Lesson 8: AI video is cheap, until it isn't

15 years. 15 lessons.
By Matt Cooper

Last week's post about AI and taste got me thinking about something else.

The cost.

AI can generate a five-second video in minutes.

Brilliant.

Now make another 20 shots. Keep the characters consistent, match the lighting, get the camera movements right and make it all work together seamlessly.

Not brilliant...far from it.

I've been speaking with other videographers, and the same frustrations keep cropping up.

Characters changing between shots. Backgrounds doing their own thing. Endless prompting, regenerating and figuring out clever little hacks to get something that should be straightforward.

And every regeneration costs money. Not to mention the hours spent wrestling it into shape.

There's a perception that AI video production is cheap and easy.

Sometimes it absolutely is and you get really lucky.

But I’m finding the reality is that you’ll likely spend a lot more time and money trying to generate a sequence that might have been quicker, cheaper and better to actually film in the first place.

Then there are the bigger questions.

What happens to pricing when AI providers need to recover their enormous running costs?

AI newsfeeds keep reporting expected leaps in AI video costs are around the corner.

And what's the environmental impact of all this? Is that a cost we can honestly accept?!

It’s all a bit depressing when you look at the big picture.

Yes, AI has made generating video incredibly fast.

But, it hasn't necessarily made finishing video incredibly fast.

And ultimately, at what cost?

Lesson 8 done. Only 7 more opinions nobody asked for coming your way.