Open DunklerPhoenix opened 2 years ago
Hi! Adding ffmepg increases the image size by something around 400mb if I remember correctly and since this is pretty optional, I decided to leave it for the time being. Uploading should work without ffmepg, but there won't be any validation.
okay. You could either create a second image with ffmpeg or you can add an option to the entrypoint script that will add ffmpeg (and additional packages which are passed by env. The second solution is for example used by photoprism. There you have a field in the config file where you can add additional packages.
Yes, we could do something like that. I'll take a look at photoprism
You can set the Env (e.g.) PHOTOPRISM_INIT: "gpu, tensorflow, nano"
Here is the entry for it https://github.com/photoprism/photoprism/blob/cb89a1a189aede610d4f8c7455232d299c1a4dca/scripts/dist/entrypoint-init.sh#L68
and there are the init scripts: https://github.com/photoprism/photoprism/tree/cb89a1a189aede610d4f8c7455232d299c1a4dca/scripts/dist
In the dockerfile is ffmpeg missing and should be added for video functionality.