How much storage does a film shoot need? There is no reliable answer based on resolution or shooting days alone.

A two-camera interview recorded in 4K can produce less data than a single-camera RAW shoot. A 12-hour call may contain only two recorded camera-hours, while a three-hour concert with four cameras can create roughly 12 camera-hours of footage. And a storage system advertised as 48 TB may provide less usable project capacity once redundancy and system overhead are considered.

The useful question is therefore not simply "How many terabytes does 4K need?" but "How much data will this production actually record, and how much usable capacity does each required storage destination need?" This guide provides a practical way to estimate that number without treating the estimate as a guarantee of final file size.

Start with the camera's actual recording data rate, then estimate the total number of recorded camera-hours across all cameras. That gives you the approximate camera-original dataset. Next, size each required complete storage destination for that full dataset, add reasonable headroom, and account for any capacity lost to RAID, formatting, or system overhead.

A useful planning model is camera data rate x total camera-hours = estimated camera-original dataset. From there, add appropriate headroom to arrive at the recommended usable capacity per complete destination.

These calculations are planning estimates, not guaranteed final file sizes. Actual storage use can vary with codec behavior, variable bitrate, Constant Quality RAW modes, audio, metadata, camera settings, and production workflow.

For the examples below, storage uses decimal units: 1 TB = 1,000 GB.

The One Storage Formula Worth Knowing

If a camera publishes its recording bitrate in Mbps (megabits per second), use:

GB per recorded hour = Mbps x 0.45

The conversion is:

Mbps x 3,600 seconds / 8 / 1,000 = GB per hour

For example, Sony lists XAVC S-I DCI 4K at 23.98p/24.00p on the FX3A at 240 Mbps.[1]

240 Mbps x 0.45 = 108 GB per recorded hour

If the camera publishes MB/s (megabytes per second), use:

GB per recorded hour = MB/s x 3.6

Blackmagic Design, for example, lists 121 MB/s for 6144 x 3456 Blackmagic RAW 8:1 at 30 fps on the Pocket Cinema Camera 6K Pro/G2 family.[2]

121 MB/s x 3.6 = 435.6 GB per recorded hour

Mbps vs. MB/s: an easy 8x mistake

The units look similar but are not interchangeable. Mbps means megabits per second, while MB/s means megabytes per second, and eight bits equal one byte. That is why 240 Mbps works out to about 108 GB/hour, while 240 MB/s would be about 864 GB/hour.

Before using any calculator or spreadsheet, confirm which unit the camera manufacturer is using.

4K is a resolution, not a data rate.

Two cameras can both record 4K while producing dramatically different amounts of data because the recording format also depends on the codec, compression ratio or quality setting, frame rate, bit depth, chroma format, whether the camera is recording RAW or processed video, whether the bitrate is fixed or content-dependent, and how long the camera actually records.

Even within one camera, changing a recording setting can more than double the hourly storage requirement. Sony lists FX3A XAVC S-I DCI 4K at 23.98p/24.00p as 240 Mbps, or about 108 GB/hour, while the 59.94p setting is 600 Mbps, or about 270 GB/hour.[1]

That is why a search such as "How much storage does 4K use?" cannot have one technically correct answer. The better question is: Which camera, codec, frame rate, and recording mode are you using?

The safest estimate starts with the manufacturer's published rate for the exact acquisition setting.

Sony publishes recording bitrates for the FX3A.[1] Blackmagic publishes BRAW storage rates by resolution and compression setting; importantly, it also notes that its Constant Quality modes can vary according to image content.[2] RED publishes R3D data rates by quality level and recording mode; for V-RAPTOR [X] 8K 17:9 at 24p, R3D MQ is listed at 298 MB/s.[3]

ARRI takes the more useful approach for a complex cinema system: its official Formats & Data Rate Calculator calculates recording duration and data rate from the selected camera, codec, sensor mode, resolution, project rate, sensor speed, media, and audio configuration.[4]

The lesson is simple: use the exact acquisition setting, not a generic 4K, 6K, or 8K assumption.

A 12-hour production day does not mean 12 hours of footage. Narrative productions may spend much of the day on lighting, blocking, rehearsals, resets, lens changes, and company moves. Cameras may roll for only a fraction of the call.

Interviews, documentaries, concerts, and events can behave very differently. A two-hour interview may create close to two hours of footage on each camera. Four cameras recording a three-hour performance produce roughly 12 camera-hours before rehearsals, soundcheck, or pickups are included.

