Q: Can you please add a storyboard feature for images

I tried building my first custom model, and one of the issues i noticed with generating images is it turns anime characters into real people. Also background scenes aren't consistent of the models. Please consider adding a storyboard feature.

If you want to know an idea of how that would work, ltx studio and dzine has something like this feature where it keeps it very consistent on the storyboard.

Adding storyboard feature would make it where its not just random pictures and consistent scenes. Very important direction and feature that should be added since its an image plan.

Tony_the_AIPLUSJul 26, 2025
Founder Team
Maxime_Seelab

Maxime_Seelab

Jul 28, 2025

A: Hi! 👋

Thank you so much for sharing this feedback — it’s super valuable.

A storyboard feature for consistent scenes and characters is indeed a great idea. We know how important style and character consistency are, especially for use cases like anime or storytelling.

We’ve forwarded your suggestion to our product team for review. While we can’t promise an immediate implementation, this is definitely on our radar for future improvements.

For your custom model, what you’re seeing is likely because the training images ( maybe photorealistic ones have been inserted ? ), which can cause anime characters to be interpreted as real people.

To improve this:
• Try using only high‑quality, clearly anime‑style images for training (no mixed styles).
• Add style keywords like “anime style,” “manga illustration,” or similar to your prompts.
• Consider increasing the number of training images focused specifically on the character’s anime features for better consistency.

Thanks again for taking the time to explain how this would help your workflow — feedback like this really helps us shape the roadmap!

If you are a paying user you can directly talk to us on the plaftorm, then we could check what's going on. ( it could be multiple cause: dataset, decorators, and we need the model name... )

Best regards,
The Seelab Team

Share
Helpful?
Log in to join the conversation