jobs so that at most one FFmpeg operation is running at a time.
- Run
make - Resolve errors (probably missing dependencies)
- Repeat until success
This project was developed on Ubuntu 18.04 (and will likely work with other platforms with a couple rough edges).
It has only been tested on Google Chrome and may encounter issues with other browsers.
To build the Docker image:
docker build -t timelapse-queue .
Also available on Docker Hub:
https://hub.docker.com/r/jheidel/timelapse-queue/
Initial implementation complete. Some further development is planned as I have time, mostly for an archival feature to convert large image sequences into a high quality 4k archive format which can later be re-processed into timelapses.
A timelapse job starts with a file browser. Images are grouped into timelapse sequences.
The timelapse job then requires some simple configuration.
Photo stacking effects are available.
Timelapse jobs go into a queue and are processed sequentially using FFmpeg.
