Comment by fatratchet | Hacker News Reader

Comment by fatratchet | Hacker News Reader

Parent Full thread

To get reliable and free storage, photo hosting is usually the easiest way. Flickr offers 1TB, picasa/g+ offers unlimited storage with some hidden quoats. Everything that allows lossless photos lets you store arbitrary data. Depending on how careful you wanna be you can store hundreds of GBs per account.

Email attachments used to be a great way a while ago but nowadays using multiple gdrive/dropbox/onedrive accounts is much easier.

They are easy to create in large numbers (especially if your ISP has dynamic IPS) and as long as you're even a little bit careful, nearly impossible to ban. Add some redundancy across different services to that and a $2 VPS that gives you tons of upload bandwidth and you've got yourself as many TBs of free,fast and reliable online storage as you want.

I spent so much time as a teenager with no money and some python skills coding storage solutions like that. I'd say it was to store movies and tv shows for myself but in retrospect I mostly did it because it was so much fun to develop.

Video hosting (i.e. YouTube) is another potential repository for massive amounts of data.

Combine that with the fact that data which is encrypted looks practically like static, and you could potentially overlay it on top of an existing video of something mundane.

You'd need to use strong ECC to get past the lossy encoding, but as things like QR codes show, that is not so hard.

The audio channel is also usable...

Trying to get the greatest entropy possible through arbitrary-strength JPEG compression would be an interesting problem to solve.

The new Google Photos storage is lossy unless you pay for it. That doesn't rule out using the images in a different way though.

I thought it was only lossy if the originals you uploaded were >16MP. I tested uploading some <16MP images and redownloading them, and they didn't seem to have undergone any lossy conversion.

Did you compare a hash of the file, or did they just look the same?

I've done similar with images for the fun of it. The simplest solution that I recall finding was to base64 the file/data, then turn to hex, then use those hex data to create pixels in RGB. I would line them up top-left to bottom-right.

Probably not the most efficient but easy and fast and the resulting images would look... interesting. For large files, the decoding would be difficult mostly just due to reading the image of so many pixels into memory. So, that's when I began fixing the image size to a smaller size and having multiple images that I would later convert to 60fps video. I could then use ffmpeg to convert images to frames and frames back to images.

I had no practical use for this but, was a fun project on a rainy afternoon.

Yeah, I wrote something that stores data to Flickr last summer: https://github.com/namwen/hoardr . I kind of had a reason but it was more for the enjoyment of getting it to work.

There is also hyperglobalmegastore https://github.com/adrian-bl/hyperglobalmegastore All data is encrypted and you can even mount your flickr 'drive' using fuse.

This project needs more stars!

Until this gets so popular that Google or Flickr start to analyse photos, and come to the conclusion to either delete those photos and videos, or to convert them and destroying the data for you. Then, years later, you need your backup and ....

I did the same thing for google photos, but just for test purpose. https://photos.google.com/album/AF1QipOjZrywipm-SSH9jVNsKVF5...

The front-end is already there : https://tahoe-lafs.org/trac/tahoe-lafs. Backends are currently being developed (https://github.com/mk-fg/tahoe-lafs-public-clouds), and there will even be a public offering from the very same guys (https://leastauthority.com/)

Reply on news.ycombinator.com