How I make cartoon animation with AI

I've always been interested in animation, and two of my kids are hoping to pursue traditional animation and filmmaking in school.

My kids are pretty anti-AI, but I still think it's important to play around with these tools and see where they're headed. Yesterday, I tried Sora 2 to see how it does with animation.

Here's my first video, a cartoon theme for the ​Mostly Technical​ podcast:

I also made one for the ​No Plans to Merge​ podcast:

As you can see, the animation still isn't... great. There are lots of weird artifacts, and it actually took me a long time to prompt these (most shots took multiple attempts). Plus, I still needed to do a lot of my own editing afterwards in a non-linear editor.

But... this was fun as hell. I stayed up past midnight last night making these. It was fun to think up scenes and gags based on inside jokes from my favorite podcasts.

Making animation with Sora 2

• Generate character assets and backgrounds in ChatGPT

  • Example of how I generate these assets in ChatGPT

• Assemble those assets into a single "storyboard" image in Sketch (example)

• Animate (using these assets) ​in Sora 2​ (example)

• Make the theme song music ​in Suno​ (example)

• Put it together, which I edit quite heavily, ​in ScreenFlow

  • Download my example project

Sora 2 prompts for creating animation

For prompting, I'm using a technique I learned from Framer.

For example, for this video, I used this prompt:

The "donor shot" refers to this storyboard graphic (which I uploaded at the same time):

How do I feel about all this?

I'll be honest: I'm conflicted.

On the one hand, the power of these tools is undeniable. I can prototype an idea in minutes that would have taken days (or been impossible) before. For a podcast intro or a quick animation concept, it's remarkable.

But then I think about my kids.

Two of them want to pursue animation. They're spending hours studying fundamentals, learning to draw, and understanding movement and timing. That kind of dedication takes discipline. It requires believing that all those hours of practice will eventually pay off.

And here I am, generating something in 30 seconds with zero skill. (Although, the quality is still not great)

I keep wondering: how do we convince the next generation to put in the 10,000 hours when AI can shortcut the whole thing? What happens when "good enough in seconds" competes with "excellent after years of work"?

I don't have a clean answer.

For now, I'm somewhere in the middle. I want my kids to explore their passions and develop the discipline necessary to achieve mastery. But I also think it's shortsighted to pretend these tools don't exist.

Maybe that's the balance: use the tools, but don't let them replace the pursuit of craft.

Cheers,
Justin Jackson

Connect with me on:
🦋 Bluesky
💼 LinkedIn
🐘 Mastodon
🧵 Threads