Asking the CEO #1: Without video compression, internet streaming would not be possible

Do you know that one letter in the digital world can take up to 32 positions? Now, consider the vastness of a live-streamed video. In a world where video accounts for 80 percent of data transmission, effective compression techniques are invaluable. The founders of Comprimato pioneered GPU accelerated design of video compression algorithms while working at the Laboratory of Advanced Network Technologies at the Faculty of IT at Masaryk University. Their research gave rise to the company Comprimato, and its software is used today by the broadcasters, film productions, but also in medical or satellite imagery applications.

Interview with Jiri Matela, Co-founder of Comprimato:

In today’s world, given how much data humanity is creating, which is run all over the world, compression is probably very important. What would it look like if we didn’t have compression software?

Jiri Matela: It would be difficult. Video streaming would not be possible. Over 80% of internet traffic is compressed video. Storing digital data is brilliantly simple – we need two characters: zero and number one – and we can encode anything into them. Then again, it is very limiting. When we compare it to the alphabet and take the letter A, it is just one character, but we write this letter using binary representation, we would need anything from eight to thirty-two zeros and ones.

In the case of image and video, the problem is several orders of magnitude higher because there is an incredible amount of data needed to represent one image.

And right here is the motivation for compression, how to make the recording and transmission of this data more efficient and denser. At the same time, there is a large amount of entropy: the 32 letters of the alphabet allow us many more combinations than ones and zeroes, plus, letters are also much easier to compress.

Image compression is also leveraging a psycho-visual tricks when the human eye cannot see all the bits represented by pixels, so there are many redundant elements. The next level is video – if we consider that we have fifty frames a second in the video, close to all the frames are almost the same. Motion is created by playing individual frames one after the other. And video compression uses just that minimal changes between frames, so much data is discarded by the compression algorithm and then reconstructed again based on similarity.

What was the path to fulfilling this dream? What does Comprimato do?

We do video compression, streaming software, and software as a service. When you think of Netflix, you probably think of movies, but there is a rather complex, sophisticated back-end behind it, a video processing infrastructure, and that’s what Comprimato does. Specifically, we specialize in live broadcast – for sports events, large video conferences, or in the USA for church services; for example, these are very common there now.

What makes Comprimato unique?

We’ve developed incredibly fast software that serves as a flexible alternative to specialized hardware designed for specific tasks. With equivalent quality and parameters, our software not only reduces costs but also enhances the density, allowing customers to process more video streams simultaneously.

So, sports fans are seeing the results of your work on their televisions?

Yes, right. I don’t know if they’re thinking about us while watching football, though 🙂

The “Asking the CEO” series will continue soon.