One conversion runs the whole thing
Camera manufacturers publish bitrates in megabits per second. Card and drive manufacturers sell capacity in gigabytes. The conversion between them is the only piece of arithmetic in this entire subject, and it is worth memorising: multiply megabits per second by 0.45 to get gigabytes per hour. That factor is 3,600 seconds in an hour, divided by 8 bits in a byte, divided by 1,000 megabytes in a gigabyte.
So a 100 Mbps codec eats 45 GB an hour. A 400 Mbps intermediate codec eats 180. A 1,900 Mbps high-end acquisition format eats 855 GB an hour — well over a terabyte for a ninety-minute interview, per camera.
Everything else on this page is multiplication on top of that. Hours of footage gives project size. A proxy workflow adds a percentage. The number of copies you keep multiplies the whole thing, and this is where budgets go wrong: a 3-2-1 backup policy does not add a bit of overhead, it triples your storage requirement. A 200 GB project is a 600 GB storage commitment, and it stays that way for as long as you keep the material.
Note that shooting more cameras multiplies too. Three cameras rolling for four hours is twelve hours of footage, not four, and this calculator wants the total across all of them.
Getting each input right
Take the bitrate from the camera's specification sheet, for the exact combination you will shoot. It is not one number per camera — it changes with codec, resolution, frame rate, chroma subsampling and bit depth, often by a factor of ten across a single camera's modes. A body that records at 100 Mbps in its efficient mode may record at 800 Mbps in its high-quality intra-frame mode, and the choice you make in the menu on the day determines whether your cards last the shoot.
If your camera quotes megabytes per second instead, switch the unit on the input rather than converting in your head; a factor-of-eight error here propagates into every figure on the page.
Hours of footage is total recorded material, across every camera. If you do not know it, estimate from your shooting ratio: finished runtime multiplied by the ratio you expect. A 12-minute video at 6:1 is 72 minutes of footage — the cost per finished minute calculator reports that ratio, and it is a cost driver in post as well as in storage.
Proxy overhead covers the low-bitrate editing copies many workflows generate. Their size depends entirely on the proxy codec you choose, so measure it once on a real project rather than trusting a default. The useful property of proxies is that they are regenerable: they can be rebuilt from the originals at any time, so they do not need to exist in every backup copy. Excluding them from the off-site copies is free storage.
Copies is the total number of copies of the material that exist, not the number of extra backups. The 3-2-1 rule, popularised by Peter Krogh in The DAM Book, means three copies, on two different media types, with one kept off-site — so enter 3, not 2.
Worked example: a four-hour shoot at 100 Mbps
A 15% proxy overhead, three copies kept, 128 GB cards, and archive storage at $20 per terabyte.
- Gigabytes per hour. 100 Mbps × 3,600 s = 360,000 megabits. ÷ 8 = 45,000 MB. ÷ 1,000 = 45 GB/hr.
- Camera originals. 4 hr × 45 = 180 GB.
- With proxies. 180 × 1.15 = 207 GB.
- Total archive. 207 × 3 copies = 621 GB.
- Recording time per card. 128 GB ÷ 45 GB/hr = 2.844 hr × 60 = 170.7 minutes, so a card lasts a little under three hours.
- Cards needed. 180 ÷ 128 = 1.41, rounded up to 2 cards.
- Archive cost. 621 GB ÷ 1,000 = 0.621 TB × $20 = $12.42.
Two cards for a four-hour day sounds comfortable until you notice the shape of it: the second card is only 41% used, and a single overrun or an unplanned second camera puts you over. Media is the cheapest insurance on a set, and the cost above says why — the entire three-copy archive for this shoot costs less than lunch.
Now shoot the same day at 400 Mbps instead. GB per hour becomes 180, the originals become 720 GB, the archive becomes 720 × 1.15 × 3 = 2,484 GB, and a 128 GB card now records 128 ÷ 180 × 60 = 42.7 minutes. The storage bill quadruples and the card swap moves from once every three hours to once every forty minutes — which changes not just your budget but how the day is run.
Reading the numbers on a real shoot
Recording time per card is the operational number. It decides whether media handling is a background task or a job someone must be assigned to. Below about half an hour per card you are swapping during takes, which means a dedicated person, a rigid labelling system, and a genuine risk of losing a card in the shuffle. Above two hours it can be handled between setups by whoever is nearest.
Card count should always be rounded generously. The calculator rounds up to the nearest whole card, but that assumes you fill each one completely and that nothing goes wrong. Real shoots overrun, cameras get added, and a card that develops an error mid-day cannot be trusted for the rest of it. Carrying double the computed number is normal practice and costs very little relative to reshooting.
