Color Basics: RGB, YUV, and Bit Depth

Why do video files store YUV instead of RGB? And what does the fraction-like number 4:2:0 actually mean?

Open a photo and the computer stores each pixel as red, green, and blue (RGB) values. But open a video file, and the encoder stores YUV instead—a term you may not be familiar with. Why don't videos use RGB? What does 4:2:0 mean, and is it a fraction? How different are 8-bit and 10-bit, really? And what are color spaces? Three questions map to three levels of understanding: the color model (why YUV exists), chroma subsampling (what 4:2:0 describes), and color depth with color spaces (8/10-bit and BT.709/BT.2020).

The Color Model: Which Three Numbers Describe a Color

The RGB Additive Model

The human retina has three types of cone cells, most sensitive to red (~560 nm), green (~530 nm), and blue (~430 nm) wavelengths. Digital color systems mimic this by combining three numeric values for red, green, and blue—this is additive color mixing.

RGB values are typically represented as 0-255 (8-bit per channel) or 0-1023 (10-bit per channel). The larger the three numbers, the brighter the color—(0,0,0) is pure black, (255,255,255) is pure white.

RGB is a display-oriented model. Monitors, phone screens, and projectors all accept RGB signals directly—each sub-pixel controls its brightness independently to blend the final color.

YUV / YCbCr: Why Video Doesn't Use RGB

If video stored RGB directly, there would be an efficiency problem: the human eye is far more sensitive to changes in luminance than to changes in chrominance. All three R, G, and B channels carry luminance information, so splitting them apart allows for more efficient compression.

YUV decomposes an RGB signal into one luminance component (Y, Luma) and two chrominance components (U/V, called Cb/Cr in digital video):

RGB → Y (luminance) + Cb (blue-difference) + Cr (red-difference)

The conversion formula (BT.601 standard, approximate):

Y  = 0.299 × R + 0.587 × G + 0.114 × B
Cb = 0.564 × (B - Y)
Cr = 0.713 × (R - Y)

Notice that the coefficients for Y are not uniform—green contributes the most to perceived brightness (0.587), blue the least (0.114), matching the spectral sensitivity of human vision.

The core advantage of YUV: since the human eye is less sensitive to color detail, you can reduce the resolution of the chrominance components with little perceptible loss—which is exactly what chroma subsampling (next section) does. This can cut the raw video data rate in half or even by two-thirds, with nearly invisible quality impact.

Note: CMYK (subtractive) is used in desktop publishing but isn't covered here since we're focused on digital video.

Chroma Subsampling: 4:4:4, 4:2:2, 4:2:0

This is the biggest "compression dividend" from YUV—drastically reducing data volume by lowering chroma resolution.

What the Numbers Mean

Chroma subsampling notation uses three numbers (e.g., 4:2:0) to describe the sampling ratio of luma to chroma within a group of pixels:

• First number (always 4): horizontal luma sampling reference
• Second number: chroma samples on the first row (relative to 4)
• Third number: chroma samples on the second row (for interlaced; in progressive equals the second row's relative count)

Intuitively, for a 4×2 pixel block:

4:2:0 is the absolute mainstream for consumer video. All Blu-ray movies, streaming content (YouTube, Netflix), and the vast majority of phone-recorded video use 4:2:0. It reduces chroma resolution to one-quarter of luma, yet the human eye barely notices.

4:4:4 is used in high-end production, computer screen capture, and text/UI content—where colored fringing around sharp edges (chroma artifacts) becomes visible at 4:2:0. 4:2:2 is a common intermediate format for broadcast and professional cameras.

When Does 4:2:0 Show Its Limits?

The reduced chroma resolution of 4:2:0 can become visible in these scenarios:

• Small red or blue text on a black background shows colored fringing at edges
• Green/blue screen chroma key compositing: insufficient chroma detail creates "crawling" edges around the keyed subject
• Skin tones in predominantly red scenes may exhibit a slightly blocky quality

For the vast majority of video content (live-action cinema, everyday recordings, animation), 4:2:0 is more than adequate.

Color Depth: 8-bit vs. 10-bit

Color depth in video typically refers to the number of bits per channel. (Note: same term as audio bit depth but a different concept entirely.)

8-bit with 256 steps per channel sounds like a lot, but it often falls short in gradient areas (skies, skin tones, light blooms). When adjacent gray levels become visibly different, color banding appears—smooth gradients turn into concentric rings or stripes.

10-bit boosts precision from 256 to 1024 steps per channel—4× the granularity. In HDR video, 10-bit is essentially mandatory because HDR's luminance range vastly exceeds SDR; 8-bit doesn't provide enough steps to cover that range without banding.

