Methodology

How the numbers are sourced, how the math runs, and what to treat as an estimate vs. a hard fact.

This page exists because before I trust a calculator with a card-purchase decision, I want to know exactly what it's doing. So here is exactly what Camera Data Calculator is doing. If anything looks wrong, email hello@cameradatacalculator.com and I'll fix it.

The math

The whole calculator runs on three formulas. Once you have a published bitrate (Mbps), the rest is unit conversion.

data_rate_MBps = bitrate_Mbps / 8

file_size_GB = data_rate_MBps × runtime_seconds / 1000

cards_needed = ceil(file_size_GB × (1 + buffer%) / card_capacity_GB)

The / 8 converts bits to bytes. The / 1000 converts megabytes to gigabytes using 1 GB = 1,000 MB, which matches how memory card capacity is labeled and sold (a "128 GB" card is 128,000 MB, not 137,438 MB). Treating storage capacity in base-10 means the math lines up with what you'll actually see on the card label.

Where the bitrates come from

Every bitrate in this calculator is sourced from manufacturer documentation. No scraped forum posts, no aggregated review-site numbers, no estimates from comparable codecs.

Nikon

Pulled from each body's official online Reference Manual table (the published Mbps figures for every codec / resolution / frame rate combination). Same source Nikon uses to spec the cameras in their own marketing.

Sony

Pulled from each body's Help Guide PDF. Sony publishes per-body bitrate tables in the spec appendix; numbers are taken directly from those tables, including the asterisk-marker variants for decimal frame rates (23.98p, 29.97p, 59.94p) which I keep as 24p/30p/60p in the interface for readability.

Canon

For the R5 Mark II: Canon's Advanced User Guide PDF. For the R5, R5 C, R3, R6 II, and C70: Canon's official published specifications, matching each model's manual appendix. Cinema RAW Light figures use Canon's HQ/ST/LT rate tables.

Apple ProRes

ProRes 422 HQ, ProRes 422, ProRes 422 LT, ProRes 422 Proxy, and ProRes RAW HQ all use Apple's published data-rate specifications. These are codec-level rates that apply consistently across any camera that records ProRes internally.

Everyone else

Panasonic, Fujifilm, Blackmagic, RED, DJI, GoPro, Insta360, OM System, Sigma, Z CAM, Leica, Apple: official manuals, spec sheets, or published codec data-rate tables from the manufacturer's own documentation. For RED REDCODE, I use the published target Mbps for each compression preset (HQ, MQ, LQ, ELQ) at each resolution and frame rate.

Variable-bitrate codecs and the buffer toggle

This is the most important caveat to understand. Many of the codecs in this calculator (N-RAW, R3D NE, ProRes RAW, XAVC HS, Cinema RAW Light, BRAW) are variable-bitrate. That means the published Mbps figure is an average. Actual recordings fluctuate with scene complexity. High-motion or detailed scenes can run 10–20% above the average; low-motion locked-off shots can run below.

This is why every result has a buffer toggle on the calculator: 0%, 5%, 10%, or 20% padding. The default is 20% because I'd rather you bring one card too many than fill a card right as the take goes long.

For Long-GOP H.264 / H.265 modes that are nominally CBR (constant-bitrate), the published rate is what you'll see in practice. For ALL-Intra modes that are CBR per-frame, the rate is also accurate to within a percent or two.

A worked example

To make this concrete, let's plan a 2-hour shoot in Apple ProRes 422 HQ at 4K UHD 30p with the default 20% buffer. I'm using ProRes here because the data rate is published by Apple, applies to every camera that records it internally, and isn't variable, so the numbers are easy to verify against Apple's own spec sheet.

Step 1. Look up the bitrate. Apple's ProRes data-rate table gives 4K UHD (3840×2160) at 29.97p as 707 Mbps.

Step 2. Convert to data rate. 707 / 8 = 88.4 MB/s.

Step 3. Multiply by runtime (2 hours = 7,200 seconds), then divide by 1,000 to get GB. 88.4 × 7,200 / 1,000 ≈ 636 GB.

Step 4. Apply the 20% buffer. 636 × 1.20 ≈ 764 GB.

Step 5. Cards needed for a 256 GB SD UHS-II V90: ceil(764 / 256) = 3 cards.

So a 2-hour ProRes 422 HQ 4K 30p shoot needs three 256 GB SD UHS-II V90 cards (or larger capacity to consolidate). That's what the calculator displays. For variable-bitrate codecs (N-RAW, R3D NE, ProRes RAW, Cinema RAW Light, BRAW), the steps are identical but step 1 uses an average rate, which is why the buffer in step 4 matters more.

Frame rate handling

Manufacturers spec frame rates as either round numbers (24p, 30p, 60p) or as exact NTSC/film rates (23.98p, 29.97p, 59.94p). The calculator uses the round numbers for the interface because that's how filmmakers actually speak. The underlying bitrate values are pulled from the manufacturer's spec for the matching exact rate (so 30p in the interface is 29.97p in the Nikon table, 60p is 59.94p, and so on). Some bodies publish slightly different rates for 24.00p (film-true) vs. 23.98p (NTSC-film); where both are published, I use the 23.98p figure as the default 24p selection, which is the more common workflow.

What is not included

Audio. Audio bitrates are negligible compared to video (typically <2 Mbps even for multitrack 24-bit). They're folded into the published video bitrate where the manufacturer reports a combined number; otherwise omitted. This will not change your card count.
Metadata, proxies, and overhead. Camera systems add small amounts of overhead for clip headers, thumbnails, and metadata. This is well under 1% on modern cameras and is already absorbed by the buffer toggle.
Filesystem overhead. The "usable" capacity of a card after exFAT formatting is roughly 99% of nominal. The buffer toggle covers this comfortably.
Anamorphic / stretched-pixel modes. These use the same bitrate as the matching standard mode at the same sensor readout.
Photo bursts mid-video. If you fire stills during a take on a hybrid body, that data lands on the same card and is not modeled here.

If a number disagrees with yours

Three possibilities:

You're looking at a different mode. 4K UHD (3840×2160) and 4K DCI (4096×2160) have different bitrates on most cameras. Same for 4K Fine vs. standard 4K, or N-RAW HQ vs. N-RAW NQ.
Your number includes audio and mine doesn't, or vice versa. Some sources publish video-only, others publish video+audio combined.
I have a bug. If you've checked the same mode and the math doesn't match, email me with the camera, codec, resolution, fps, and the source you're comparing against. I'll fix it.

Updates

The bitrate dataset is versioned with the site (footer of every page). Whenever I add a camera or correct a number, that version increments. Browse the dataset history on the project's git repo if you want a paper trail; if there's enough interest I'll publish a changelog page.

← Back to the calculator