Search papers, labs, and topics across Lattice.
WanToFight is a generative game engine designed to simulate real-time, two-player interactions in The King of Fighters '97, addressing the limitations of existing engines that focus on single-player or non-real-time scenarios. The system integrates a streaming autoregressive generator, a Player Association module for character identity binding, and a keyboard injection module, achieving 30FPS performance on a single high-end GPU. This work represents a significant advancement in generative game engines by successfully combining multi-player control, real-time inference, and complex interactions in a single framework.
Real-time multiplayer combat in generative game engines is now a reality, thanks to WanToFight's innovative architecture.
We present WanToFight, a generative game engine that simulates real-time, two-player The King of Fighters '97 (KOF~'97) gameplay from keyboard input. Prior generative game engines target either single-player first-person settings or non-real-time cooperative scenarios; multi-player control, real-time inference, complex physical interaction, and adversarial gameplay have not been jointly addressed. WanToFight closes this gap with three components built on the Wan-1.3B video diffusion transformer: a streaming autoregressive generator with block-causal attention and a rolling KV cache; a visually grounded Player Association module that binds each player's keyboard signal to a character identity; and a gated, locally causal keyboard injection module trained with a single-player-to-full-gameplay curriculum. A four-step DMD-distilled student paired with a pruned VAE decoder sustains 30FPS at 512x384 on a single NVIDIA RTX 5090 over the duration of a complete match. To our knowledge, WanToFight is the first generative game engine to combine multi-player control, real-time inference, complex physical interaction, and adversarial gameplay in one system.
