Learn
What Is Chroma Subsampling?
Video files are large. A single frame of uncompressed 4K video contains over eight million pixels, each described by multiple values for brightness and colour. At 24 frames per second, the data adds up rapidly. Every compression technique used in video production exists to reduce that data to something manageable, and chroma subsampling is one of the oldest and most effective of them.
Chroma subsampling works by reducing the amount of colour information in a video signal while keeping the brightness (luminance) information intact at full resolution. The result is a smaller file or a lower data rate, with a reduction in image quality that, in most viewing conditions, is invisible to the human eye. The reason it works is rooted in biology: human vision is far more sensitive to differences in brightness than to differences in colour. Chroma subsampling exploits this asymmetry, discarding colour data that the eye is unlikely to miss.
How human vision justifies it
The human eye contains two types of photoreceptor cells: rods, which are sensitive to light and dark, and cones, which detect colour. There are roughly 120 million rod cells in the human retina, compared with only about 6 million cones. The disparity is significant. We perceive fine spatial detail primarily through luminance (brightness differences), not through chrominance (colour differences).
This means that if you reduce the resolution of the colour information in an image while leaving the brightness information at full resolution, most people will not notice the difference under normal viewing conditions. The edges of objects, the texture of surfaces, and the fine detail that makes an image look sharp are all carried by luminance. Colour provides broad washes of hue and saturation, but it does not need to be resolved at the same level of detail.
This biological fact is the foundation on which chroma subsampling is built. It is not a workaround or a compromise born of technical limitation; it is a principled approach to compression that takes advantage of how the visual system processes information.
The notation explained
Chroma subsampling is described using a three-number notation, written with colons: 4:4:4, 4:2:2, 4:2:0. These numbers can seem opaque at first glance, but the system is logical once its structure is understood.
The notation describes how colour samples are distributed across a small block of pixels, specifically a block that is four pixels wide and two pixels tall. Each of the three numbers refers to a different aspect of the sampling.
The first number is the sampling reference width. It is almost always 4, indicating a reference block four pixels wide. This number establishes the baseline against which the other two are measured.
The second number is the number of colour (chroma) samples in the first row of the block. In 4:4:4, every pixel in the first row has its own colour sample: four out of four. In 4:2:2, only two out of four pixels in the first row have unique colour samples; the other two share colour information with their neighbours. In 4:2:0, again only two out of four pixels in the first row have colour samples.
The third number is the number of colour samples in the second row. In 4:4:4, it is four: full colour in both rows. In 4:2:2, it is two: the second row has the same colour sampling as the first. In 4:2:0, it is zero: the second row has no independent colour samples at all and instead uses the colour information from the first row.
This notation system is a compact way of describing how much colour data has been retained relative to the luminance data. The lower the second and third numbers, the more colour information has been discarded.
4:4:4: no subsampling
In a 4:4:4 signal, every pixel retains its full colour information. There is no subsampling at all. Every pixel has its own luminance value and its own chrominance values, resulting in the highest possible colour fidelity.
4:4:4 is used in contexts where colour accuracy and edge quality are paramount. Visual effects work, particularly green screen and blue screen compositing, benefits significantly from 4:4:4 because the edges between the coloured screen and the subject need to be as clean and well-defined as possible. When colour information is subsampled, those edges become softer and less precise, making clean keying more difficult.
Graphics, animation, and text rendering also benefit from 4:4:4, since hard edges between differently coloured elements can show artefacts at lower chroma resolutions.
The trade-off is data size. 4:4:4 files are substantially larger than their subsampled equivalents. A 4:4:4 ProRes file, for instance, will be roughly 50% larger than the same content encoded as ProRes 422. For productions with heavy storage requirements, this matters. Managing large 4:4:4 files across a production demands reliable cloud storage and thoughtful file organisation.
4:2:2: the professional standard
4:2:2 is the most common chroma subsampling scheme in professional video production. It halves the horizontal colour resolution while retaining full vertical colour resolution. In practical terms, every pair of horizontally adjacent pixels shares a single set of colour values, while each pixel retains its own luminance.
The reduction in data is meaningful: a 4:2:2 signal carries roughly two-thirds the data of a 4:4:4 signal of equivalent bit depth. This saving, combined with the fact that the quality reduction is negligible for most production purposes, makes 4:2:2 the standard for professional acquisition and broadcast.
Most professional cameras record in 4:2:2 or offer it as an option. Codecs commonly used in production, such as Apple ProRes 422 and Avid DNxHR, are built around 4:2:2 subsampling. Broadcast standards, including many HD and UHD delivery specifications, specify 4:2:2 as the minimum.
For video editors and colourists, 4:2:2 provides a good balance between quality and practicality. Colour grading at 4:2:2 produces clean results for most material. Keying against a green screen is more challenging than at 4:4:4 but still viable with well-lit screens and careful technique. The files are large enough to support professional work but small enough to be practical for storage and transfer.