For storage planning, think in total camera-hours. One camera recording for three hours creates three camera-hours; two cameras each recording for three hours create six camera-hours; four cameras each recording for three hours create twelve. If cameras record for different durations, add their individual recorded hours rather than relying on one average.

This is why "How much storage for a 10-day shoot?" is incomplete. Ten shooting days can produce 10 camera-hours or 100 camera-hours depending on the production.

During early pre-production, a crew may not yet know how many camera-hours will be recorded. In that case, shooting ratio can provide a rough planning estimate.

A 10:1 shooting ratio means approximately ten minutes are acquired for every minute expected in the finished project.

For a 90-minute film:

90 minutes x 10 = 900 recorded minutes = 15 recorded hours

If that project were recorded at 240 Mbps:

15 hours x 108 GB/hour = about 1.62 TB of camera originals

But shooting ratio should not be used as a second multiplier once realistic recorded camera-hours are already available. Both methods are trying to estimate the same thing: total acquisition time.

For multicamera work, define whether the shooting-ratio estimate already includes all cameras. Otherwise the same footage can easily be counted twice.

In practice, use camera-hours once the recording plan is reasonably known. Use shooting ratio only when you have an expected final runtime but still need an early estimate of how much material the production may acquire.

Suppose a production expects 5 TB of camera originals. That does not automatically mean 5 TB of storage is enough. The camera-original dataset tells you how much media the production will create. The storage policy answers a different question: how many complete destinations must hold that media? If the workflow requires two independent complete destinations, then each destination needs enough usable capacity for the entire 5 TB dataset, plus whatever headroom the production chooses.

The storage plan may also need room for production audio, stills and reference photography, proxies or dailies, offload reports and metadata, project files, temporary working media, and schedule changes such as overtime, pickups, or an additional shooting day. A practical planning buffer can help, but there is no universal industry percentage. The worked examples below use 20% headroom only as an illustration, not as a required rule.

These three ideas are often mixed together, but they solve different problems. Capacity asks whether there is enough usable space for the dataset. Redundancy asks whether a component can fail without immediately making the storage unavailable. Verification asks whether the copied files arrived intact and match the source.

A production can have enough capacity and still have an unsafe data workflow. A RAID can tolerate certain drive failures without proving that a source copy was correct. A checksum-verified copy can be valid while still existing on only one physical storage system.

Professional tools such as ShotPut Pro and Silverstack Lab are built around verified offloads, checksums, reports, and production media management.[6][7] ASC Media Hash List provides a standardized way to carry file-integrity information through production and post.[8] For the copy, verification, and card-release side of the workflow, see UnifyDrive's separate guide to a checksum-verified on-set backup workflow.[9]

The goal of these examples is not to create universal presets. It is to show how the same method behaves at very different acquisition rates.

Example 1: Two-Camera Sony FX3A Interview Shoot

Assume a two-camera Sony FX3A interview shoot using XAVC S-I DCI 4K at 23.98p and 240 Mbps. Each camera records for two hours per day over three shooting days, the workflow requires two independent complete destinations, and this example uses 20% planning headroom.

At 240 Mbps, the camera generates about 108 GB/hour. Across two cameras, two recorded hours per camera per day, and three days, that becomes 12 camera-hours. The estimated camera-original dataset is therefore 12 x 108 GB = 1.296 TB. With 20% example headroom, each complete destination should provide about 1.56 TB of usable capacity.

If the production requires two independent complete destinations, each should be sized for approximately 1.56 TB or more. The important number is not "3.12 TB on one device." It is the full dataset capacity required on each destination.

Example 2: High-Data-Rate RED RAW Production

RED lists 298 MB/s for V-RAPTOR [X] 8K 17:9, 24p, R3D MQ.[3] That works out to 1,072.8 GB/hour. If one camera records for two hours per day over five days, the production creates 10 camera-hours. The estimated camera-original dataset is therefore 10 x 1,072.8 GB = 10.728 TB. With 20% example headroom, each complete destination should provide about 12.87 TB of usable capacity.

A second independent complete destination would need comparable capacity again.

The contrast between these examples is the reason a generic "4K vs. 8K" rule is not enough. Acquisition mode and recorded time matter far more than the resolution label alone.

Storage hardware is often described by raw installed capacity, while production planning should be based on the capacity that remains after the selected storage layout and system overhead are taken into account.

