intoPIX experts answer the web's most frequently asked questions about video compression.
Why do we need video compression?
Because video produces ridiculous amounts of data. As an example: you take around 300-400 holiday photos with your phone and a “storage almost full” message will pop up. Now, still images are usually “lossy” compressed at 10:1 with JPEG. So uncompressed, you could only take 30-40 images. Cinema movies are often shot at 24 or 48 frames per second. Without compression, that means it will produce the same amount of RAW data within one second, as the ones that filled up your phone’s storage. Then, imagine how much that is for a full movie. To spare you from calculating, generally said, an average 90 minute blockbuster creates up to 1TB of raw video data. And because nobody wants to start streaming Netflix two days before they actually fall into their couch, you need video compression.
Where is video compression used?
Due to the previously mentioned data amount issues, compression is basically used everywhere video needs to be managed, transmitted, or stored.
It is oftentimes not visible to the end-user if a video is compressed or not, so here is a list of example where compression is used and why it is used specifically there:
To offer best quality, larger resolutions, higher frame rates, higher dynamic range (HDR) and reduce the amount of hard drives film studios send to movie theaters.
Advanced Driver-Assistance Systems
To build autonomous driving vehicles that are safe, highly connected - within the car and with the outside world - while keeping power consumption low.
To upgrade from HD to 4K or even 8K video, without costly infrastructure renovations and deliver videos to all our devices (from TV to tablets to mobile phone) all across the world.
Professional AV installations
To minimize the investment in cables (e.g. Cat5 instead of Cat6) and reduce room needed for cables (e.g. single 1GB Ethernet instead of multiple) in office buildings, lecture halls, houses of worship, smart homes, transportation terminals, etc.
VR / AR gear
To allow users a wireless experience through private Wi-Fi or 5G technologies using the most reactive video compression.
To capture and send high resolution images (e.g. 16K x 16K pixels) all the way down to earth in a reasonable time.
To reduce the amount of electronic signals sent, managed, and stored within a device, thus reducing memory, interfaces, size and power consumption.
And many other applications!
How does video compression work?
Simply said, you want to transform the video data from a large stream of pixel information to a more efficient stream of bits.
Most algorithms use 3 main steps to do so:
Decorrelation of information
The pixels of a picture are reorganized to ease entropy coding efficiency, sort the data, classify the data, ... The most popular mechanisms for this are the “Discrete Cosine Transform” (DCT) and the “Discrete Wavelet Transform” (DWT).
Codes redundant information into bits to make it smaller. In a simple way, let’s say you have an image of a tomato. Here, the algorithm will assign the shortest code to a very frequent red, while colors that appear much less frequent will be assigned with a much longer code. This helps you minimize the final code in a whole.
Here you eventually differentiate between lossy and lossless compression. Depending on the targeted compression ratio or targeted bitrate, you’ll keep only the most relevant data for achieving the best possible result.
How to choose the right video codec?
The most important thing is to identify your needs and the challenge you want to solve first, as there are many codecs out there, that all serve a different purpose. It is easy to find good codecs:
JPEG, MPEG2, MPEG4, JPEG2000, JPEG2000 ULL, HEVC, TICO, VP9, AV1, JPEG-XS, TICO-RAW, FlinQ, etc.
... but the difficult task is to find the codec that suits your requirements. In any case, it is always best to contact compression experts like intoPIX, to evaluate your options.
How do you implement video compression in your product?
You can get COTS codecs that are already embedded on a board or you can customize your solution and then choose the platform: they can run on FPGAs, you can embed it into an ASIC, or use it software-based on GPU or CPU.
Why don’t we use video compression all the time?
In fact, compression is used more often than one would think.
This is especially the case in end-user applications. However, much of professional gear still relies on uncompressed video due to the fear of quality loss and latency – which is unjustified considering today’s availability of ultra-low latency, lossless quality video codecs. But eventually, in the verge of HDR, 4K, and 8K video, even broadcasters who had solely dealt with uncompressed video for ages, are now starting to adapt lightweight compression technologies such as TICO or TICO-XS. Thanks to its line-based latency and lossless quality it behaves just like uncompressed, but only requires (at minimum) a 6th of the bandwidth.
So you’re still using uncompressed video? Go ahead and contact us to find out how compression can simplify your video transport and lead to great cost savings.