Colour depth, also called bit depth, is the number of bits used to store the colour of each pixel, which determines how many distinct colours an image can contain. An 8-bit-per-channel image holds about 16.7 million colours, enough for photographs to look smooth and natural, while lower depths hold fewer colours and higher depths hold far more. Colour depth shapes how rich an image looks, how large its file is, and whether gradients appear smooth or banded. This guide explains what bit depth means, the common depths, and when each matters.

What bit depth actually means

A digital image stores each pixel as a number, and bit depth is how many bits that number uses. More bits mean more possible values, and therefore more possible colours. One bit gives two values, black or white. Eight bits give 256 values. Because colour images use three channels, red, green, and blue, the total is the per-channel figure cubed. This is why bit depth is the direct measure of how many colours an image can represent, and why a higher depth allows finer distinctions between shades.

Bits per channel versus bits per pixel

The terminology causes confusion because bit depth is quoted two ways. An "8-bit" image usually means 8 bits per channel, which is 24 bits per pixel across the three RGB channels. So "8-bit" and "24-bit" can describe the same image, one counting per channel, the other per pixel. When transparency is included as a fourth channel, that adds another 8 bits for 32 bits per pixel. Knowing which figure someone means avoids the common mix-up, and per-channel is the more useful way to think about it.

1-bit and indexed colour

At the low end, 1-bit images store only two colours, black and white, useful for line art and fax-style graphics. Indexed colour goes a step up, storing a small palette, up to 256 colours in 8-bit indexed mode, and referencing each pixel by its palette slot. This is how GIF and 8-bit PNG work, and it keeps files tiny for simple graphics with few colours. The limit shows on photographs, where 256 colours cannot render smooth gradients, producing visible bands instead of a natural blend.

8-bit per channel: the standard

The workhorse depth for everyday images is 8 bits per channel. This gives 256 levels per channel and about 16.7 million total colours, which is more than the eye can distinguish in most scenes, so photographs look smooth and natural. JPG, standard PNG, and WebP all commonly use 8-bit-per-channel colour. For the web, screens, and ordinary photography, this is the standard and the right choice, offering full-looking colour at a sensible file size. It is what "true colour" refers to.

High bit depth: 10, 12, and 16-bit

Professional and HDR work uses higher depths. 10, 12, or 16 bits per channel give billions of colours and far finer gradations, which matters for heavy editing and high-dynamic-range imagery. The benefit is headroom: with more levels between shades, you can push exposure and colour in editing without the gradient banding that 8-bit would show. The cost is much larger files. Camera RAW, professional editing, and HDR displays use high bit depth, but for a finished web image, 8-bit is delivered because screens and formats expect it.

What is colour banding

Banding is the visible artifact of too few colours. When a smooth gradient does not have enough distinct shades, it breaks into visible steps or bands instead of blending seamlessly, most obvious in skies and soft shadows. Low bit depth is the usual cause, which is why an indexed 256-colour image bands where an 8-bit-per-channel one does not. Editing in high bit depth and only reducing to 8-bit at the end helps avoid it, because the extra levels absorb the adjustments that would otherwise create bands.

Bit depth and transparency

Transparency is stored as an extra channel, the alpha channel, with its own bit depth. An 8-bit alpha channel gives 256 levels of transparency, from fully opaque to fully clear, allowing soft, feathered edges and partial see-through effects. This is why a 24-bit RGB image with 8-bit alpha is described as 32-bit. Formats like PNG and WebP support this full alpha channel, while GIF offers only a single fully-transparent colour, which is why GIF edges look hard and PNG edges can be smooth.

Bit depth and file size

Higher bit depth means more data per pixel and therefore larger files. Doubling the bits per channel roughly doubles the raw data before compression. This is why 16-bit images are reserved for editing and archiving rather than delivery, and why finished web images are 8-bit. There is no point serving a 16-bit image to a screen that cannot show the extra colours; it only wastes bandwidth. Match the bit depth to the purpose: high for editing headroom, standard 8-bit for delivery.

How bit depth relates to colour space

Bit depth and colour space are different but related. Bit depth is how many steps there are; colour space is the range those steps cover. A wide colour space with low bit depth spreads few steps across a broad range, risking banding, while high bit depth gives enough steps to fill a wide space smoothly. This is why wide-gamut HDR content benefits from high bit depth: the extra range needs extra steps to stay smooth. The two work together to determine how rich and smooth an image looks.

Choosing the right bit depth

For most work the choice is easy. Use 8-bit per channel for finished web images and everyday photography, since it looks full and keeps files reasonable. Edit in 16-bit if you are making heavy tonal or colour adjustments, to preserve headroom and avoid banding, then export to 8-bit for delivery. Use indexed colour only for simple graphics with few colours where tiny files matter. Match the depth to the task and you get smooth images at sensible sizes without paying for colours no screen will show.

