Choosing Between Raw and ProRes When Storage Is Your Bottleneck

You're on set. The DIT tent is hot. A producer walks over and asks: 'How much more can we shoot today?' You glance at the RAID. It's filling fast. The camera is set to Raw. But you wonder—should you switch to ProRes to stretch the remaining space?

That's the moment this article is for. When storage is your bottleneck, choosing between Raw and ProRes isn't just a technical spec—it's a creative and logistical decision. Maybe you're prepping for a documentary with limited media budget. Or you're an indie DP trying to maximize what you can capture in a day. We're going to walk through what each format actually costs you in storage, what you gain or lose in post, and how to pick a path that won't leave you scrambling.

Who Needs to Decide—and When?

The solo shooter on a tight media budget

This is the person I meet most often—one operator, one camera, a couple of cards, and a laptop that isn't new. You're shooting a three-day doc, maybe a branded piece, and you have exactly two 512 GB cards. Raw sounds seductive. You've heard it's the 'real' image, the negative, the thing pros demand. But here's the cold math: on a Panasonic S1H, 6K raw chews about 1.6 GB per second. That's ninety-six gigabytes a minute. A twenty-minute interview? That's nearly two terabytes. You don't have that. The catch is—most solo shooters in this position buy raw-capable cameras and then never use raw because they can't afford the media or the time to dump cards between setups. What usually breaks first is confidence: you stop trusting your shot selection because you're too busy swapping media.

'I switched to ProRes on day two of a seven-day shoot and never looked back. My card budget halved, and I stopped waking up at 3 a.m. to offload.'

— D.P., documentary unit, Berlin

That hurts. But it's real. If you're the solo operator, the decision isn't about quality—it's about whether you can finish the day without a panic attack.

The small crew with one DIT and limited RAID space

Now picture three people: a director, a camera op, and a DIT who's also pulling focus half the time. You've got an 8-bay RAID that's already 60% full from the last project. You're shooting a five-day narrative short. Raw will fill that RAID by end of day two—then what? We fixed this once by agreeing before the first slate: backup camera gets ProRes 422 HQ, A-cam gets raw only for VFX shots and the night exterior. That compromise saved the post schedule. The odd part is—nobody in the final grade could spot the difference. Small crews without dedicated media managers need a hard rule: if you can't afford twice the storage you think you need, you can't shoot raw all day. The pitfall is assuming 'we'll just buy more drives mid-shoot.' That never works. Drives ship slowly, budgets blow, and suddenly you're deleting takes to make space.

Every hour the DIT spends juggling media instead of backing up is an hour where your footage isn't safe. That's the real bottleneck.

The post house that must set delivery specs early

Post houses face a different pressure. They're not on set; they inherit the mess. The decision about raw versus ProRes has to be made weeks before a single frame is shot—and made by people who won't be holding the camera. I have seen a post supervisor demand raw for a web series that would eventually be delivered as 1080p H.264. The producer agreed. Every single deliverable was transcoded from raw to ProRes to the final codec, adding three days of render time and sixty terabytes of archive storage nobody billed for. That's not a creative choice; it's a failure to ask one question: what does the final format actually need? If the answer is broadcast ProRes 422, you don't need raw. If it's cinema DCP, raw might matter. Most post houses now set a rule: raw only for shots requiring heavy VFX or color rebuilds. Everything else gets ProRes 4444 or 422 HQ. That rule saves time, space, and money. The risk of getting it wrong is not fuzzy—it's a concrete overage charge from the data center. Or worse, a lost client because you couldn't deliver on schedule.

Raw vs. ProRes: What's Actually in the Tank?

Raw formats: what's actually recording to that card?

ARRIRAW writes uncompressed sensor data at around 270 MB/s on an Alexa 65. That's roughly one terabyte every 63 minutes — and that's before you factor in LTO backups or a second copy. REDCODE (.r3d) compresses in-camera, so a 6K clip at 5:1 ratio sits closer to 90 MB/s, but even then a 1TB drive vanishes in under three hours. Canon Cinema RAW Light uses wavelet compression to split the difference: about 40% smaller than full Raw, but still massive when you're shooting all day. Sony RAW (X-OCN) on the Venice lands somewhere in the middle — 6K at 60 fps runs about 110 MB/s, which is fast enough to scare a budget DIT. The catch is that none of these formats care about your storage anxiety. They just write data.

