Bit depth is how many steps the camera uses to describe brightness in each colour channel. More steps do not make a picture sharper or more colourful. They give a grade more room before the gaps between steps become visible.
How many colours does each bit depth store?
Each extra bit doubles the number of levels per channel, and a pixel has three channels, so the totals climb fast:
| Bit depth | Levels per channel | Colours in total | Legal range for video |
|---|---|---|---|
| 8-bit | 256 | about 16.7 million | 16 to 235, so 220 usable levels |
| 10-bit | 1,024 | about 1.07 billion | 64 to 940, so 877 usable levels |
| 12-bit | 4,096 | about 68.7 billion | Mostly raw and cinema formats |
The legal range column is the one people miss. Most video does not use every code value: in 8-bit, black sits at 16 and white at 235. That leaves 220 steps from black to white, not 256. What Rec.709 is explains where those numbers come from.
Can you see the difference straight from the camera?
Rarely. A well exposed clip in a standard picture profile looks almost identical in 8-bit and 10-bit on a normal screen, because 220 steps is enough to describe a finished image without visible jumps. Most of what you watch online is 8-bit.
Where 8-bit sometimes shows even ungraded is in very smooth gradients: a clear blue sky, a studio wall, a sunset. If the encoder is also starved of bitrate, those areas can show faint contour lines. That is usually compression making the problem worse, and it is the same problem.
Why does 10-bit matter when you grade?
Because grading stretches the image, and stretching spreads the steps apart.
Say a sky occupies 30 levels in an 8-bit file. Add contrast and those 30 levels now cover a range that would ideally need 60. There are still only 30 distinct values, so neighbouring areas jump from one shade to the next with nothing in between. That jump is banding. In 10-bit the same sky starts with roughly four times as many levels, so the stretch has spare values to fill the gaps.
The same thing happens with a strong creative LUT, which is why a look that is fine on one clip can band on another. Lowering the amount helps; LUT intensity explained covers how each app does it.
Why does log footage need 10-bit most?
Log is the extreme case of stretching. A log profile squeezes the camera's whole dynamic range into a flat, grey signal, so each stop of light gets fewer code values than it would in a normal profile. Then the conversion to Rec.709 stretches it all back out. Log footage explained covers why the picture looks grey to begin with.
In 10-bit that round trip is comfortable. In 8-bit it is a gamble: expose well, convert gently, and it can hold; underexpose and lift the shadows, and it bands and turns noisy. The conversion LUT itself is not the cause, and choosing the right Rec.709 LUT will not rescue an 8-bit file that was pushed too far.
The camera guides on this site make the same point per brand. The S-Log3 guide and the D-Log guide both recommend 10-bit recording wherever the body offers it.
Is bit depth the same as chroma subsampling or bitrate?
No. They are three separate settings that often appear together in camera menus:
- Bit depth (8 or 10-bit) is how many levels each channel has.
- Chroma subsampling (4:2:0, 4:2:2) is how much colour detail is stored relative to brightness detail. 4:2:2 keeps more, which helps keying and fine colour edges.
- Bitrate is how much data per second the encoder may use. Too little and fine gradients turn blocky regardless of bit depth.
Cameras often tie them together, so a 10-bit mode may only exist with HEVC (H.265) or with 4:2:2. Read the menu as one combination, not three independent choices.
Which should you record?
| Situation | Record | Why |
|---|---|---|
| Log profile, any camera | 10-bit | The conversion stretches the image hard |
| Footage you plan to grade heavily | 10-bit | More room before banding |
| Standard profile, little or no grading | 8-bit is fine | Smaller files, easier playback |
| Older computer, long projects | 8-bit, or 10-bit with proxies | Some 10-bit files are heavy to play back |
| Delivery for the web | Either source; deliver 8-bit | Most platforms and screens play 8-bit |
On action cameras the same logic applies: the GoPro LUT guide notes that newer models offer a 10-bit mode, which takes a flat profile and a LUT better.
What should you do with 8-bit footage you already have?
Grade it gently. Get exposure and white balance right first, keep contrast changes modest, and watch skies and walls on a large screen before you commit. A creative look that bands at 100% often holds at 70%. For log shot in 8-bit, a conversion that keeps the shadows where they are is safer than one that lifts them.
When a whole card needs the same conversion, the work is applying it consistently and pulling back the clips that strain. BatchTone reads H.264 and HEVC in 8 or 10-bit, applies one .cube across a folder of clips and photos, and gives every file its own strength slider, so the sky shot that bands can sit lower than the rest. A frame from each clip is previewed with the LUT applied before anything is exported, the audio is copied untouched, and the originals are never modified. It runs on macOS 12 or later on Apple Silicon. Batch applying a LUT to video lists the formats, and there is a seven-day trial with everything unlocked.
Common questions
Do I need a 10-bit monitor to benefit from 10-bit footage?
No. The benefit of 10-bit recording is in the grade, which happens on numbers, not on the screen. A 10-bit display helps you see subtle banding while you work, but footage recorded in 10-bit holds up better even if you grade on an ordinary screen.
Is 10-bit the same as HDR?
No. HDR delivery uses 10-bit because it covers a much wider brightness range, but plenty of 10-bit footage is standard dynamic range. Recording 10-bit log and delivering ordinary Rec.709 is the most common use.
Does exporting to 8-bit throw the advantage away?
No. The grade happens on the 10-bit source, so the stretched values are calculated with all the in-between shades available. Rounding the finished result to 8-bit for delivery costs far less than grading an 8-bit source.
Why are 10-bit files sometimes slow to edit?
They are usually HEVC, which is harder to decode than H.264, and sometimes 4:2:2, which some hardware decoders do not support. Proxies or an intermediate format such as ProRes make editing smooth again.