Hacker News new | ask | show | jobs
by mjrpes 485 days ago
It's interesting what this situation would be like if HDD capacity hadn't stopped its exponential growth in 2010: https://imgur.com/a/lWdcjX7

We would be at a penny per terabyte of space. If AV1 in HD can store 400 hours of video per TB, the roughly 24TB to store a 24/7 stream over the course of a year would cost only 25 cents. Providers could keep all video content indefinitely.

Perhaps there's some benefit to this exponential growth coming to an end. Imagine a surveillance state that had near limitless storage and could keep 24/7 recordings indefinitely of cameras on every street, house, vehicle, etc.
2 comments
Dylan16807 484 days ago
Low quality video mixed with some key snapshots and full audio would only be about half a terabyte per year. So even with current pricing, a surveillance state can easily pay $3-10 dollars to store that if it wants to.
link
fc417fc802 484 days ago
Unfortunately that remains a concern. The current research on ML based video codecs is yielding almost unbelievable size results.
link
pdimitar 484 days ago
Where can I read more?
link
fc417fc802 484 days ago
I guess this [0] might be a decent starting point. Also consider size comparisons of various gaussian splatting techniques (ex [1] & [2]). Static 3D scenes aren't technically the same thing as a 2D time series of frames, but since videos are generally composed of highly spatially related data there's bound to be overlap (ex [1] has already been extended to support dynamic scenes).

And keep in mind that security footage has an awful lot of both static and repetitive objects. It's an ideal target for more complicated approaches to compression.

[0] https://github.com/ppingzhang/Awesome-Deep-Learning-Based-Vi...

[1] https://maincold2.github.io/c3dgs/

[2] https://jasonlsc.github.io/nerfcodec_homepage/

link
pdimitar 484 days ago
Wow those are some extreme savings!

Thank you, your comment went straight to favorites.

link