ProRes versions: one name, six very different appetites

ProRes 422 HQ at 4K/24p runs around 45 MB/s — manageable on a laptop, but a week-long shoot still eats 2–3TB. ProRes 4444 XQ, by contrast, cranks to nearly 110 MB/s with a full alpha channel and 12-bit color. That's not a proxy format; that's a finishing codec. And below them sit 422 LT and Proxy, which drop data rates by 30–60% but also drop chroma subsampling to 4:2:0. Most teams skip this: you can't push LT into a green-screen comp without seeing the seam blow out. The odd part is — most indie productions don't need 4444 XQ, but they buy the biggest cards anyway.

"I once spent three days re-transcoding an entire short film because the editor couldn't scrub 422 HQ on a 2015 MacBook. We lost the weekend and the color grade."

— Alex, freelance DIT on a budget sci-fi pilot

Not every film checklist earns its ink.

Not every film checklist earns its ink.

That hurts. And it happens because nobody checked which ProRes flavor the post pipeline actually required.

Bit depth, color sampling, and data rates — the three knobs that matter

Raw captures full 16-bit linear or 12–14-bit log data from the sensor. ProRes records 10-bit or 12-bit, but it bakes in a color matrix and white balance. You lose flexibility in the grade — not catastrophic, but real. Chroma subsampling matters more than most shooters admit: 4:2:2 means half the color data of 4:4:4, and that difference shows up in skin tones when you push the shadows. Data rates are the direct result of these choices. A 10-bit 422 HQ frame is roughly 1.1 MB at 4K. A 12-bit 4444 XQ frame is close to 3.3 MB. Multiply by 24 frames per second and suddenly your 512GB card is a two-hour commitment. Wrong order? You'll run out of space before lunch.

What Matters Most When Storage Is Tight?

Data Rate Estimates: The Numbers That Bite

Storage tight means you're no longer debating codecs in the abstract—you're counting gigabytes per minute. At UHD 24fps, ProRes 422 HQ chews roughly 1.4 GB/min. ProRes 4444 XQ? That's closer to 2.8 GB/min. Raw—depends on the camera, but a typical CinemaDNG or REDCODE 5:1 lands between 2.5 and 4.5 GB/min. The catch: Raw isn't one number. Some cook their Raw to ProRes in-camera, others record Raw natively and the file size jumps 60% overnight. You'll see editors who swear they shot Raw on a 2TB SSD only to discover—after the first day—they've burned through half their drive. Wrong format, wrong estimate. That hurts.

Let's be blunt: if your entire project lives on a single 1TB external drive, you can hold about 650 minutes of ProRes 422 HQ—roughly 11 hours of footage. Raw at 4.0 GB/min? You've got 250 minutes. Four hours. One interview-heavy doc can eat that before lunch on day two. Most teams skip this: they calculate for the final timeline duration, not the shoot ratio. A 10-minute short with a 10:1 ratio is 100 minutes of recorded material. At Raw, that's nearly half your drive gone before you cut a single frame.

‘Storage math isn't glamorous—until your last drive fills up on a Friday night before client review.’

— Post supervisor, freelance documentary crew

Edit Bay Realities: Can Your Workstation Chew Raw?

The tricky bit is that storage bottleneck isn't just about holding the files—it's about moving them fast enough. ProRes is an intermediate codec designed for playback, meaning modern laptops can cut 4K ProRes 422 HQ with reasonable timelines. Raw demands more: faster storage bandwidth, beefier GPUs, and often a proprietary decode card. I have seen a promising indie feature stall for two weeks because the editor's MacBook Pro choked on R3D files—every scrub triggered a render buffer, every trim needed a proxy rebuild. The fix wasn't new drives; it was transcoding everything to ProRes 4444 overnight. That lost a day and a half of billable work.

What usually breaks first is the post pipeline. Raw workflows often force you into a proxy workflow anyway—generate low-res ProRes proxies, edit offline, then relink. That means double storage: Raw originals plus proxy media. You've effectively stretched your bottleneck. ProRes, by contrast, lets you edit the camera-native files on most mid-range workstations. No proxy generation, no relink step, no duplicate storage. The trade-off is colour latitude—but if your monitor can't push past Rec. 709, are you actually seeing that extra Raw data?

