LESSON 86
Codecs and bitrates
H.264, H.265, ProRes and RAW, compared.
IN THIS LESSON
- What a codec does
- H.264 and H.265
- ProRes and RAW
- Bitrate and chroma
- Practice: what to record
What a codec does
A codec is the method used to squeeze video into a smaller file and unpack it again for playback. Raw picture data is enormous, so almost every camera compresses. Most delivery codecs look for parts of a frame, and parts of neighbouring frames, that repeat, and store only the changes. That saves a great deal of space but means the computer must rebuild many frames from others before it can show one. Editing codecs make different trade-offs: they compress each frame on its own, so files are bigger but every frame can be decoded immediately.
H.264 and H.265
H.264, also called AVC, was finished in 2003 and is still the most widely used video format, found on Blu-ray, streaming services and broadcast. Its successor H.265, or HEVC, was published in 2013 and can reach the same quality at roughly 25 to 50 percent less data. As an example from the standard's own history, 4K at 30 frames needs about 50 megabits a second in HEVC against about 100 in older H.264 profiles. HEVC's Main 10 profile also stores ten bits per colour, which matters for grading. Both are patent-licensed and both are heavy to decode on older computers.
ProRes and RAW
Apple ProRes, introduced in 2007 with Final Cut Studio 2, is an intermediate codec built for post-production. It is lossy but designed to look visually lossless, compresses each frame on its own, and its 422 versions use 4:2:2 colour at 10 bits. ProRes 422 HQ runs at about 220 megabits a second for HD. RAW goes further and stores the sensor's data before the camera turns it into a finished picture. CinemaDNG, an Adobe-led open format announced in 2008, holds each frame as a DNG image, and Apple added ProRes RAW in 2018.
Bitrate and chroma
Bitrate is how much data each second of video gets, usually in megabits a second. More bits mean fewer compression marks, especially in fine texture, water, leaves and grain. Many codecs vary the rate with the content, so an average figure hides peaks. Chroma subsampling is another saving: because our eyes are less sharp for colour than for brightness, 4:2:2 and 4:2:0 store colour at reduced resolution. That is invisible to most viewers, but it limits how hard you can push a green-screen key or a grade.
Practice: what to record
This is working practice. For quick social pieces, the camera's H.264 or H.265 at its highest bitrate is fine. If you plan a heavy grade, a green screen or a long edit, record ProRes or another intra-frame codec at 10 bits if the camera allows. Save RAW for jobs where you control the light poorly and want the most room to recover it, and budget for the storage. Whatever you pick, test it before the job: record a minute, copy it to the edit computer and scrub through it. If playback stutters, plan to make proxies, and if the card fills faster than you expected, buy more cards rather than dropping the bitrate on the day.
| Codec | Year | Built for | Notes |
|---|---|---|---|
| H.264 / AVC | 2003 | Delivery and capture | Most common format; long-GOP |
| H.265 / HEVC | 2013 | Delivery and capture | 25 to 50 percent less data than H.264 at equal quality |
| Apple ProRes | 2007 | Editing and finishing | Each frame stored alone; 422 HQ about 220 Mb/s in HD |
| CinemaDNG | 2008 | RAW capture | Open, Adobe-led; one DNG image per frame |
Try it today
- 01Record the same 10-second shot in two codecs your camera offers.
- 02Compare the file sizes and write down the megabits per second.
- 03Push the exposure two stops in your editor on both clips.
- 04Look closely at shadows and skies for blocks and banding.
- 05Choose a default codec for your next shoot and note why.