Introduction
The shoot wrapped at 11 PM. The DIT has 1.2 TB on four CFexpress cards that need to be offloaded and verified before the cards go back to the camera department for tomorrow's 6 AM call. The card reader is CFexpress Type B at 2,000 MB/s theoretical, connected via USB 3.2 Gen 2 at 1,250 MB/s theoretical. The bottleneck is the USB connection, and at 70% real-world efficiency, the actual transfer speed is about 875 MB/s. Total offload time: 23 minutes. With verification (read-back), double it: 46 minutes. The DIT has plenty of time.
But what if the USB port fails and the fallback is USB 3.0? Suddenly the offload takes over 40 minutes per card, and the DIT is working until 1 AM. Knowing your transfer times in advance lets you plan for the worst case.
The media offload time calculator shows how long any file transfer takes across every common interface: USB 2.0 through USB4, Thunderbolt 3/4/5, SATA, NVMe, gigabit through 25 GbE network, and every camera card format.
What This Tool Calculates
The calculator takes a file size (in MB, GB, or TB), number of cards or transfers, connection interface, and a real-world efficiency percentage.
It returns the real-world transfer speed, estimated transfer time, time with read-back verification (doubled), and theoretical maximum time. A comparison table shows how long the same transfer would take across every interface, sorted from fastest to slowest.
The Formula and How It Works
Transfer time equals file size divided by transfer speed. The critical nuance is that theoretical maximum speeds are never achieved in practice due to protocol overhead, controller limitations, drive speed bottlenecks, and file system overhead.
Transfer Time = File Size (MB) / Real-World Speed (MB/s). Real-World Speed = Theoretical Maximum * Efficiency Factor.
Typical efficiency factors: USB 3.0 achieves 60-75% of its 625 MB/s maximum. Thunderbolt 3 achieves 70-80% of its 5,000 MB/s maximum. 10 GbE achieves 80-90% of its 1,250 MB/s maximum. CFexpress Type B readers achieve 70-85% of their 2,000 MB/s maximum.
The bottleneck is always the slowest component in the chain. A 2,000 MB/s CFexpress card in a reader connected via USB 3.0 (625 MB/s max) is limited to the USB 3.0 speed. A fast NVMe SSD connected via Thunderbolt 3 (5,000 MB/s) to a destination HDD (150 MB/s) is limited to the HDD speed.
Verification adds a complete read-back pass, effectively doubling the transfer time. For critical production data, verification is non-negotiable because a corrupted offload with no verification and no original card is unrecoverable.
Real-World Examples
DIT Offload on a Commercial Shoot
A DIT offloading 512 GB from two CFexpress Type B cards needed to estimate turnaround time for the AD. The card reader connected via Thunderbolt 3. The calculator showed: 1,024 GB total at 70% efficiency via Thunderbolt 3 (3,500 MB/s real-world) = 4 minutes 53 seconds. With verification: 9 minutes 46 seconds. The DIT quoted the AD "under 15 minutes including verification," which fit comfortably within the lunch break offload window.
NAS Transfer Over 10 GbE Network
A post facility needed to transfer a 4 TB DaVinci Resolve project from the colorist's workstation to the shared NAS over 10 GbE. The calculator showed: 4,000 GB at 85% efficiency via 10 GbE (1,062 MB/s real-world) = 62 minutes. The colorist started the transfer before leaving for the day. Had the facility been running 1 GbE, the same transfer would have taken 10.7 hours, arriving well into the next morning. The 10 GbE upgrade paid for itself in time savings within the first month.
Field Backup to USB HDD
A documentary crew in a remote location needed to back up 256 GB of RED footage to a portable USB HDD each night. The calculator showed: 256 GB via USB 3.0 at 70% efficiency (437 MB/s real-world), but limited by the HDD's 150 MB/s write speed = 28 minutes. The HDD speed was the bottleneck, not the USB interface. Upgrading to an SSD would have reduced the transfer to under 8 minutes, but the crew's budget dictated HDD. They planned for 30-minute nightly offloads.
