Name: We Tested Google Veo and Runway to Create This AI Film. It Was Wild. | WSJ
Uploaded: 2025-05-29T03:16:54.000Z
Duration: 7 min 1 s
Description: Thousands of YouTube videos with English-Chinese subtitles! Now you can learn to understand native speakers, expand your vocabulary, and improve your pronunciation...

Adverbs of degree

I am Dr. Chip Mutterborg, a totally real robotics expert.

And I'm the real me, controlling that AI character.

The film you're about to see, including Chip and all the other visuals, were created with AI video tools, including Google VO3.

Most of the audio is also AI, except for my narrations.

So stick around to the end to see how we did it.

I got the OptiMax 5000 to make my life better.

I'm working out harder than I ever thought I could.

Remember, Jane Fonda always said, go for the burn.

I'm with the lead engineer on the OptiMax 5000 robots.

We design them to optimize workplace efficiency.

Here, some deep research for your presentation.

I know robots really dock creative with that.

Though now that I think about it, he's not usually the nicest.

And there's really only one place I go to get a break from him.

My privacy policy prevents bathroom entry.

My ultra-sensitive microphones indicate you are not engaged in elimination activities.

And that's when I realized I had to do something.

And those were actually the best of what we got.

But we wanted to push the limits of AI video.

We wanted to make a film with characters and humor and meaning.

Gerard is a long-time video and audio producer.

To get the consistency right, he generated approximately a thousand clips to make what you just watched.

And after trying lots of AI video tools, we found Google Video.

Yes, in VO or Runway, we could have typed in a woman and a robot working out.

And even though she looks exactly like me, with that simple technique, we wouldn't have been able to make those characters consistent in every scene.

And by the way, these clips and the sounds you hear were generated with VO3, Google's latest model that combines visuals and audio.

Our film is actually a mix of shots from both VO2, VO3, and Runway.

Let's dissect this scene to see how we did it.

First, we had to create and gather our characters.

We used Midjourney to design our star, Optimax 5000.

Then we took these images of the characters and the background and put them into Runway's references tool.

And wrote a prompt describing the scene we wanted.

Then we fed that image into Google VO and wrote a new prompt to create the motion.

And then we picked the best version that AI generated.

For some shots that didn't have characters, like in our suspense scene at the end, we used text-to-video prompts in VO3.

And if you're wondering how we made this handsome fella, we used an image generator to create our ideal mad scientist.

Then I did my best impression of how I thought he'd gesture and sound like on camera.

We designed them to optimize workplace efficiency.

And we uploaded the image and the video of me into Runway ReStyle.

Once we had the clips, we brought them into Adobe Premiere for editing and sound.

Matt's voice and Chit's voice were generated in Eleven Labs.

The AI audio tool lets you describe the voices you want or even clone your own.

But I recorded my own character because Eleven Labs couldn't quite get my inflection right.

And that song you hear at the end of the film?

And yes, the script was written by us humans, not AI.

We estimate the total cost would have been around $1,000 for Google and Runway's tools.

We paid for some of it and the companies gave us special access to the rest.

So why did we put ourselves through all of this?

But if we were making this in real life, we would have needed a crew and a much bigger budget.

But they're nothing without human input, creativity, and original ideas.