Why You Can't Just Paste a Video

You can drop a photo into a document, an email, a slide. Drag it, shrink it, put it in a corner. Nobody thinks about it.

Try that with a video and you see the tools erroring out or just doing nothing.

A document is a page. A photo can sit on it because a photo is finished . Open the file, the pixels are right there.

A video can't do that. It needs machinery running just to produce anything, even one still image, even the thumbnail you see before you press play. There is no frame sitting in that file ready to be shown. It has to be made on demand.

So when you paste a video into a document, you're not placing an object on a page. You're handing the page a box of Ikea furniture with no tools and no instructions.

.mp4, .mov, .avi, .mkv, .webm — eh, what?

Those are all videos. And very far apart in how they work.

And it's worse. Two files that both say .mp4 can be built completely differently inside: different compression, different frame rates, different audio.

So before any app can show you a video, it has to figure out what it's actually holding. Photos skip that step. That is a whole lot of processing just for basic analysis.

Below I'll explain how a video actually gets played. Some formats are more forgiving about it, but they're typically larger. Tradeoffs. Nothing's perfect.

It's also big

Ten minutes on your phone can be 5 GB . That's a huge amount of data to just get from here to there, to crunch through.

Every tool that touches it has to move all of that around before anything appears on screen. Still not a millisecond operation, especially not sending it over the internet.

Most frames aren't pictures

A video at thirty frames per second isn't thirty photos. That would make the files multiples larger, unusable. So it saves one real frame, then for the next frames it only saves what moved .

Your kid runs across the playground. The slide behind him doesn't move, so the slide isn't saved again. The frame just contains him, plus a note: "everything else, find it back there" .

Which means when you ask for a single image out of a video, a thumbnail, a preview, the software can't reach in and take one. It has to find the last complete frame, then replay every change since then to reconstruct the picture you asked for. It needs the machinery and calculations just for grabbing that picture.

A photo hands you its pixels. A video hands you a scattered mess and expects complex software to make sense of it.

Change anything, and it's a brand new video file

Trim two seconds off the front and it gets even messier, to the point where nothing can make sense of it at all. Frames are pointing at other frames that no longer exist. The coordinates are simply wrong. All of it has to be recalculated.

Same for resizing. Same for changing format. A photo you resize gets redrawn. A video you resize gets remade.

And yet WhatsApp does it fine, just paste the video

Because those apps aren't pages. They're built for video, and they're much bigger and heavier for it. And they still don't handle it seamlessly. Paste anything slightly large and you'll watch the loading circle for a while. Longer still when you hit Send.

That's why it feels like nothing there and impossible in a document. WhatsApp comes with a player and a transcoder, software built to work with data that runs in real time instead of what technicians call "a state" .

How we do it?

We work with full video platforms in LINK-V . Videos are very cool technology and very difficult to handle at scale. To figure out how to store them, how many versions of one video. How to play them to users without lagging. We've managed to build tools thanks to which we handle videos easier now. But even when we don't work to hard on video processing anymore, servers definitely do.