The archive figure is a recurring commitment, not a one-off purchase. That 621 GB has to exist somewhere for as long as the project might be revisited, on media that itself has a lifespan. Drives fail, and a copy nobody has verified is not a copy. Budget for periodic verification and for migrating the archive to new media every few years.
Finally, sanity-check the sustained write speed as well as the capacity. A card rated for a peak burst may not sustain a high-bitrate intra-frame codec, and the failure mode is a recording that stops mid-take. Above roughly 2,000 Mbps this becomes the binding constraint rather than capacity, and the number to check on the card is its sustained rating, not the large number on the label.
Bitrate, storage rate and card endurance
| Bitrate | GB per hour | 64 GB card | 128 GB card | 512 GB card |
|---|---|---|---|---|
| 25 Mbps | 11.25 | 341 min | 683 min | 2,731 min |
| 50 Mbps | 22.5 | 171 min | 341 min | 1,365 min |
| 100 Mbps | 45 | 85 min | 171 min | 683 min |
| 200 Mbps | 90 | 43 min | 85 min | 341 min |
| 400 Mbps | 180 | 21 min | 43 min | 171 min |
| 800 Mbps | 360 | 11 min | 21 min | 85 min |
| 1,200 Mbps | 540 | 7 min | 14 min | 57 min |
| 1,900 Mbps | 855 | 4 min | 9 min | 36 min |
Read down any card column and the endurance halves exactly as the bitrate doubles. The bottom-left cell is the one to notice: at high-end acquisition bitrates a 64 GB card holds a single four-minute take.
What this calculation leaves out
- Decimal versus binary gigabytes. Cards and drives are sold in decimal GB of 1,000 MB; operating systems report binary GiB of 1,024 MiB. A 128 GB card shows as about 119 GiB, roughly 7% less than the label, and this calculator uses the decimal convention the labels do.
- Filesystem overhead and formatting. A freshly formatted card offers slightly less than its rated capacity, and some cameras reserve space for metadata and clip structure.
- Variable bitrate codecs. Long-GOP formats vary with scene complexity, so a quoted bitrate is a maximum or an average rather than a guarantee. Intra-frame codecs are much closer to constant.
- Audio and metadata tracks. Multichannel audio, timecode and camera metadata add to the file, typically a small percentage but not zero on low-bitrate video.
- Render files, caches and colour-managed exports. An edit generates conform renders and cache files that can rival the proxy overhead, and they live on the working drive rather than the archive.
- Sustained write speed. Capacity and speed are different specifications. A card with plenty of room can still fail to record a high-bitrate codec if its sustained rate is below the demand.
- Archive medium lifespan. Drives fail and optical media degrade. A three-copy policy assumes those copies are verified periodically and migrated to fresh media over the years.
Where storage sits among the other production constraints
Storage is the cheapest thing on a shoot and the most expensive thing to have got wrong. The archive in the worked example costs about twelve dollars, against a production day whose labour cost runs into the thousands — a comparison the cost per finished minute calculator makes concrete. There is no version of this arithmetic in which skimping on media or copies is a rational saving.
What storage decisions do affect materially is time. A higher bitrate means more data to offload, more to back up, more to hand to an editor and more to keep. Ingest and verification of a large card is not instant, and a shoot that generates two and a half terabytes has a real offload window built into its schedule. The related delay at the other end of the pipeline — export and upload — is covered by the video export and upload time calculator.
The bitrate choice itself is the same trade-off that governs live streaming, with the network constraint replaced by a capacity constraint: how many bits per pixel does this content actually need? The reasoning is worked through in the livestream bitrate calculator, and it applies to acquisition codecs too — heavy motion and fine detail justify the higher setting, a locked-off interview generally does not.
Key terms
- Codec bitrate
- Data rate the camera writes, in megabits per second. It varies within one camera with codec, resolution, frame rate, bit depth and chroma subsampling.
- Proxy
- A low-bitrate copy of the footage used for editing on modest hardware. Regenerable from the originals, so it does not need to be preserved in every backup copy.
- 3-2-1 rule
- Keep three copies of the data, on two different media types, with one copy off-site. Popularised by Peter Krogh in The DAM Book and now standard practice in media and IT backup.
- Sustained write speed
- The rate a card can maintain indefinitely, as distinct from its peak burst rate. High-bitrate intra-frame codecs are limited by this rather than by capacity.