Research map / Video / Video generation
GameFactory: Creating New Games with Generative Interactive Videos
Jiwen Yu, Yiran Qin, Xintao Wang, Pengfei Wan, Di Zhang, Xihui Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use cross-attention instead of concatenation for keyboard actions (ablation) | action control module · Only-Key subset | Cam: 0.0853 -> 0.0439 | Improved |
| use cross-attention instead of concatenation for keyboard actions (ablation) | action control module · Only-Key subset | Flow: 22.37 -> 7.79 | Improved |
| use concatenation instead of cross-attention for mouse movement (ablation) | action control module · Mouse-Small subset | Cam: 0.0798 -> 0.0685 | Improved |
| use concatenation instead of cross-attention for mouse movement (ablation) | action control module · Mouse-Large subset | Flow: 325.18 -> 249.54 | Improved |
| use concatenation instead of cross-attention for mouse movement (ablation) | action control module · Mouse-Large subset | Cam: 0.1362 -> 0.1021 | Improved |
| change key/mouse fusion mechanism (ablation) | action control module · Only-Key / Mouse CLIP, FID, FVD | CLIP/FID/FVD | No clear effect |
| add multi-phase decoupled training with domain adapter instead of one-phase (ablation) | GameFactory · open-domain scene generalization | Cam: 0.1134 -> 0.0997 | Improved |
| add multi-phase decoupled training instead of one-phase (ablation) | GameFactory · open-domain scene generalization | Flow: 76.02 -> 54.13 | Improved |
| add multi-phase decoupled training instead of one-phase (ablation) | GameFactory · open-domain scene generalization | FID: 167.79 -> 121.18 | Improved |
| add multi-phase decoupled training instead of one-phase (ablation) | GameFactory · open-domain scene generalization | Dom: 0.7345 -> 0.7565 | Improved |
| train on GF-Minecraft instead of VPT (ablation) | action control model · in-domain evaluation | FID: 156.69 -> 125.85 | Improved |
| train on GF-Minecraft instead of VPT (ablation) | action control model · in-domain evaluation | Flow: 107.67 -> 43.48 | Improved |
Closest papers by meaning
- GameGen-X: Interactive Open-world Game Video Generation (ICLR 2025 · similarity 0.92)
- Diffusion Models Are Real-Time Game Engines (ICLR 2025 · similarity 0.88)
- From Zero to Hero: Training-Free Custom Concept Spawning in World Models (NeurIPS 2026 · similarity 0.88)
- Genie: Generative Interactive Environments (ICML 2024 · similarity 0.87)
- Moonshine: Distilling Game Content Generators into Steerable Generative Models (AAAI 2025 · similarity 0.86)
- GROOT: Learning to Follow Instructions by Watching Gameplay Videos (ICLR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