UP6 supports six PCIe 4.0 M.2 slots and up to 48 TB of installed all-flash capacity.[5] UnifyDrive currently documents support for RAID 0, RAID 1, RAID 5, RAID 6, and UDR on UP6.[10][11]

The important difference between these modes is not simply speed. Each one makes a different trade-off between usable capacity and drive-failure protection:

The 40 TB and 32 TB figures above are nominal RAID math, not guarantees of the exact project space a formatted UP6 will show. Filesystem and system overhead, reserved space, SSD sizing, firmware implementation, and the way volumes are configured can reduce final available capacity. RAID 1 and UDR are especially poor candidates for a single universal "six drives equals X TB" figure because their usable capacity depends on how the mirror or protected folders are configured.[11]

RAID also solves only one part of the risk model. RAID 1, 5, 6, and UDR can improve availability when supported drive failures occur, but a redundant array inside one UP6 remains one storage system, not an independent backup destination.[9][11]

The practical rule is therefore simple: compare the production dataset with usable capacity after the chosen storage configuration, not only with the headline raw capacity.

Capacity tells you how much space is required. It does not decide the best storage architecture by itself.

Portable SSD

Portable SSDs are simple, fast, and easy to hand off. They work well for smaller datasets, shuttle media, individual destinations, and straightforward single-user workflows. The friction tends to appear as drive count grows: more labels, more cables, more adapters, more individual devices to track, and more manual decisions about which copy lives where.

RAID

RAID combines multiple drives into one storage set and can trade capacity for performance and/or drive-failure tolerance depending on the RAID level. RAID should not be treated as a substitute for an independent copy. It primarily changes how one storage system behaves when individual drives fail.

NAS

A NAS becomes more relevant when production needs centralized capacity plus some combination of direct access, network access, shared storage, organization, or fewer scattered devices. A portable NAS moves that model closer to acquisition instead of limiting it to an edit suite or studio.

For a deeper comparison of these roles, see Portable SSD vs Portable NAS for On-Set Backup.[12]

There is no exact terabyte threshold; the shift usually becomes visible when several portable SSDs are being rotated every day, A/B or shuttle destinations are harder to track, the crew needs both direct-attached and network access, footage and working files are scattered across unrelated devices, or card ingest, storage, networking, and power protection are being handled by several separate boxes. Once daily production volume is regularly measured in terabytes or multiple users need organized access to the same working storage, the question changes from "Which SSD should we buy?" to "How should production storage be organized as a system?"

That is where a portable all-flash NAS can become useful.

A production can have enough terabytes and still have slow card turnaround.

Offload performance depends on the complete path:

camera media -> reader -> interface -> host/offload software -> destination -> verification

A fast storage array cannot make a slower camera card read faster. Multiple simultaneous destinations can also increase I/O load, while checksum verification and reporting are deliberate parts of professional workflows.

If turnaround time is the problem rather than capacity, identify the active bottleneck instead of assuming that buying a larger storage system will solve it. For that problem, see How to Speed Up Your DIT Offload Workflow: Find and Fix the Real Bottleneck.[13]

As datasets grow, storage planning can shift from choosing individual drives to managing capacity, ingest, connectivity, local control, power continuity, collaboration, and recovery options as one system.

UnifyDrive UP6 is designed for that mobile storage layer. Its current hardware specification includes six PCIe 4.0 M.2 slots with up to 48 TB of installed all-flash capacity, dual Thunderbolt 4 ports rated up to 40 Gbps, 10GbE, a UHS-II SD slot, a CFexpress Type B slot with Type A supported through an adapter, a 6-inch touchscreen, Wi-Fi 6 with AP mode, Plug & Backup for PC-free initial copies from supported media, and a built-in UPS rated by UnifyDrive for up to two hours under its internal test conditions. The unit is listed at approximately 1.3 kg without the silicone case.[5]

The storage layer is configurable rather than fixed. UP6 supports RAID 0, RAID 1, RAID 5, RAID 6, and UDR, so a crew can choose a different balance of capacity and drive-failure tolerance depending on whether the device is being used for working storage, a protected project pool, or another defined role in the media plan.[10][11] S.M.A.R.T. drive-health information can also surface warning conditions before a failing SSD becomes an emergency.[9][14]

The more interesting part for a production team is that protection does not stop at the RAID level. Group Files and permission controls can be used to organize shared project folders and manage who has access.[12] When a collaborator deletes a file from Group Files, UnifyDrive documents an administrator-controlled Recycle Bin that can capture the deletion instead of making it immediately permanent; its current workflow documentation also describes this protection for files deleted over SMB.[14] That does not replace a backup, but it can reduce the chance that a routine collaboration mistake becomes an irreversible loss.