Bit depth in common formats

Each image format supports particular bit depths. JPG is always 8 bits per channel, PNG supports 8 and 16-bit plus indexed colour, and GIF is limited to 8-bit indexed with a 256-colour palette. WebP and AVIF handle 8-bit and, for AVIF, higher depths suited to HDR. This is part of why format choice matters: a GIF cannot hold a smooth photographic gradient, while a 16-bit PNG can preserve editing headroom. Knowing each format's depth explains what it can and cannot represent.

HDR and bit depth

High dynamic range imagery depends on high bit depth. HDR content spans a much wider range of brightness, and needs 10 bits per channel or more to render it without banding. An 8-bit image simply does not have enough levels to describe the full range from deep shadow to bright highlight smoothly. This is why HDR photos, video, and displays all specify higher bit depths. As HDR screens become common, higher bit depth is moving from a professional concern to a mainstream one.

Editing in high bit depth

Even when the final image will be 8-bit, editing in a higher depth pays off. Working in 16-bit gives thousands of extra levels between shades, so heavy adjustments to exposure, contrast, and colour do not create banding. Each edit nudges values around, and with 8-bit's limited steps those nudges can open visible gaps in gradients. High bit depth absorbs them. The professional habit is to edit in 16-bit and only convert to 8-bit as the last step before export, preserving smoothness through the whole process.

Why screens and delivery use 8-bit

Most screens and web formats deliver 8-bit for a practical reason. The eye can barely distinguish more than 8-bit's colours in a normal image, so the extra data of higher depth is wasted on delivery. Serving a 16-bit image to an 8-bit screen only doubles the file size for no visible benefit. This is why the workflow separates editing from delivery: keep high depth while you work for the headroom, then export to 8-bit for a file that looks identical but is far smaller.

Reducing colours deliberately

Sometimes lowering bit depth is the goal, not a compromise. Reducing a simple graphic to indexed colour with a small palette makes the file tiny, ideal for logos, icons, and flat illustrations with few colours. The trick is choosing a palette that covers the image's actual colours, and using dithering, scattering pixels of available colours to simulate missing ones, to soften any banding. For the right kind of image, deliberate colour reduction is a powerful compression technique rather than a loss.

Common bit depth mistakes

A few misunderstandings cause trouble. Saving a photograph as an indexed 256-colour image bands the gradients, because there are not enough colours for smooth blends. Editing heavily in 8-bit rather than 16-bit introduces banding that could have been avoided. And serving 16-bit images to the web wastes bandwidth on colours no screen shows. Each is easy to avoid: use 8-bit per channel for photos and delivery, 16-bit for heavy editing, and indexed colour only for simple graphics. Match the depth to the task.

The bottom line on bit depth

Bit depth sets how many colours an image can hold, and the right choice is usually straightforward. Use 8 bits per channel for finished web images and everyday photography, where its 16.7 million colours look full and natural at a sensible file size. Edit in 16-bit when making heavy adjustments to avoid banding, then export to 8-bit for delivery. Reserve indexed colour for simple graphics with few colours. Match the depth to the task and you get smooth images without paying for colours no screen will ever show.

What is colour depth or bit depth?

It is the number of bits used to store each pixel's colour, which sets how many colours an image can contain. An 8-bit-per-channel image holds about 16.7 million colours.

What does 8-bit colour mean?

Usually 8 bits per channel, giving 256 levels each and about 16.7 million colours across RGB. It is the standard for web images and photography, looking smooth and natural at a sensible file size.

Why do I see bands in my gradients?

That is colour banding, caused by too few distinct shades to render a smooth gradient. Low bit depth is the usual cause; editing in high bit depth and exporting to 8-bit at the end helps avoid it.

When should I use 16-bit colour?

For heavy editing and HDR work, where the extra levels give headroom to adjust exposure and colour without banding. Export the finished image to 8-bit for web delivery.

What is the difference between bits per channel and bits per pixel?

Bits per channel counts each of red, green, and blue separately; bits per pixel is the total. An 8-bit-per-channel image is 24 bits per pixel, or 32 with an alpha channel.

What bit depth does JPG use?

JPG is always 8 bits per channel, about 16.7 million colours. PNG supports 8 and 16-bit plus indexed colour, while GIF is limited to 8-bit indexed with a 256-colour palette.

Why does HDR need high bit depth?

Because HDR spans a much wider brightness range and needs 10 bits per channel or more to render it without banding. An 8-bit image lacks enough levels to describe the full range smoothly.

Should I edit in 8-bit or 16-bit?

Edit in 16-bit for heavy adjustments, since the extra levels prevent banding, then export to 8-bit for delivery. The final image looks identical but the editing stays smooth.

What is dithering?

Dithering scatters pixels of the available colours to simulate ones the palette lacks, softening the banding that low bit depth causes. It lets a small indexed palette approximate a smoother gradient in simple graphics.

Does higher bit depth mean a larger file?

Yes. More bits per channel means more data per pixel, so doubling the depth roughly doubles the raw size before compression. This is why 16-bit is used for editing and 8-bit for web delivery.