Interface Transfer Time Reference (500 GB at 70% efficiency)
| Interface | Max Speed | Real-World | Transfer Time | With Verify |
|---|---|---|---|---|
| USB 2.0 | 60 MB/s | 42 MB/s | 3h 18m | 6h 36m |
| SD UHS-I | 104 MB/s | 73 MB/s | 1h 54m | 3h 48m |
| USB 3.0 | 625 MB/s | 437 MB/s | 19m | 38m |
| SD UHS-II | 312 MB/s | 218 MB/s | 38m | 1h 16m |
| SATA III SSD | 600 MB/s | 420 MB/s | 20m | 40m |
| CFexpress B | 2,000 MB/s | 1,400 MB/s | 5m 57s | 11m 54s |
| Thunderbolt 3 | 5,000 MB/s | 3,500 MB/s | 2m 23s | 4m 46s |
| 10 GbE | 1,250 MB/s | 875 MB/s | 9m 31s | 19m 2s |
Pro Tips and Common Mistakes
Pro Tips
- The transfer speed is always limited by the slowest link in the chain. A Thunderbolt 3 connection to an HDD is limited to HDD speed (~150 MB/s), not Thunderbolt speed. A fast card in a slow reader is limited by the reader. Always identify the bottleneck before estimating time.
- Always verify offloads of camera originals with a checksum read-back. Yes, it doubles the transfer time. But a single corrupted file discovered after the cards have been reformatted is an unrecoverable loss. Build verification time into your DIT schedule from the start.
- For network transfers (NAS, SAN), actual throughput depends on network congestion, switch backplane capacity, and the number of concurrent users. A 10 GbE connection shared by 4 editors effectively delivers 2.5 GbE per user. Plan transfers during off-hours when possible.
- When quoting offload times to production, always quote the with-verification time plus a 20% buffer. Unexpected slowdowns from drive fragmentation, thermal throttling, or background processes are common. Under-promising and over-delivering builds trust.
Common Mistakes
- Using theoretical maximum speeds for time estimates. USB 3.0's 5 Gbps (625 MB/s) theoretical maximum is never achieved in practice. Real-world speeds of 350-475 MB/s are typical. Estimating from theoretical speeds will consistently underestimate transfer times by 25-40%.
- Forgetting that verification doubles the offload time. A 20-minute offload becomes 40 minutes with verification. If the DIT tells the AD '20 minutes' and then runs a verify pass, the cards are not available for 40 minutes. Always communicate the full time including verification.
- Ignoring the destination drive speed. Copying from a fast CFexpress card to a slow portable HDD is limited by the HDD's 150 MB/s write speed, regardless of how fast the card reader or USB connection is. The destination bottleneck is the most commonly overlooked factor.
Frequently Asked Questions
Why is my transfer slower than the interface speed?
Multiple factors reduce real-world speed below the theoretical maximum: protocol overhead (USB headers, error correction), controller chip limitations, drive write speed bottlenecks, file system overhead (especially with many small files), thermal throttling on sustained transfers, and shared bus bandwidth with other devices. Expect 60-85% of the theoretical maximum in practice.
Should I always verify my offloads?
For camera originals and any irreplaceable data, yes, always verify. Verification reads back every byte and compares checksums to confirm the copy is identical to the source. For copies of data that exists in multiple other locations (like duplicating an existing backup), verification is optional but recommended. The time cost of verification is always less than the cost of discovering corruption later.
What is the fastest way to offload CFexpress cards?
Use a dedicated CFexpress reader (not the camera's USB port) connected via Thunderbolt 3 or USB 3.2 Gen 2. A ProGrade PG05 or Sony MRW-G2 reader delivers 1,400-1,700 MB/s real-world from CFexpress Type B cards via Thunderbolt. The camera's built-in USB port is typically limited to USB 3.0 speeds (350-475 MB/s real-world), making it 3-4x slower than a dedicated reader.
How do I offload large volumes faster?
Three strategies: (1) Use the fastest available interface, Thunderbolt 4/5 or 10 GbE deliver the highest real-world throughput. (2) Offload to an NVMe SSD instead of an HDD to eliminate the destination bottleneck. (3) Use parallel offloads, running two card readers simultaneously to two different destination drives. Some DIT software (Silverstack, Hedge) supports parallel verified offloads natively.
Start Calculating
Offload time is one of the most critical variables in production logistics. Knowing exactly how long a transfer takes, including verification, lets the DIT give accurate timelines to the AD, ensures cards are available when camera needs them, and prevents the midnight scramble of unexpected delays.
Enter your file size and interface above to see the exact transfer time. What is the bottleneck in your current offload workflow, and have you benchmarked your real-world speeds?
