Hosting on AWS Elastic Beanstalk and S3

I copy this quote from @bastianallgeier

We used a file upload hook that uploaded the original to S3 and afterwards replaced the file on disk with a 1px spacer. The S3 ID, original width and height were stored in the meta text file and the URL component would then use that ID from the text file to create the correct URL. This way, Kirby will still think it has normal files on disk, while they are actually tiny and don’t take up much space at all. But the originals are all in S3