Real-time interactive game world models. Press a key, and the model draws the next frame of the game world.
12 projects · 10 papers · 10 open-source codebases · 7 models · 6 datasets
YeyingLab_GameWorldModels_web.mp4
Yeying Lab is led by Yeying Jin, Staff Researcher at Tencent and Adjunct Faculty at NUS ECE. We build video world models that players can steer, edit and play in real time: agents that run the game while a video model renders it, action and language control that stays spatially consistent, and NPCs that read the game state and react to players.
These are technical explorations, not commercial games.
| Project | In one line | Links |
|---|---|---|
| MaLiang | Paint a map, and the world follows. | Coming soon |
| Wizard Manipulate Any World in Real Time via Agentic Control |
Manipulates any world in real time. | arXiv · Project · Code · 🤗 Model · 🤗 Data |
| UnboundRun | Renders endless parkour at 40 fps. | Coming soon |
| GameCrafter An Agentic Framework for Multi-Gameplay Game World Generation |
Builds worlds from a prompt. | Paper · Project · Code |
| GameDirector Decoupling Gameplay Logic from Rendering for Player-Configurable Game World Models |
Splits logic from rendering. | arXiv · Project · Code · 🤗 Model · 🤗 Data |
| Project | In one line | Links |
|---|---|---|
| SCOPE Simulating Cross-game Operations in Playable Environments for FPS World Models |
Learns first-person action control. | arXiv · Project · Code · 🤗 Model · 🤗 Data |
| ActionRoPE Action as Coordinates for Spatially Consistent Control in Planar-Camera Game World Models |
Keeps revisited scenes stable. | arXiv · Project · Code · 🤗 Model |
| H3-World Turning Language Understanding into World Control |
Turns language into world control. | arXiv · Project · Code |
| Incantation Natural Language as the Action Interface for Multi-Entity Video World Models |
Commands many characters in plain language. | arXiv · Project · Code · Play · 🤗 Model · 🤗 Data |
| Project | In one line | Links |
|---|---|---|
| WorldMind Decoupled Game World Model for State-Aware NPC Behavior |
Makes NPCs reason, then act. | arXiv · Project · Code |
| ReactiveGWM Flexible Control and NPC Reactivity in Game World Models |
Lets NPCs react to players. | arXiv · Project · Code · 🤗 Model · 🤗 Data |
| StatePlay State-Aware Game World Models for Mechanics-Consistent Generation |
Models frames and game states. | arXiv · Project · Code · 🤗 Model · 🤗 Data |
SCOPE and Incantation are joint projects with external collaborators.
Yeying Jin's first-author work on restoring images taken at night and in bad weather, 2021–2024.
| Project | In one line | Links |
|---|---|---|
| Raindrop Clarity A Dual-Focused Dataset for Day and Night Raindrop Removal · ECCV 2024 |
Benchmarks raindrop removal by day and night. | arXiv · Code · Challenge |
| DeS3 Adaptive Attention-driven Self and Soft Shadow Removal using ViT Similarity · AAAI 2024 |
Removes self and soft shadows with diffusion. | arXiv · Code |
| Nighttime Dehaze Enhancing Visibility in Nighttime Haze Images Using Guided APSF and Gradient Adaptive Convolution · ACM MM 2023 |
Clears haze and glow at night. | arXiv · Code |
| S-Aware Network Estimating Reflectance Layer from A Single Image: Integrating Reflectance Guidance and Shadow/Specular Aware Learning · AAAI 2023 |
Recovers reflectance free of shadows and highlights. | arXiv · Code |
| Night Enhancement Unsupervised Night Image Enhancement: When Layer Decomposition Meets Light-Effects Suppression · ECCV 2022 |
Brightens night images and suppresses glare. | arXiv · Code · Demo |
| Fog Removal Structure Representation Network and Uncertainty Feedback Learning for Dense Non-Uniform Fog Removal · ACCV 2022 |
Removes dense, uneven fog. | arXiv · Code |
| DC-ShadowNet Single-Image Hard and Soft Shadow Removal Using Unsupervised Domain-Classifier Guided Network · ICCV 2021 |
Removes hard and soft shadows without paired data. | arXiv · Code |
We're hiring 6-month research interns in world models and coding agents, in Singapore and through Tencent campus recruiting in China. Details