Recording format is the one decision you cannot fix later. You can always compress a lossless master down, but you can never get back detail a lossy encoder threw away at record time. So it is worth understanding what each option actually costs and buys you.
We build KeepTake, which records in all four of the formats below, so these numbers come from what the app actually produces at 1080p30. Your exact sizes will vary with resolution, frame rate, and how much motion and noise is in the frame, but the ratios hold.
The four options at a glance
| Format | Container | Quality | Size at 1080p30 |
|---|---|---|---|
| H.264 | MP4 | Lossy, good | Smallest, varies with bitrate |
| H.265 (HEVC) | MP4 | Lossy, better per byte | ~1.3 GB per 30 min |
| FFV1 | MKV | Mathematically lossless | ~27 GB per 30 min |
| Uncompressed rawvideo | MOV | Bit-for-bit off the camera | ~8 GB per minute |
Uncompressed: the honest baseline
Uncompressed rawvideo stores every pixel of every frame with no encoding at all. At 1080p30 that is roughly 8 GB per minute. A 30-minute take is around a quarter of a terabyte.
Why would anyone do this? Because it is the only format with zero generation loss and zero encoder involvement. Nothing is estimated, nothing is smoothed, and the file costs nearly no CPU to write. If you are capturing a master for heavy color work or archival of something unrepeatable, and you have the disk for it, uncompressed is the pure option.
The practical constraints are disk space and disk speed. You need a drive that can sustain the write rate, which any modern internal SSD can, but many external spinning disks cannot. This is also where crash-safe writing matters most, because a quarter-terabyte file you cannot play is an expensive mistake. We wrote about how KeepTake protects long recordings in a separate post.
FFV1: lossless without the absurdity
FFV1 is a lossless video codec used by archives, including national libraries, precisely because it is mathematically lossless: decode it and you get back the exact pixels that went in. At 1080p30 webcam content it lands around 27 GB per 30 minutes in our recordings, roughly a ninth of uncompressed.
That is still a lot of data, but it is in workable territory. An hour-long masterclass recording is around 55 GB instead of 480 GB. FFV1 costs more CPU than raw (it is doing real compression work), but it is designed to compress and decompress reliably in real time.
KeepTake records FFV1 into MKV, and the container choice is deliberate. MKV is far more tolerant of interruption than MOV or plain MP4, so a crash mid-recording still leaves you with a playable file.
If “lossless” matters to you but 8 GB per minute does not fit your life, FFV1 is the answer.
H.265: the sane default for keeping
H.265 (HEVC) is a modern lossy codec, and at webcam content it is remarkably efficient. At 1080p30, a constant-quality H.265 recording comes out around 1.3 GB per 30 minutes. That is roughly 20x smaller than FFV1 and about 180x smaller than uncompressed.
Lossy does not mean bad. At a sensible quality setting, H.265 webcam footage is visually indistinguishable from the source for almost every purpose: talking-head videos, courses, interviews, meetings. KeepTake records H.265 in constant-quality mode with a quality slider, which means the encoder maintains a consistent visual quality and spends bits only where the image needs them, instead of chasing a fixed bitrate. On Apple Silicon the encoding runs on the VideoToolbox hardware encoder, so it barely touches the CPU.
The one real caveat: if you plan aggressive color grading, heavy post-processing, or multiple re-encode generations, lossy source footage gives the process less to work with. For footage that goes more or less straight to viewers, H.265 is the right call.
H.264: maximum compatibility
H.264 is the older lossy standard. It compresses less efficiently than H.265, so files are larger at the same quality, but literally everything plays it: old software, old TVs, every editor, every browser. If your recording has to go somewhere unknown or into a picky toolchain, H.264 MP4 is the safe handoff format.
So which should you record in?
Our honest recommendations:
- Default choice: H.265 constant quality. Great quality, small files, hardware encoded.
- Footage you will edit and grade seriously: FFV1 lossless. It costs ~27 GB per 30 minutes and gives your edit a perfect source.
- Unrepeatable, must-be-perfect masters with disk to spare: uncompressed, at ~8 GB per minute, eyes open.
- Compatibility above all: H.264.
Record big, compress later
Here is the workflow we actually use: record in FFV1 or high-quality H.265, edit or review, then batch-compress what you are keeping. KeepTake has a built-in compressor tab for exactly this. It batch-converts recordings to hardware H.265, software x265, or AV1 for the smallest archives, and its safe-delete verifies the new file’s integrity before moving the original to the Trash. It never permanently deletes anything, because a compressor that deletes originals on faith is how you lose footage.
Storage is cheap and re-shooting is expensive. When in doubt, record one step higher in quality than you think you need. You can always compress down afterwards; you cannot compress up.