We've been quietly working behind the scenes, testing different approaches, and listening to your feedback.
​
You've been telling us the same thing for months: "I can get one character to look great, but the moment I try to add a second character to the scene, everything falls apart."
​
Most AI tools (including our first version) struggle with this. But we refused to accept "good enough."

So we went back to the drawing board. Spent months testing different models, analyzing thousands of failed attempts, and working directly with beta users who were willing to try our experimental approaches.
​
The result? We launched Multiple Character Consistency V2 - a brand new model that's one of the best-in-class out there right now, specifically optimized for cartoon generation.

But here's what really made the difference: we also figured out the exact way to structure prompts that works perfectly with this new model. When you combine the v2 model with the 3-step prompting method I'm about to show you, you'll get better results than any other AI tool out there for creating scenes with two characters.

Our beta users who tested this multiple character consistency workflow are telling us it's "one of the best multiple character consistency tools they've used in a long time" and that "it's very clever in understanding natural prompts, and the images are super high quality and consistent."

Currently, this works for two characters in a single scene - we're working on supporting more than two characters soon.

Now, the key to getting amazing results with this v2 model is how you structure your prompts.

The 3-Step Prompting Method We Developed

We discovered that AI needs structure, not complexity. Most people try to cram everything into one long sentence, and that's where things go wrong.

But first, you need to start with the right character references. Upload high-quality front-view full-body images of each of your two cartoon characters. This gives the AI the clearest understanding of who each character is before you start positioning them in scenes.

Don't have those character images yet? No problem - you can create them easily using our Character Turbo V2 in our app. It's designed specifically for generating clean, front-facing full-body character references.

Then, break down your prompting like this:

Step 1 - Define your character tags first

Always start by telling the AI model who's who:

• @character1 is a boy and @character2 is a girl
• @character1 is a dog and @character2 is a child

Step 2 - Set positions before anything else

Before you add any actions, tell the AI exactly where each character should be:

• @character1 is standing in the front, @character2 is behind and to the left of @character1
• @character1 is sitting next to @character2

Step 3 - Then add what they're doing

Only now do you layer in the actions and interactions:

• @character1 is looking at @character2 with a surprised look
• @character1 is giving a present to @character2
• @character1 and @character2 are talking and smiling

Here's what a complete prompt looks like:

@character1 is a boy and @character2 is a girl. @character1 is standing to the left and @character2 is standing to the right. @character1 is waving hello to @character2

or

@character1 is a boy and @character2 is a girl. @character1 and @character2 are sitting together. They are laughing and sharing ice cream

The key? Keep it simple. Don't try to be clever or creative with your prompting.
Just follow the sequence: define tags → set positions → add actions.

See It In Action

We put together a quick 1-minute video showing this exact workflow. it's pretty satisfying to see how consistent the characters stay across different scenes.

​

Your Turn

If you've been struggling with getting two characters to stay consistent across different scenes, give this approach a try. Most people see immediate improvement on their first attempt.

The workflow is available in Multiple Character V2 right now. Just remember: tags first, positions second, actions last.

And here's something we learned - you can always regenerate or experiment with different prompts. That's actually how you get the best results. Don't be afraid to try variations.

What are you working on? Reply and let us know - we love seeing what you create.

Community Win: UCLA Guest Speaking

We had the coolest experience that we just had to share with you.

Anika Duncan, who teaches "The Business of Entertainment, Media & Sports" at UCLA, invited Sachin, our co-founder, to be a guest speaker this summer. Sachin taught the students about developing AI Cartoon Storyboards with Consistent Characters AI.
​
She shared her thoughts and appreciation on LinkedIn, Read her full post here.​

If you're an educator (we know many of you are), this experience reminded us how much students love seeing real AI tools in action. Theory is great, but showing them something they can actually use? That's when the lightbulbs go off.
​
Are you using our tool in educational settings? We'd love to hear about it.

Student Spotlight: Erica Weinstein created an entire storyboard for a romantic comedy called "The Not-So-Secret Lives of Teachers" using Consistent Character AI. The 8-scene presentation shows consistent characters throughout - exactly what we taught in the workshop.

Talk soon,
Sachin and Diana
​

P.S. We've been getting amazing submissions from the community. If you create something cool using this workflow, share it with us at support@consistentcharacter.ai - we feature the best ones in our monthly showcase that goes out to 35,000+ readers.