UP6 also includes Recovery Assistant for a different failure mode: files accidentally deleted from supported memory cards or external media. UnifyDrive states that the tool can scan for deleted files that have not yet been overwritten and restore recoverable items to Internal Storage or Group Files.[12][14] Recovery success is not guaranteed. The correct response after an accidental deletion is to stop using the affected media immediately so new writes do not overwrite the deleted data.[14]

These functions address different risks. RAID and UDR are about storage availability when drives fail; S.M.A.R.T. monitoring is about early warning; the Group Files Recycle Bin addresses accidental deletion in shared storage; Recovery Assistant is a last-resort path for supported source media; and the built-in UPS adds a power buffer around active storage operations. None of them makes verification or independent copies unnecessary.[9][14]

Taken together, those capabilities let UP6 consolidate several hardware and management layers of a compact production data setup: high-capacity flash storage, direct workstation connectivity, card ingest, networking, local control, team access, and several recovery-oriented safeguards.

That does not make UP6 a replacement for professional DIT software, checksum verification, QC, or independent backup destinations. ShotPut Pro supports full-checksum verification, ASC MHL output, and offload reports, while Silverstack Lab provides checksum-based verification, ASC MHL workflows, and production data-management tools.[6][7] ASC MHL itself is designed to track file integrity and chain of custody across production and post.[8]

UP6 is therefore better understood as one storage destination or working-storage layer within a wider production workflow, with additional controls around collaboration and recovery.

Plug & Backup can provide a PC-free initial landing copy from supported media. But "copy complete" should not automatically mean "card cleared." Card release should still follow the production's verification, copy, and QC policy.[9]

Sometimes, but the answer depends on the production math. A moderate-data-rate project with limited recorded hours may fit comfortably in far less capacity. A multicamera RAW project can consume tens of terabytes quickly. And 48 TB of raw installed capacity does not necessarily mean 48 TB of usable project space after the selected storage configuration and system overhead.[5][10]

If the production requires multiple independent complete copies, calculate the usable capacity required per destination first. Do not combine the capacity of all required copies and compare that total with one storage system.

Before choosing drives or building the DIT/data-management package, confirm:

1. What exact camera, codec, resolution, frame rate, and compression/quality setting will be used?

2. What recording data rate does the manufacturer publish for that setting?

3. How many camera-hours are realistically expected?

4. If camera-hours are unknown, what shooting ratio is reasonable for an early estimate?

5. Does that estimate already include all cameras?

6. How many complete storage destinations does the production require?

7. What usable capacity remains after the selected RAID/storage layout?

8. Does the same storage also need to hold audio, proxies, stills, reports, or project files?

9. How much operational headroom is appropriate for the production?

10. What software and checksum method will verify the copies?

11. What conditions must be met before a camera card can be released and reformatted?

Once those questions are answered, storage becomes a production calculation rather than a guess.

How many TB do I need for a film shoot?

There is no universal number. Estimate the camera-original dataset from the recording data rate and total camera-hours, then size every required complete destination for the full dataset plus appropriate headroom. High-data-rate RAW and multicamera productions can require many times more capacity than compressed single-camera shoots.

How much storage does one day of filming use?

It depends on how many camera-hours are actually recorded, not how long the crew works. At 240 Mbps, one recorded camera-hour is about 108 GB. Four camera-hours would be about 432 GB before additional copies and headroom. A high-data-rate RAW format can produce several times more data over the same recording time.

How much storage does 4K video use per hour?

4K has no single file size. For example, Sony FX3A XAVC S-I DCI 4K at 23.98p/24.00p is 240 Mbps, or about 108 GB/hour, while the 59.94p setting is 600 Mbps, or about 270 GB/hour.[1] The codec and recording mode matter as much as the resolution.

How much storage do I need for a 10-day shoot?

Use the exact acquisition data rate and estimate total recorded camera-hours across the ten days. A ten-day production with one recorded hour per day is very different from a ten-day multicamera documentary recording six camera-hours per day. "Ten days" alone is not enough information to estimate storage accurately.

How much storage do I need for a feature film?

If recorded camera-hours are not known yet, shooting ratio can provide an early estimate. Multiply the expected final runtime by the planned shooting ratio, then apply the recording data rate. Once the production has a realistic camera-hour estimate, use that instead of shooting ratio.

How much storage does Blackmagic RAW use?

It depends on resolution, frame rate, and BRAW compression or quality mode. Blackmagic lists 6144 x 3456 BRAW 8:1 at 30 fps at 121 MB/s, approximately 435.6 GB/hour.[2] Constant Quality modes can vary according to image content, so a fixed hourly number is not appropriate for every BRAW setting.