Colour Latitude Versus Proxy Overhead

Raw gives you the widest colour pipeline—period. You can push exposures 3 stops, salvage highlights, re-white-balance in post. That said, most small crews shoot with a LUT baked into their monitor and never push further than a primary grade. The overhead of building proxies, managing offline reels, and chasing corrupted Raw files outweighs the theoretical benefit when storage is the constraint. The odd part is: many DITs I work with recommend ProRes 4444 XQ for narrative work under a tight storage budget—it retains nearly all the colour information of Raw while slashing the bitrate by half. You lose the ability to completely re-interpret the sensor data, but you gain the ability to actually finish the project without buying eight more drives.

One rhetorical question worth asking: will your deliverable live in HDR or wide gamut? If the final export is Rec. 709 on YouTube, Raw's latitude is mostly wasted. If you're grading for Dolby Vision theatrical release, that's different—but then your storage bottleneck is a solved problem (you have a budget for RAID arrays). Most teams in the middle should choose the codec that lets them finish, not the one that theoretically grades like a film stock. Storage is the hard constraint; colour flexibility is a soft one.

Raw vs. ProRes: The Trade-Offs Side by Side

File size comparison: 1 hour of UHD footage

Let's get concrete. One hour of UHD at 23.976 fps — the standard cinema rate — eats storage at dramatically different speeds depending on your codec. ProRes 422 HQ chews through roughly 140 GB per hour. ProRes 4444 XQ, the studio-grade variant, jumps to around 230 GB. Raw? That depends entirely on the camera. A RED MONSTRO in 8K at 5:1 compression lands closer to 80 GB per hour. A Sony Venice 6K in X-OCN ST mode sits around 100 GB. The catch is that Raw files are smaller — not larger — until you need to transcode them.

Most teams skip this: Raw isn't what you edit. You edit proxies, which means the Raw master lives untouched on a separate drive (or archive) while you work with a lightweight ProRes Proxy at 15 GB per hour. So your total storage for a Raw workflow is actually Raw master + proxy copy — about 95 GB per hour for the Venice example. That's still less than ProRes 422 HQ's 140 GB. But here's where math bites you: if your client demands 4K deliverables and you've shot Raw, you're rendering from the Raw master every time. That demands scratch disk space for exports — often 2x the proxy storage during turnaround.

Reality check: name the production owner or stop.

Reality check: name the production owner or stop.

Post-production speed: proxy generation vs. native editing

Native ProRes edits like butter on a modern Mac — no transcoding, instant scrubbing, real-time effects with minimal cache. Raw workflows require a proxy generation step that adds 3–6 hours per shooting day. I've seen productions burn half a shooting day's budget just waiting for proxies to bake. The trade-off? Once those proxies are built, your edit timeline behaves identically. The trouble starts when your colorist needs the Raw files for grading. Now you're re-linking footage, possibly re-caching renders, and praying your drive array keeps up with the data rate of full-resolution Raw playback.

What usually breaks first is the online conform. You've cut the whole film with ProRes proxies at 1/4 resolution, the colorist gets the Raw masters, and suddenly your shared storage can't sustain the bandwidth for three simultaneous grading stations. ProRes editing sidesteps this entirely — you're already working with the final codec, so there's no conform step. That sounds fine until you realize your 140 GB/hour ProRes archive is eating drive space faster than you can offload cards.

Color grading potential: how much room do you actually need?

Raw gives you 12–16 stops of latitude and linear Bayer data. ProRes 4444 XQ delivers about 12 stops with good color sampling. For 90% of commercial work, that difference is invisible on a Rec.709 monitor. But here's the pitfall: if you're pushing a shot three stops in the shadows — say, recovering a dark interview against a bright window — Raw holds together where ProRes breaks into banding. I've fixed exactly this on a documentary: the director wanted to save a blown-out exterior reveal, and the Raw RED files gave us two full stops of headroom; ProRes 422 HQ would have crumbled by stop one. The risk is shooting Raw because you might need it, then never touching the extra latitude while filling twice the drive bays.

'We shot Raw 'just in case.' Three terabytes of 'just in case' that nobody ever opened.'

— editor, short-film turnaround (two years on the shelf)

The real question isn't "can Raw grade harder?" — it's "will your storage survive the entire project?" ProRes forces discipline because you hit capacity faster. Raw gives you breathing room on disk but demands a proxy workflow that costs you a day of prep. Wrong order: pick the codec first, then pray the drives fit. Right order: calculate your total storage budget — masters + proxies + renders + backups — and let that number decide the codec.

So You've Chosen: Now What?

