Your footage exists once at capture and never again. Here's a complete photo and video storage pipeline — cards, field offload, NVMe scratch, active project storage and cold archive — built so nothing is ever a single point of failure.
A creator's data only exists once at the moment of capture, and never again. The job of a storage workflow is to get those irreplaceable files off a fragile card and into a safe, fast, well-organised pipeline — from capture to edit to permanent archive — without ever leaving a single point of failure.
Two things make creative storage harder than ordinary computing. First, the volumes are enormous and lumpy — a single day's shoot can dwarf a year of a typical user's writes, arriving all at once. Second, the data is both irreplaceable and long-lived: you may need that raw footage again years later for a re-edit or a remaster. So the pipeline has to handle a fast, bursty intake at one end and decades-long preservation at the other, with fast working storage in the middle. No single drive does all three jobs well, which is why creators think in tiers rather than in one big disk.
Everything starts on a memory card — SD or UHS-II for most cameras, CFexpress for high-bitrate 8K and rapid bursts. Two rules dominate here: buy for sustained write speed (the Video Speed Class, not the headline read number) so the camera never drops frames, and treat the card as temporary. A card in a camera bag is your only copy until you offload it, and that is the riskiest your footage will ever be. Use reputable brands; the card market is full of capacity-faking counterfeits.
The instant you can, copy cards to a rugged external SSD. Flash has no moving parts to skip or shatter when a bag gets dropped on location, and a fast USB or Thunderbolt unit clears a card in minutes. The discipline that saves careers: offload to two destinations before formatting the card. A second external SSD, or the SSD plus a laptop, gives you two copies in the field — so a lost drive on the trip home is an annoyance, not a catastrophe.
This is the one place fast storage genuinely transforms the work. Multi-stream 4K/6K timelines, cache, and render scratch hammer the drive with random and sequential I/O, and an NVMe SSD keeps scrubbing and playback fluid where a hard drive would stutter. Keep your active project, cache and media on NVMe while you cut it. Endurance matters more than for a typical user because editors write enormous volumes daily — prefer TLC NAND and watch the drive's rating, as covered in our SSD endurance guide.
| Stage | Storage | Why |
|---|---|---|
| Capture | CFexpress / SD card | In-camera; buy for sustained write speed |
| Field offload | Rugged external SSD (×2) | Shock-proof, fast, two copies before format |
| Editing scratch | Internal NVMe SSD | Cache, render & multi-stream timelines |
| Active project | NAS or fast HDD/SSD | Shared, roomy, backed up nightly |
| Archive | HDD + LTO tape / optical | Cheapest per TB; cold, off-site, immutable |
Once a shoot lands, the working copy lives on a roomy, backed-up volume — an internal HDD array, a NAS, or a fast SSD for smaller projects. For heavy 6K/8K work, a proxy workflow is the great equaliser: edit against low-resolution proxy clips that play back smoothly even from a hard drive, then relink to the full-resolution originals only at export. Proxies let modest storage handle footage that would otherwise demand the fastest drives money can buy.
Finished projects shouldn't squat on expensive fast storage. Move them to bulk HDDs for the warm archive, and once you're storing many terabytes, LTO tape — which can reach single-digit dollars per terabyte of media and sits on a shelf for decades without power — for the deep cold copy. Archival Blu-ray adds an immutable, off-site leg for your most irreplaceable masters.
Storage hardware is only half the battle; a consistent structure is what lets you find a clip two years later. Adopt a dated folder convention (year/shoot/card) the moment footage lands, keep originals untouched and read-only, and let your editor's catalog or asset manager point at that canonical copy. Avoid scattering working versions across random drives — that's how a “final” master gets lost. A tidy hierarchy also makes archiving and restoring trivial, because each project is a self-contained, copyable unit.
A creator's pipeline is only as fast as its slowest link. A blazing external SSD plugged into an old USB port runs at the port's speed, not the drive's; a fast NVMe scratch disk paired with a saturated NAS connection stutters on playback; a card reader on USB 2.0 turns a five-minute offload into half an hour. Match interfaces end to end — reader, cable, port and drive — and for multi-stream editing, confirm your machine actually has the bandwidth (Thunderbolt or fast USB) the footage demands.
None of this is safe until it's redundant. Apply the 3-2-1 rule: three copies, two media types, one off-site — from the moment of capture, not just at archive time. The field double-offload is your early 3-2-1; the archive stage extends it for the long term. Compare every drive in this pipeline by real value in our $/TB rankings, and gauge how much capacity your shoots demand with the capacity calculator.
Shock-proof, fast portable drives to copy cards onto twice before you format.
Compare cards, external SSDs, NVMe scratch and archive drives by cost per terabyte.
Open the $/TB rankings →