I have a friend who’s a singer. He showed me how much storage 1 4 min music video took. 40 min per camera, 3 cameras, 4K. Unedited raw footage came out at ~2.5TB. It’s just insane to me. My PC has 1TB of storage.
4K = 3840 x 2160 pixels = 8,294,400 pixels per frame
Assuming 8 bits of color depth (lowest common depth) = 66,355,200 bits per frame (66MB)
At 24fps (again, low end as mentioned by another commenter) = 1,592,524,800 bits per second (1.5GB)
95.5GB per minute of footage per camera, assuming literally no compression.
95.5GB × 40 = 3.8TB × 3 cameras = 11.4TB
Raw uncompressed video takes up a LOT of space.
(Yes I used 1000 instead of 1024 for convenience)
Well, TB, GB, MB are decimal (powers of 10). TiB, GiB, MiB are binary (powers of 2). So that’s correct.
Yeah but that’s a BS measure they created decades later. We never used MiB in the 80s and 90s.
To clarify, it was always powers of 2 in the old days.
Kilo and mega prefixes have been used for far far longer than computers have been around and they mean 1000 and 1000000. It’s only in computing where these prefixes were corrupted because they were deemed “closed enough” and caused confusion because people expect the terminology they use everywhere else to also apply to their computer.
You’re not wrong- but when the change happened, it seemed to be around the time HDD manufacturers started taking flak for selling high capacity drives that didn’t match formatted capacity by larger and larger margins. OS level files sizes were always 2^n - it was a real PitA when Apple switched to base 10 for gui size reporting. I believe terminal utilities still measure file size in 1024 bytes.
Formatted size will always be smaller than raw size as you are storing a file system and there are lots of different ones with different space requirements.
Storage is measured in binary base. All other measurements are marketing.
Except the marketing in this case is actually more accurate. Using ‘kilo-’ to refer to 1024 of something is a straight up misnomer. The Ancient hackers really cursed us with an ambiguous convention for counting bytes.
Thanks, I was too lazy to explain that haha
Wait til you hear how big the raw footage from a major picture is.
It’s ~69MB for DIN A3 with 32bit depth.
It’s 0.06TB for 10 metres times 10 metres, 300dpi, 32 bit
… I think he meant a different kind of major picture. But still, your math checks out.
Oh, are movies called major pictures?
“Major motion picture” is a term thrown out there for big studio productions as opposed to made for tv or home movies.
Wait until you record in 8k, or use 20 cameras like during a sporting event.
And people wonder why CCTV footage is always low res and grainy. I worked at a site with something north of 100 cameras and a requirement for 30 days storage. Their server room was mostly just racks and racks of hard drive chassis all wired up to the DVR system.
What’s the size per day? High end cinema here has 30TB/day coming off set.
Which is why there’s almost no 4k live sports. The processing is just insane today.
Almosteverything you ingest from day to day on your PC, in terms of multimedia content, is compressed.File compression is a necessity, as it makes transmission of such media over the internet much more teneble. But uncompressed video carries a lot more information – not necessarily all useful information, but it is there.
4k with 10b/pixel (assuming, could be 12) x 60fps (assuming) x 360s (4min) x 3 cameras:
3840 * 1920 = 7,372,800 pixels
7372800 * 3 channels (RGB) = 22118400 values
22118400 * 10/8 = 27648000 bytes
27648000 * 60 = 1,658,880,000 bytes/sec
1658880000 * 360 = 597,196,800,000 (0.6TB for 4 minutes)
.6TB * 3 cameras = 1.8TB
RAW means RAW, no encoding, no cute channel games, no nothing, just fuck-off I/O.
edit: My bad, that’s 6 minutes, so assume 12 bits per second and you get the same size for 5 minutes.
Modern compression algorithms are quite literally just magic at this point.
edit2: FUCK! Forgot audio!
So… 192khz * 32bits/8(bits per byte) = 768,000 bytes/sec * 300 (5 minutes) = 230,400,000, so 230MB, doesn’t really move the needle.
Something i don’t see mentioned anywhere is that mysic videos typically are recorded at 1/3 of the normal speed. So the original files will be three times as long as you think would be necessary.
I have a Blackmagic 6K camera that records RAW, and a ninja monitor that also records RAW. My 2TB SSD doesn’t hold much footage.
psa: use ffmpeg to re encode smartphone videos for massive space savings
If you don’t mind the (slight but nonetheless present) quality loss. Not quite worth it to me since I can always copy large videos off my phone to free up space.
Storage space requirements are insane once you’re not talking about end product. In this case, you need the full recording for every angle, and if you’re actively editing you probably want a buffer as well.
Another example with the same principle: Geophysical survey data. The end product is usually a bunch of SEG-Y files with associated JPEGs, totallengde a few gigs. But the raw survey data is more close to a petabyte: It’s s8mewhat reasonable to think of it as an audio recording consisting of four mono channels * 1000 recording stations * a month duration.
Raw data adds up fast.
4K really weighs heavy in matter of storage space, just multiply 1080p by 4. If you can get that the raw rushes for a 1080p clip would be around 500 Go, then it makes sense that the same in 4k is 4 times bigger. My friends who stores a lot of rushes buys a new hard drive each year or two now that he uses 4k
I mean, do the maths. A high res picture is 8-15 MB in size, and a 4k video has 60 fps (frames per second). Let’s take 12 MB as average picture size, that gives you 12MB x 60 frames x 60 seconds = 43.2 GB/min. With some advanced compression algorithms you can maybe get the individual frame size down to half, but that’s really it.