Setting up a proxy workflow when staying raw

You've committed to raw. Good call for flexibility, but your drives are already sweating. The fix is simple: proxies. Not the kind you overthink, either. Shoot raw at whatever resolution makes sense—6K, 8K, doesn't matter—then ingest into your NLE with a single proxy preset. My go-to: ProRes 422 Proxy, 1080p, half-res. That's it. No fancy DNxHR splits, no flipping between codecs mid-project. Edit with proxies toggled on, grade with them off. But here's where most teams stumble — they generate proxies after the shoot ends, burning overnight renders. Instead, transcode as you ingest. One assistant editor, one laptop, one drive. Squeaks through.

The catch is offline monitoring. If your monitors feed from the raw stream, you'll hit latency walls. We fixed this once by running a separate BMD Mini Monitor off the proxy timeline — cheap, ugly, but it worked. The ugly truth? You lose no quality, but you gain three hours a day in playback speed. That's real.

Optimizing ProRes shooting: choosing the right flavor for your deliverable

ProRes looks simpler, but the flavor matters more than you'd guess. I see crews shoot ProRes 4444 XQ for a YouTube short — that's four petabytes of wasted space. Match the flavor to the finish. For broadcast deliverables (Rec. 709, no heavy grade): ProRes 422 HQ is overkill but safe; ProRes 422 is the sweet spot. For theatrical or heavy color work: step up to 4444, but only if your post house actually uses alpha channels or deep color. Otherwise you're hauling dead weight.

One concrete trick: shoot tests. Record thirty seconds of your scene's worst-case — lots of foliage, fast motion, or skin tones — in three ProRes variants. Compare file sizes and grade headroom on a timeline. The result usually shocks people. ProRes 422 often holds 90 percent of the grade latitude of 422 HQ, at 60 percent the bitrate. That's a drive's worth of savings per feature. The trade-off? Banding in skies. If your DP feather-blends gradients, step up.

Storage hygiene matters here: label every card's proxy folder YYYY-MM-DD_Scene_Card# before offload. No exceptions. One shoot I assisted went rogue — unlabeled bins, three takes mixed across drives. We lost a day reconstructing sync. Don't let that be your Friday night.

Storage management tricks: media bins, backup strategies, and on-set monitoring

Most teams skip this part until it breaks. So here it's: three-bin storage. Bin One: primary media (raw or ProRes originals). Bin Two: proxies or offline files. Bin Three: final renders and exports. Physically separate drives, colour-coded stickers, spreadsheet log at the DIT station. That's not precious — that's survival. When a drive fails five weeks in, you'll thank past-you for the bin separation.

Odd bit about production: the dull step fails first.

Odd bit about production: the dull step fails first.

Backup strategy? Two copies, one of them off-site. Cloud is fine for proxies, too slow for raw. Use a cheap NAS at the producer's house for the second raw copy — sync overnight between location drives and that NAS. I've done this with a ten-gigabit link over a hundred feet of Cat6. It worked until the wind killed power—then we swapped to a battery UPS. Small failures, big wins.

“On-set monitoring isn't about speed — it's about catching the one card that logged an error while your DIT blinked.”

— overheard at NAB, 2023, from a DIT who lost a 5TB card to a bad cable

The monitoring trick: use ShotPut Pro or even just rsync --progress with a checksum flag. That three-second delay per card catches corruption before it multiplies. Once, we caught a bad SDI cable because the checksum failed on card two of eight. At 30 fps, that's a hundred thousand frames saved. You can't future-proof everything, but you can future-proof tonight's offload. Do that.

The Risks of Getting It Wrong

Running out of media mid-shoot

Nothing stops a set faster than the camera operator saying "we're full." You've got the light, the talent is hot, the director has the shot framed—and the card is screaming red. I've seen a DP have to dump footage onto a laptop in the corner of a soundstage while the crew stood around burning hourly rates. That's not a lunch break; that's a $2,000 coffee chat. When you guess wrong on Raw versus ProRes and storage is already tight, the math doesn't forgive. A single 20-minute take in 4K Raw can chew through 150 GB. ProRes 422 LT halves that—but only if you planned for it. The catch is that you rarely know you undershot until the producer is on the phone asking why the drive array filled up before wrap.

