A Thousand Compresses: How a Video Slowly Disappears

Video compression is like washing clothes—each time, a little fades. After a thousand rounds, almost nothing is left. Here's how it works.

A Thousand Compresses: How a Video Slowly Disappears

Sometimes, a simple demonstration cuts through layers of technical jargon to reveal the essence of a concept. This happened to me recently with video compression. It's a topic I've explored before, but let’s be honest—talking about codecs, encoding, and decoding can feel a bit abstract.

It’s a world of technical terms, hardware versus software debates, and questions like, “Which codec should you use?” But recently, I came across a brilliant demonstration that brought it all into focus.

The Thousand Upload Experiment

Marques Brownlee did something that caught my attention. He uploaded and downloaded a video with audio to YouTube Shorts and TikTok—not once, not twice, but a thousand times. The result? A fascinating visual representation of compression in action.

Each time the video was processed, there was some lossy quality loss, but it wasn’t noticeable at first. After four or five rounds, you could still recognise the original. But after a thousand cycles? The result was hilariously unrecognisable—almost entirely purple, a glitchy ghost of the original content.

It was both funny and enlightening, like watching the life cycle of your favourite T-shirt. When you wash clothes a few times, the wear and tear are imperceptible. But after years of use, the colours fade, the fabric thins, and the item eventually becomes unwearable.

Video compression follows a similar principle: the lossy compression losses accumulate, and the quality degrades with repeated cycles. The same effect can be seen with JPEG images—each time you re-save, they lose quality due to cumulative compression, eventually leading to noticeable degradation.

👕
Technical Callout: YouTube Shorts primarily uses VP9 and H.264 codecs, while TikTok relies on H.264. Both platforms apply lossy compression to optimise streaming, sacrificing quality for efficiency. With each re-encoding, visible degradation occurs as data is discarded, resulting in a progressive decline in video fidelity.