Surveillance video compression explained

Guard watching surveillance video

Surveillance cameras generate large volumes of data. Without compression, the video that needs to be stored and transmitted would quickly become difficult to manage. Compression reduces those requirements while preserving the detail needed and the ability to investigate.

With compression in place, running dozens or hundreds of cameras 24/7 becomes entirely feasible. But scale brings its own challenge. Even compressed video adds up across a large deployment, and storage costs grow with every camera and every day of retention. The more efficient the compression, the lower the running cost over the lifetime of the system.

crowded city street in dublin

How compression got smarter

Early surveillance systems compressed each frame independently, resulting in high bitrates and storage costs. The shift came with video codecs that didn't just compress single frames but looked at what changed between them. Encoding only the differences brought bitrates down dramatically and made higher-resolution cameras viable at scale.

The more significant shift came not from a new codec, but from intelligence applied to encoding itself. Instead of compressing every part of every frame equally, smart encoding analyzes scene content in real time. Important areas get more bandwidth, while static backgrounds get less.

How video compression works

Overview highway terminus intersection

Video doesn't travel from camera to screen in a single step. The camera captures raw image data, encodes it, sends it over a network and delivers it to storage or a viewing client. Compression happens at the encoding stage. This takes place inside the camera itself. That's the only point in the chain where the original, uncompressed image still exists, which makes it the most important step to get right.

The component that does the work is called a codec, short for coder-decoder. It's a pair of algorithms working together: one compresses the video at the camera, the other decompresses it at the receiving end. Codecs from different standards aren't compatible, so an H.264-compressed video can't be decoded by an H.265 decoder. That's why modern cameras often support multiple codecs.

More advanced codecs require more computing power, which introduces a small delay between capture and display. For most surveillance use cases, the effect is barely noticeable. Applications that rely on live interaction, such as PTZ control, are generally more sensitive to latency.

What compression actually does

Not all data in a video frame is equally valuable. A static wall or an unchanged stretch of road doesn’t need to be re-encoded continuously. Compression identifies and removes that redundancy while preserving what matters: motion, faces, license plates, events.

Compression works at two levels:

  • Within a single frame, called intraframe compression

Details that the human eye can't perceive are removed. The frame looks the same. The file is a fraction of the size.

  • Across frames over time, called interframe compression

Only the pixels that have changed are encoded. Everything else remains unchanged and doesn’t need to be transmitted again. In a typical surveillance scene, much of the image remains the same from one frame to the next, significantly reducing the amount of data that needs to be stored and transmitted. 

Surveillance video compression is lossy by design. Unlike lossless compression, which preserves every bit of the original data, lossy compression removes information unlikely to affect the video's usefulness. The alternative would generate files that are too large for continuous recording and transmission. 

Lossless compression is mainly used for document storage and is normally not used for video applications except film production and digital archiving, where preserving every detail outweighs storage and  bandwidth requirements. Lossless bitrate reduction is simply too low; bitrate would end up far beyond what network or storage infrastructure can accommodate.

Illustration showing how frames change over time Image to the left: Within a single frame, Image to the right: Across frames over time

Compression standards: H.264, H.265 and beyond

The codec you choose shapes everything downstream: bandwidth requirements, storage costs, compatibility with VMS platforms and client devices, and the quality of data available for AI analytics. Three standards are worth noting.

H.264

Advanced Video Coding (AVC)

H.264 has been the backbone of network video for over 20 years. It is reliable, widely compatible, and works with virtually every VMS platform and client device on the market.

The limitation is bandwidth. Maintaining visual clarity at high resolutions requires high bitrates, and across large camera deployments, this adds up. H.264 also doesn't support video wider than 8K, which is increasingly relevant as camera resolutions rise.

20+ years in use

H.265

High Efficiency Video Coding (HEVC)

H.265 improved on H.264 by delivering comparable quality at 15 to 40 percent lower bitrates. The challenge has always been licensing. Restrictive patent terms made it effectively impossible for browser vendors to include native decoders, leaving many users to install them manually.

