Video
Generating short shots from text, images and reference material, then upscaling them to production quality.
Mundi Lab · Research
Building our own models for video, animation and 3D.
Mundi Lab is the studio’s research team. We are developing our own generation models for video, animation and 3D, so our products are not limited to what off-the-shelf models can do.
Why build our own
Generic models make generic worlds.
A children’s game needs the same characters, the same art direction and the same warmth from one scene to the next. That is hard to get from general-purpose models, so we are building our own.
Control over style
Consistent characters and art direction across every image, shot and asset.
Built for production
Outputs designed to fit real pipelines, from animated shots to meshes for real-time engines.
Our own infrastructure
We already run generation models on the studio’s own GPU infrastructure, under our control.
Research areas
Three fields, one goal: living worlds.
Animation
Bringing characters and scenes to life with motion that stays true to their design.
3D
Stylized meshes and textures, ready for real-time engines like Unreal Engine.
From lab to product
Research that ships.
Lab work doesn’t stay in the lab. Our models will power Visual Run, the workspace our team already uses to produce Mundopia.
Research
Mundi Lab
Video, animation and 3D models
Workspace
Visual Run
Where our team uses them
Product
Mundopia
What children play
Where we are
A step-by-step path to our own models.
- Today Our own generation infrastructure Video generation and upscaling models run on our own GPUs and are already used inside Visual Run.
- In progress Our own models We are developing in-house models for video, animation and 3D, shaped by Mundopia’s art direction.
- Next In production Ship them in Visual Run and use them to produce Mundopia.
Research partnerships.
Researchers, labs and partners interested in generative video, animation or 3D: let’s talk.