[338 / 53 / ?]
Quoted By: >>51072449 >>51072484 >>51072524 >>51072536 >>51072570 >>51072724 >>51072830 >>51072879 >>51073040 >>51073121 >>51073474 >>51074020 >>51074601 >>51074748 >>51076048
Continued from
>>51062769
Original post:
What if you could use youtube as your own cloud backup solution?
I've been experimenting a little, an when you take into account the artifacts caused by youtubes compression algorithm, you can fit 2x 40L QR codes per 1080p frame and stay within the noise profile needed to read the QR code. 40L has a numeric capacity of 7089 characters. So you get 14178 numeric characters per frame. Assuming I store my data in binary, I can store 1772,25 byte = 1,77 KB/frame.
Youtube allows me to upload videos that are upto 11 hours in duration and 128 GB in size.
Assuming 60fps, that's 60*39600=2376000 frames per video = 4.20552 gigabytes/video.
All that's left is making a winrar archive of the data, encrypt it and split it into 0,885 KB chunks. Convert each chunk into a 40L QR code, assemble 2 each into a 1080p frame, and assemble each batch of 2376000 frames into an 11 hour video. Upload to YouTube and voila, free storage! Repeat ad infinitum or until you've uploaded all your data. To retrieve your data, simply download the videos, strip the frames, read the QR codes and assemble binary info into file. Extract and enjoy.
Does anyone know of any better barcode systems I could use? QR isn't very effective, but it survives youtubes lossy compression. I've been thinking about using colored pixel clusters, but I need to take into account the chroma subsampling. I'm sure the efficiency could be improved by an order of magnitude by a more efficient barcode though.
-----
Updates? Progress? Let's pick this up where we left it off yesterday.
>>51062769
Original post:
What if you could use youtube as your own cloud backup solution?
I've been experimenting a little, an when you take into account the artifacts caused by youtubes compression algorithm, you can fit 2x 40L QR codes per 1080p frame and stay within the noise profile needed to read the QR code. 40L has a numeric capacity of 7089 characters. So you get 14178 numeric characters per frame. Assuming I store my data in binary, I can store 1772,25 byte = 1,77 KB/frame.
Youtube allows me to upload videos that are upto 11 hours in duration and 128 GB in size.
Assuming 60fps, that's 60*39600=2376000 frames per video = 4.20552 gigabytes/video.
All that's left is making a winrar archive of the data, encrypt it and split it into 0,885 KB chunks. Convert each chunk into a 40L QR code, assemble 2 each into a 1080p frame, and assemble each batch of 2376000 frames into an 11 hour video. Upload to YouTube and voila, free storage! Repeat ad infinitum or until you've uploaded all your data. To retrieve your data, simply download the videos, strip the frames, read the QR codes and assemble binary info into file. Extract and enjoy.
Does anyone know of any better barcode systems I could use? QR isn't very effective, but it survives youtubes lossy compression. I've been thinking about using colored pixel clusters, but I need to take into account the chroma subsampling. I'm sure the efficiency could be improved by an order of magnitude by a more efficient barcode though.
-----
Updates? Progress? Let's pick this up where we left it off yesterday.