These compatibility issues haven't gone away. For deployments where client compatibility isn't a concern, H.265 still delivers real efficiency gains. However, for new deployments, a stronger option is now available.

Limited browser support

AV1

Alliance for Open Media (AOM)

AV1 was developed by Alliance for Open Media, a coalition including Google, Amazon, Microsoft and Netflix, and released in 2018 as a royalty-free, open-source codec. Designed for modern video streaming, it delivers significantly higher compression efficiency than earlier standards. It also supports features such as toggleable overlays that embed metadata directly into the video stream.

Compared with H.264, AV1 typically reduces bitrate by around 40 percent while maintaining similar image quality. It also benefits from broad support across modern browsers, operating systems and mobile platforms, making adoption easier and reducing licensing concerns.

For surveillance, that translates into lower storage costs and less network strain. Preserved image detail also means AI analytics work with better visual data. Cameras that support AV1 can run H.264 and H.265 alongside it, making transitions easier to manage.

Recommended for new deployments

AI’s role in video compression

Content-aware encoding uses AI to identify which parts of a scene contain information that matters. More bitrate can then be allocated to those areas, while static background elements are compressed more aggressively. The result is lower bandwidth consumption without a corresponding loss of useful detail.

Image quality matters for analytics. Object detection and license plate recognition are typically performed on the camera before compression, since the original image contains more information than the encoded stream.

Generative AI takes a fundamentally different approach. Instead of encoding the original image, it reconstructs parts of the video during playback. That may be acceptable for entertainment streaming. Surveillance footage often needs to support investigations and evidentiary use, which places different demands on accuracy. 

Stefan Lundberg is talking about AV
"Consider a retail chain operating 2,000 cameras across 150 locations, all streaming at 1080p or higher. Using H.264, that generates terabytes of data per day. Switching those streams to AV1 could cut storage needs by roughly 40%." Stefan Lundberg, Senior Expert Engineer at Axis Communications

The value of getting video compression right

When compression is configured well, it reduces bandwidth and storage demands significantly without compromising the image quality that makes surveillance video useful. The impact extends beyond storage savings, affecting everything from retention times to network capacity and cloud costs.

A symbol showing a computer monitor with a cogwheel and a highlighted dollar sign.

Longer retention at no extra cost

Lower bitrate means less storage consumed per camera per day. The same hardware that previously held 30 days of video recordings can now hold 60. For organizations with retention requirements driven by insurance or forensic needs, that's a direct cost saving without any infrastructure investment.
AXIS Camera Station Device Compatibility Tool icon

More cameras on the same network

Bandwidth is a fixed resource. When each camera consumes less of it, the same network infrastructure supports more cameras, or higher-resolution cameras, without upgrades. Useful for both new deployments and expanding existing systems.
A symbol showing a camera connected to a cloud with a play symbol

Lower cloud costs

Transmitting video to the cloud is expensive, and costs scale directly with bitrate. Efficient compression helps keep storage and bandwidth requirements under control, particularly in larger deployments where even small reductions can have a significant impact over time.

A note on comparing compression performance

Bitrate alone is not a meaningful measure of compression performance. A camera can achieve a lower bitrate by discarding more visual information. This, in turn, may reduce image quality and limit the video's usefulness. What matters is how much detail is preserved at a given bitrate. What does the video actually look like, and does it retain the detail needed for analytics and forensic use? That comparison may be hard. Two cameras placed side by side don't capture identical images. Because no two scenes are identical, compression performance is best evaluated in the environment where the camera will actually be used.

Exploring video compression solutions

Servers and window with cityscape sunset

Axis Zipstream

Content-aware compression technology that reduces bandwidth and storage by 50 percent or more, while keeping forensic-grade image quality intact.
AXIS Q1715 mounted above escalator

AV1 Codec

The open, royalty-free codec that delivers high-quality video at low bitrates. Native support across browsers, operating systems and cloud platforms, without the licensing complications that held earlier standards back.
Man watching data insights dashboard

Video management software

Video management software is the interface through which operators interact with the surveillance system. It provides access to live video, recordings and system events from one place.

Implementation Considerations

Compression affects more than storage and bandwidth. The settings you choose influence both system efficiency and the quality of the recorded video.

