[thrdbndndn]

I don't get how it works.

> Encoding: Files are chunked, encoded with fountain codes, and embedded into video frames

Wouldn't YouTube just compress/re-encode your video and ruin your data (assuming you want bit-by-bit accurate recovery)?

If you have some redundancy to counter this, wouldn't it be super inefficient?

(Admittedly, I've never heard of "fountain codes", which is probably crucial to understanding how it works.)

[Jaxan]

Yes it is inefficient. But youtube pays the storage ;-). (There is probably a limit on free accounts, and it is probably not allowed by the TOS.)

[genidoi]

Right, you just pay daily in worrying when, not if, youtube will terminate your account and delete your "videos".

[brandonli28]

Hey there, Brandon here (developer). I've uploaded an explanation video here, which might be useful to watch :D

https://youtu.be/l03Os5uwWmk?si=nJDwz4s7_E4WFOwC

[grishka]

He encodes bits as signs of DCT coefficients. I do feel like this is not as optimal as it could be. A better approach IMO would be to just ignore the AC coefficients altogether and instead encode several bits per block into the DC. Not using the chrominance also feels like a waste.

[brandonli28]

This actually won't work against YouTube's compression. The DC coefficient is always quantized, rounded, scale, and any other things. That means that these bits are pretty much guaranteed to be destroyed immediately. If this is the case for every single block, then data is unrecoverable. Also, chrominance is not used on purpose, because chrominance is compressed much more aggressively compared to luminance.

[sdenton4]

Yeah, I would assume that transcodes kill this eventually...