4:2:0: the delivery standard
4:2:0 reduces colour resolution in both the horizontal and vertical dimensions. Only two colour samples exist in the first row of the reference block, and the second row has none of its own, borrowing instead from the first row. The result is colour information at one quarter the resolution of the luminance.
This is the chroma subsampling scheme used for nearly all consumer and distribution video. Blu-ray discs, streaming services, web video, and most broadcast television use 4:2:0. The H.264 and H.265 codecs that dominate internet video delivery are typically configured for 4:2:0 subsampling. The video file formats you encounter in daily viewing are almost universally 4:2:0.
For viewing on a screen at normal distances, 4:2:0 is visually indistinguishable from 4:2:2 or 4:4:4 in most content. The reduction in colour resolution is simply not perceptible under typical conditions. This makes it an excellent choice for delivery, where file size and bandwidth directly affect the viewer's experience (buffering, load times, and storage on devices).
However, 4:2:0 is less suitable as an acquisition or working format when the footage will undergo significant manipulation in post-production. Heavy colour grading can reveal the reduced colour information, particularly around high-contrast edges where colour transitions are abrupt. Chroma keying at 4:2:0 is more difficult and less clean than at higher subsampling rates. For this reason, shooting at 4:2:2 or higher and delivering at 4:2:0 is a common and sensible workflow.
When chroma subsampling matters
For most viewing situations, chroma subsampling is invisible. A viewer watching a film in a cinema, streaming a series at home, or scrolling through social media will not see the difference between 4:2:0 and 4:2:2 in normal content. The system works because it is designed around the capabilities of human vision.
Where subsampling becomes a meaningful consideration is in production and post-production workflows where the image will be manipulated.
Green screen and blue screen compositing is the most frequently cited example. The process of keying (separating the subject from the coloured background) depends on clean, well-defined colour edges. At 4:2:0, the colour information around the subject's edges is at lower resolution than the luminance, which can produce fringing, rough edges, and visible artefacts in the composite. Shooting at 4:2:2 or 4:4:4 gives the compositor cleaner data to work with.
Intensive colour grading is another case. When the colourist makes large adjustments to hue, saturation, or colour balance, the reduced colour information in a 4:2:0 signal can lead to visible banding or artefacts, particularly in areas with smooth colour gradients (skies, skin tones, out-of-focus backgrounds). Working with 4:2:2 or 4:4:4 source material gives the colourist more room to push the image without degradation.
Graphics and text overlaid on video can also be affected. Sharp-edged graphic elements placed over 4:2:0 video may interact poorly with the subsampled colour data, producing visible edges or colour fringing that would not appear with higher subsampling.
For content creators working with footage that will go through minimal post-production (a quick trim, a simple grade, and export), 4:2:0 acquisition is often perfectly adequate. The distinction becomes important when the pipeline involves significant manipulation of the image.
Practical implications for camera and codec choice
Understanding chroma subsampling helps inform decisions about which camera to use, which codec to record in, and how to configure the recording format for a given project.
When selecting a camera for a shoot that involves green screen work, the ability to record 4:2:2 internally (or to output 4:2:2 or 4:4:4 via HDMI or SDI to an external recorder) is a meaningful specification. Many consumer and prosumer cameras record only 4:2:0 internally but offer higher subsampling via their external outputs. An external recorder capturing 4:2:2 ProRes from a camera that would otherwise record 4:2:0 H.264 internally can make a significant difference to the quality of downstream compositing.
Codec selection is closely tied to subsampling. ProRes 422 is, as its name indicates, a 4:2:2 codec. ProRes 4444 supports 4:4:4 (the extra "4" indicates an alpha channel). DNxHR SQ and HQ are 4:2:2. H.264 and H.265, in their most common profiles, are 4:2:0. Understanding these relationships helps video editors and videographers choose formats that match their production needs without carrying unnecessary data overhead.
For projects that will involve a proxy workflow, the subsampling of the proxy files is less critical than that of the original camera media, since the proxy is used only for editing performance and the final grade and output work with the full-resolution originals. However, if the proxy will be used for rough compositing or colour work, retaining 4:2:2 in the proxy can be useful.
Storage and bandwidth implications are worth noting. Moving from 4:2:0 to 4:2:2 roughly doubles the chroma data (though not the total file size, since luminance is the larger component). Moving from 4:2:2 to 4:4:4 doubles it again. For productions managing terabytes of footage, these differences affect storage costs, transfer times, and backup strategies. A cloud workspace that handles large media files efficiently makes working with higher-quality formats more practical across distributed teams.
The relationship to bit depth
Chroma subsampling is often discussed alongside bit depth, and the two interact in important ways. Bit depth describes how many levels of brightness or colour each sample can represent: 8-bit provides 256 levels per channel, 10-bit provides 1,024, and 12-bit provides 4,096.