Beltway with cars marked with yellow boxes

Run AI analysis before you compress

Where AI runs in the workflow makes a real difference. AI-powered analytics, such as object detection and license plate recognition, should run on the camera before compression. Compressed video contains less visual information than the original, which affects how reliably analytics can interpret what they see.

Codec compatibility across your system

Every device in the chain needs to support the same codec, including cameras, video management software, storage servers, and viewing clients. Cameras that support H.264, H.265 and AV1 provide greater flexibility, making it easier to work with existing systems while supporting future upgrades.

Quality, bitrate, and bandwidth

Compression settings need to balance image quality against storage and bandwidth requirements. Too little compression increases resource consumption, while too much can obscure details important to investigations and day-to-day operations.

The right setting depends on the scene and the purpose of the recording. Test compression performance in your actual environment, not under ideal conditions. Design tools that estimate bitrate per camera and case can help you plan. 

Storage dimensioning

Bitrate determines how quickly storage fills up. Before deploying or expanding a system, calculate the expected bitrate per camera, then multiply by the number of cameras and the required retention period. 

A system that seems affordable at deployment can quickly become expensive if storage runs out sooner than expected, or if cameras are added later without revisiting capacity.

AV1 becomes the baseline

Support for AV1 continues to expand across cloud platforms and the wider software ecosystem. As decoder support becomes commonplace, many of the adoption barriers seen with H.265 are starting to fall, making AV1 a more practical choice for new surveillance deployments.

Cloud puts a premium on efficiency

Moving video to the cloud makes compression efficiency more important because storage and transmission costs scale with bitrate. In larger deployments, where video is retained for extended periods, even small reductions in bitrate can have a significant impact over time.

As cloud-based surveillance grows, image quality alone will no longer be sufficient as a measure of compression performance.

Smarter encoding, more granular control

Development is increasingly focused on making existing codecs work harder. Encoding adapts dynamically to scene content, time of day and activity levels. Features such as toggleable overlays embed metadata directly in the video stream, delivering video and structured data together without additional bandwidth.

Storage cost remains a key driver

Storage is one of the largest ongoing expenses in a surveillance system, and demand continues to grow as resolutions increase and retention periods lengthen. More efficient compression helps offset this growth. Lower bitrates reduce storage requirements and can also reduce the amount of hardware needed over the system's lifetime.

New standards will keep coming

Video compression is an active area of development, and new standards will continue to emerge. Migrating between standards takes time, and existing systems need to keep running. Successful new codecs respect what's already deployed and remain compatible with it.

Not all of the new standards will gain traction. Codec adoption depends as much on licensing and industry momentum as on technical performance. For most organizations, long-term flexibility is more valuable than trying to predict which codec will dominate next.

A note on generative AI compression

The industry is exploring generative AI as a compression method. Recordings may need to support investigations or serve as evidence, so the video must accurately reflect what the camera captured.

Businessman with tablet in his hands

Smart compression in practice

A white paper on how content-aware compression reduces storage requirements while preserving the detail needed for investigations.

Axis site designer genetec streamvault calculator

AXIS Site Designer

Estimating storage and bandwidth requirements before deployment removes much of the guesswork. AXIS Site Designer helps you configure devices, calculate system requirements, and generate a complete bill of materials.

Stefan Lundberg is talking about AV

AV1 explained

Stefan Lundberg, Senior Expert Engineer at Axis, explains what AV1 means for network video and how it compares with H.264 and H.265.

Servers and technician with laptop

Surveillance video storage

Where and how you store video determines what your surveillance system can deliver. The right storage architecture keeps video recordings accessible, costs predictable, and your system ready to scale.

Axis object analytics time in area

Video analytics

Modern video analytics can automatically identify what is happening in recorded video. Explore how AI and machine learning help organizations get more value from their video systems.

operator surveillance system room

Image usability

Surveillance video is only as useful as the detail it captures. Image usability is what determines whether a face can be recognized, a license plate read, or an event understood. The right balance of resolution, lighting and processing turns video into actionable evidence.