Is RAID the same as a backup?

No. RAID can provide performance and/or drive-failure tolerance within one storage system. It does not create an independent physical copy, and it does not verify that copied media matches the source.

Why is 48 TB not always 48 TB usable?

48 TB can describe raw installed drive capacity. RAID parity or mirroring, formatting, filesystem/system overhead, and reserved space reduce the amount available for project data. Production planning should therefore compare the dataset with the usable capacity of the selected storage layout, not only the raw number printed in the specifications.

Is 48 TB enough for a feature film?

It may be, but there is no reliable answer without the codec, data rate, total camera-hours, storage configuration, and number of required independent destinations. A compressed single-camera feature and a multicamera RAW feature can differ by tens of terabytes.

The most reliable storage plan starts before the first card reaches a DIT or data wrangler.

Start with the acquisition format. Estimate total camera-hours. Calculate the approximate original dataset. Decide how many complete destinations the production requires. Compare that requirement with usable rather than raw capacity. Then choose the storage architecture that supports the workflow without compromising verification, card-release discipline, or handoff.

For smaller projects, that may be a few portable SSDs. As datasets and drive counts grow, centralized flash storage, direct high-bandwidth connectivity, and local network access can become more useful.

UP6 is designed to consolidate several of those storage and connectivity functions into a mobile system while still fitting alongside professional offload software, checksum verification, QC, and independent backup destinations.

Explore UnifyDrive UP6: https://unifydrive.com/products/unifydrive-up6

Camera and Codec Data

[1] Sony. "ILME-FX3A Specifications." Sony USA.

https://www.sony.com/electronics/support/camcorders-and-video-cameras-interchangeable-lens-camcorders/ilme-fx3a/specifications

[2] Blackmagic Design. "Blackmagic Pocket Cinema Camera - Tech Specs." Blackmagic RAW recording formats and storage rates.

https://www.blackmagicdesign.com/products/blackmagicpocketcinemacamera/techspecs

[3] RED Digital Cinema. "V-RAPTOR [X] 8K VV Operation Guide - R3D Quality."

https://docs.red.com/955-0225/955-0225_V2.0%20Rev-B%20RED%20PS,%20V-RAPTOR%20%5BX%5D%208K%20VV%20Operation%20Guide%20HTML/Content/4_Menus/b_ProjSet/R3D_Quality.htm

[4] ARRI. "Formats & Data Rate Calculator."

https://www.arri.com/en/learn-help/learn-help-camera-system/tools/formats-and-data-rate-calculator

Product and Production Workflow

[5] UnifyDrive. "UnifyDrive UP6." Product specifications and workflow features.

https://unifydrive.com/products/unifydrive-up6

[6] Imagine Products. "ShotPut Pro." Verified offload, checksum, ASC MHL, and reporting functionality.

https://www.imagineproducts.com/product/shotput-pro/mac

[7] Pomfort. "Silverstack Lab." Data management, verified backup, QC, and reporting.

https://pomfort.com/silverstacklab/

[8] ASC Media Hash List. "Media Hash List."

https://mediahashlist.org/

[9] UnifyDrive. "On-Set Data Management for Small Crews: A Checksum-Verified DIT Backup Workflow."

https://unifydrive.com/blogs/features/on-set-data-management-for-small-crews-a-checksum-verified-dit-backup-workflow

[10] UnifyDrive. "Secure Creative Data Anywhere with UnifyDrive UP6 Mobile NAS." RAID 0, 1, 5, 6 and UDR support.

https://unifydrive.com/blogs/features/why-the-unifydrive-up6-is-the-new-survival-gear-for-modern-creators

[11] UnifyDrive. "RAID Decoded: The Essential Breakdown of Minimum Disks, Speed, and Fault Tolerance."

https://unifydrive.com/blogs/features/raid-0-1-5-6-10-explained-choosing-the-right-raid-level-for-your-unifydrive-nas

[12] UnifyDrive. "Portable SSD vs Portable NAS for On-Set Backup: Which Fits a DIT Workflow Better?"

https://unifydrive.com/blogs/features/portable-ssd-vs-portable-nas-video-creators

[13] UnifyDrive. "How to Speed Up Your DIT Offload Workflow: Find and Fix the Real Bottleneck."

https://unifydrive.com/blogs/features/how-to-speed-up-your-dit-offload-workflow-find-and-fix-the-real-bottleneck