A 10-bit 4:2:2 signal carries more total colour information than an 8-bit 4:4:4 signal, because the increased precision of each sample compensates for the reduced number of samples. In practice, many professional workflows use 10-bit 4:2:2 as an optimal balance: the higher bit depth provides smoother gradients and more grading headroom, while the 4:2:2 subsampling keeps data rates manageable.
Understanding the interplay between subsampling and bit depth helps in evaluating camera specifications and codec options. A camera that offers 10-bit 4:2:2 recording is providing high-quality acquisition data that supports professional post-production. An 8-bit 4:2:0 recording, while perfectly adequate for delivery, offers less room for manipulation. Understanding colour space alongside subsampling and bit depth provides a more complete picture of how colour is captured, processed, and delivered.
Chroma subsampling and compression
Chroma subsampling is one layer of compression in a multi-layered system. A typical video codec applies several compression techniques in sequence: chroma subsampling reduces the colour data, spatial compression reduces redundancy within each frame, and temporal compression reduces redundancy between frames.
In codecs like H.264 and H.265, chroma subsampling is applied first, reducing the data before the codec's more complex compression algorithms take effect. The result is a compound saving: less data enters the encoder, and the encoder then compresses that data further.
Production codecs like ProRes and DNxHR apply lighter spatial compression and no temporal compression (each frame is encoded independently), which preserves editability and random access at the cost of larger files. These codecs rely more heavily on chroma subsampling for their data savings, which is why the subsampling scheme is part of the codec's name (ProRes 422, ProRes 4444).
Understanding this relationship helps clarify why changing the subsampling scheme has a significant effect on file size and data rate. It also explains why the same codec at different subsampling levels can produce noticeably different results when the footage is pushed in post-production.
Frequently asked questions
Can I see the difference between 4:2:2 and 4:2:0 on my monitor?
In most content viewed at normal distances, no. The difference becomes visible when the image is examined closely, particularly around high-contrast colour edges, or when the footage is heavily manipulated in post-production. For standard viewing of finished content, 4:2:0 and 4:2:2 are visually equivalent.
Does chroma subsampling affect still images?
Yes, though it is less commonly discussed in that context. JPEG images use a form of chroma subsampling (typically 4:2:0) as part of their compression. RAW and TIFF files retain full colour information. The same perceptual principle applies: the eye is less sensitive to colour detail than to luminance detail.
Should I always shoot in 4:4:4?
No. 4:4:4 produces the highest colour fidelity but also the largest files. For most production work, 4:2:2 is sufficient. 4:4:4 is most beneficial for VFX-heavy work, green screen compositing, and situations where the image will undergo extreme colour manipulation. Shooting 4:4:4 for a talking-head interview that will receive a light grade and be delivered at 4:2:0 produces no visible benefit.
Why do streaming services use 4:2:0?
Bandwidth and file size. Streaming services need to deliver video over the internet to millions of simultaneous viewers with varying connection speeds. 4:2:0 provides the best balance of quality and data efficiency for delivery. The quality reduction is imperceptible to viewers, while the data savings are substantial.
What does 4:1:1 mean?
4:1:1 reduces horizontal colour resolution to one quarter while maintaining full vertical colour resolution. It was used in some early DV and DVCPRO formats. It carries less colour information than 4:2:2 but distributes the reduction differently from 4:2:0. It is rarely encountered in modern production.
Does chroma subsampling affect audio?
No. Chroma subsampling applies only to video. Audio compression uses entirely different techniques. The audio formats used in video production (PCM, AAC, AC3) have their own compression methods unrelated to chroma.
Can I convert 4:2:0 footage to 4:2:2?
You can transcode 4:2:0 footage to a 4:2:2 codec, but doing so does not restore the colour information that was discarded. The transcoded file will be larger and will play well in professional editing applications, but the underlying colour resolution remains at 4:2:0. You cannot add information that was never captured.
How does chroma subsampling relate to HDR?
HDR (High Dynamic Range) video typically uses 10-bit colour depth, which interacts with chroma subsampling. Many HDR delivery specifications call for 10-bit 4:2:0, which provides more colour precision per sample than standard 8-bit 4:2:0. The higher bit depth partially compensates for the subsampling by providing smoother colour transitions. HDR acquisition for professional work often uses 10-bit 4:2:2 or higher.
What subsampling do smartphone cameras use?
Most smartphone cameras record in 4:2:0, using codecs like H.264 or H.265. Some recent models offer higher subsampling options in their "pro" or "cinema" modes, but 4:2:0 remains the default. For most smartphone content, which undergoes minimal post-production, this is entirely appropriate.
Does LUT application behave differently at different subsampling levels?
Applying a LUT to 4:2:0 footage can reveal subsampling artefacts if the LUT makes aggressive colour changes, particularly in areas with subtle colour gradients. The same LUT applied to 4:2:2 or 4:4:4 footage will produce cleaner results. For light creative grades, the difference is negligible; for heavy transformations, it can be visible.