Note: 10-bit video requires a 10-bit display panel to render fully. 8-bit panels can simulate 10-bit via FRC (Frame Rate Control) dithering—close to, but not equal to, native 10-bit.

Color Depth and Bandwidth

The bandwidth scales dramatically—at 4K 60fps, 8-bit 4:4:4 raw bandwidth is about 12 Gbps (near HDMI 1.4's limit). Switching to 10-bit pushes it to roughly 18 Gbps (requiring HDMI 2.0 or higher). This is why a TV's HDMI version directly determines whether it can carry 4K HDR 10-bit signals.

Color Space: BT.709, BT.2020, and Gamut

Color space (gamut) defines "what range of colors can be represented." Even within the same RGB or YUV model, different color spaces cover different portions of the visible spectrum.

• BT.709 has been the HD video standard since the 1990s; all SDR video is based on it
• DCI-P3 is the standard gamut for digital cinema projection, widely adopted by Apple and high-end monitors ("Display P3")
• BT.2020 is the target gamut for UHD / 4K HDR video—far larger than BT.709, though very few consumer displays can fully cover it today

When a video's color space doesn't match the display: - BT.2020 video on a BT.709 screen → colors are "clipped" into the smaller gamut, appearing duller than intended - sRGB image on a DCI-P3 display → usually color-managed by the OS or application, mapped correctly

Color Depth vs. Gamut

These two concepts are often confused:

• Color depth = precision (how many steps per channel)
• Gamut = range (how large an area of color the space covers)

A larger gamut (BT.2020) requires higher color depth (10-bit or more) to avoid banding in transitions—just as a larger area needs a finer grid to remain smooth.

Common Misconceptions

1. "RGB is uncompressed and raw; YUV is lossy." RGB and YUV are both color models—neither is inherently lossy or lossless. Chroma subsampling (e.g., 4:2:0) is lossy, but the YUV model itself is not a compression scheme.
2. "10-bit panels display more vivid colors than 8-bit." Vibrancy depends on the gamut (BT.709 vs. DCI-P3). 10-bit determines gradient smoothness, not color intensity.
3. "4:2:0 discards 75% of the color data—quality must be terrible." 4:2:0 discards chroma information that the human eye is not sensitive to; it's nearly imperceptible in regular viewing. The impact is only significant in professional color grading or chroma keying.
4. "HDR just means brighter, so 8-bit compression is fine." HDR's luminance range is far wider than SDR; 8-bit precision isn't enough to avoid banding across that range. HDR standards require 10-bit minimum.
5. "A BT.2020 monitor shows 'true HDR'." Virtually no consumer display fully covers BT.2020. HDR displays typically target DCI-P3 or smaller gamuts, using tone mapping for out-of-range colors.

Practical Tips

• Everyday video publishing: shoot and export at YUV 4:2:0 / 8-bit / BT.709—the most widely compatible configuration. All streaming platforms work with this baseline.
• Footage for color grading: aim for 10-bit 4:2:2 or better when shooting. The stretching and compression applied during grading amplifies 8-bit quantization errors. If you only have 8-bit material, avoid extreme hue and saturation adjustments.
• Green/blue screen keying: use 4:2:2 or 4:4:4 footage when possible. With 4:2:0, applying a slight blur to the chroma channels before keying can mitigate jagged edges.
• Screen recording (tutorials/demos): 4:4:4 is ideal (sharp text and UI elements). On 4:2:0, chroma artifacts around static interface elements can be noticeable. Look for "lossless color" or similar options in your recording software.
• Watching HDR content: ensure the display, HDMI cable version, and playback software all support HDR; a missing link in the chain breaks the experience. Enable "Use HDR" in the OS settings.
• Don't use P3 mode on sRGB monitors: some monitors offer a P3 gamut mode but their panels are natively sRGB—this is a software mapping and causes oversaturation. True wide gamut requires panel-level support.

Further Reading

• Resolution and Clarity: From 480p to 4K — pixel count and how it relates to perceived sharpness
• Video Filters: From Color Adjustments to Effects — the color principles behind filter tool parameters (brightness/contrast/saturation/hue and which YUV components they affect)
• HDR vs. SDR: The Next Generation of Picture Quality — extending from color space to HDR's color and luminance standards
• LUTs and Color Grading: The Color Science Behind Filters — 3D LUTs are essentially look-up transforms within a three-dimensional color space

This site's video filter tool has sliders for brightness, contrast, saturation, and hue—after reading this article, you'll know that adjusting saturation essentially scales the UV components' Euclidean distance from the origin, and adjusting brightness changes the Y component value.

Related Tools

• Filter Video
• Video Converter