^{Read time - 3 minutes}

I just got back from Freepik’s Upscale Day in London, here are the 3 biggest shifts I saw.

Takeaway 1: AI films are now realistic enough to compete with real production

At the event, I watched a film called
“The Chronicles of Bone” by Kavan the Kid, you can watch it here.

This isn’t some rough AI video demo.

It felt like a real cinematic production: Story, atmosphere, characters, pacing.

Actually, if you’ve got 15 minutes… you HAVE TO watch it👇

A 15-minute cinematic clip like this would traditionally cost:

This is why the industry gets disrupted.

Takeaway 2: AI ads are about speed, not perfection

One of the speakers, Ikenna Mokwe, really reinforced this.

He told us that after he started creating AI ads, he landed over £100K in client work in just 4 months, working solo!

He also showed one of his viral AI ads, a satirical AI spot about commuting in London called “13 Ways to Annoy People on the Commute.”

It’s funny, slightly absurd… and exactly the kind of content that performs right now.

It actually won Best Spec Ad at the Bionic Awards in London.

This is actually worth watching👇

Takeaway 3: The people in the room are already way ahead.

But here’s the good news: It’s not too late to enter the AI space.

Right now, supply and demand in AI ads and film are completely out of sync.

Which means:

If you spend just a few hours learning this every day…
you’re already ahead of 99% of people.

See my recap reel from the whole London trip right here.

Even as a solo creator, without a team, you can now:

And that gives you a huge advantage in both cost and time!

This community is the best place to learn it all! But if you want to just start generating amazing images and videos, Freepik has all the best tools in one place.

Luma AI just launched a new image model called UNI-1.

And it works very differently from what we’re used to.

Most AI image tools work the same way:

You write a prompt → it generates an image.

But this new model changes that.

It doesn’t just generate… it thinks first.

Uni-1 is one of the first models that combines reasoning + image generation in one system.

Instead of jumping straight to output, it actually:

And then generates the image, see examples for yourself below👇

That said…

UNI-1 is impressive.

But my go-to is still Nano Banana 2.

It’s faster, more consistent, and just works better for what I need personally.

You can see recent examples I posted here, along with tutorials on how I created them.

Which image model is your favorite?