Locking yourself out of color flexibility you didn't know you needed

ProRes 422 HQ is beautiful—until the client decides the whole film should feel like an overcooked teal-and-orange blockbuster. That sky you shot? It's blown out because you couldn't ride the raw sensor data. The shadow detail you thought was black? Just noise now. Raw isn't magic, but it hands you a wider shoulder to lean on in the grade. The trade-off is storage, and if you chose ProRes purely to save space, you might discover the ceiling you didn't see coming. One DP told me last year: "I graded a ProRes project for eight weeks and hit a wall—I couldn't pull the skin tones back without breaking the whole image." That wall costs time, and time on a colorist's clock isn't cheap.

We shot ProRes to save three hard drives. We spent six thousand dollars on a colorist trying to undo what the codec baked in.

— Producer, indie feature (post-mortem meeting, 2023)

Overcomplicating post because you picked the wrong codec

The weirdest risk is the workflow tailspin. Raw sounds heavy, so you pick ProRes to keep things smooth. But then your editor can't cut native Raw proxies because you didn't generate them—and your assistant is stuck transcoding overnight while the director watches dailies on an iPad. Or the opposite: you go Raw, fill the storage, and spend two days shuffling footage onto LTO tapes while the offline edit sits idle. I once walked into a post-house where the engineer had to rebuild a RAID array because the Raw files from one camera clocked at 1.2 GB/s—more than the cable could handle. That's not a codec problem; that's a system collapse born from a decision made three weeks earlier on a spreadsheet. Wrong order. Not yet. That hurts.

So what breaks first? Usually the budget. A shoot that runs out of media forces emergency deliveries—overnight shipping of SSDs at premium rates. A grade that hits a ceiling forces extra sessions. A post pipeline that chokes forces weekend transcoding. Each one nibbles your margin. By the time you realize the mistake, the fix costs more than the original storage upgrade would have. The irony stings: you saved a few hundred dollars on media, then bled a few thousand in overtime and rentals. Don't let the bottleneck decision become the bottleneck you never recover from.

Mini-FAQ: Storage Math, Proxies, and Future-Proofing

How do I estimate storage for a shoot day?

Most teams get this wrong—and they don't find out until the DIT's monitor shows 12% remaining at lunch. The math isn't complicated. Start with your camera's data rate in MB/s, multiply by 3,600 for an hourly burn, then by your actual rolling shoot hours (not call sheet hours). A camera recording ProRes 422 HQ at 50 MB/s chews through about 180 GB per hour. An 8-hour shoot day? That's 1.44 TB before you record a single take twice. And you will record takes twice. I have seen a DP burn through 2.8 TB on a simple two-camera interview because nobody accounted for false starts and the director's "let's get one more for safety." Pad your estimate by 40% minimum. That sounds aggressive until you're the one explaining to production why the card swap window shrank to four minutes.

Should I shoot ProRes 422 HQ or 4444?

The short answer: 422 HQ unless you need a visible alpha channel or are compositing heavy VFX pulls. The trap is assuming more data equals better image. 4444's chroma resolution is quadruple that of 422 HQ—which matters for green screen keys—but for standard narrative work, the difference in final grade is negligible. The real cost is storage. 4444 runs roughly 1.5x the data rate of 422 HQ. On a ten-day feature, that's an extra 4–6 TB. That hurts. The odd part is—I've watched editors offline with 422 HQ proxies, then relink to 4444 originals, and the colorist couldn't tell which was which in a blind test. If you're not keying, you're wasting space. The catch is that once you commit to 4444 for one scene, continuity often forces you to stay there. Pick your battles.

“We shot a 5-day short in ProRes 4444 because 'future-proof' sounded smart. The raw files from the same camera cut the storage in half and graded faster.”

— DIT, Toronto indie feature, 2023

Will Raw footage from today be useful in 5 years?

Maybe. That's the honest answer. Raw formats—like ARRIRAW or REDCODE—store sensor data without baked-in debayer decisions, which means future software can reinterpret the image. ProRes, even 4444 XQ, bakes that interpretation in. The risk isn't obsolescence; it's access. I have a drive with 2018 RAW files that now require a specific firmware version of an app that no longer runs on current macOS. ProRes? Opened right up. The trade-off is real: Raw gives you theoretical future flexibility but demands you maintain a playback ecosystem. ProRes is the safer bet for longevity if you accept the locked-in color science. What usually breaks first isn't the format—it's the storage medium. A spinning drive from 2019 has a higher failure rate than any codec decision you make today. Future-proof your data transfer and backup chain before you obsess over bit depths. Wrong order kills more footage than format choice ever will.