Kai-Peng Zhang

Static 3D worlds now possess vivid, looping motion, seamlessly transitioning from any viewpoint without explicit motion capture.

Language instructions now directly control real-time, high-fidelity 360° video generation, enabling fluid interactive exploration from a single panorama.

LLM/VLM agents significantly outperform non-agentic methods in preserving unedited parts during iterative 3D asset modification, but current agentic edits are prohibitively slow.

Most video generation models fail to accurately render fine-grained, program-specified world events, with PROWBench revealing significant discrepancies between intended logic and visual output.

KaiNinja, a part-level extension of TRELLIS.2 built on a dual-volume form of its O-Voxel representation, is the first 3D generative model trained on agent-authored part data and finds that whole-object fidelity improves over the same backbone fine-tuned on the same dataset.

AlayaVista is presented, a camera-controllable streaming video world model that decouples panoramic world evolution from perspective observation synthesis, and MUGEN, a large-scale real-world panoramic video dataset containing 1,318 hours of videos at resolutions of at least 4K, together with rich semantic and geometric annotations.

Monolithic 3D scenes with hundreds of heavily occluded objects can be cleanly parsed into individual editable meshes without ever training on multi-object data.

ARISE-RL transforms agent training by enabling robust self-evolution through a novel rubric-mediated co-evolution framework, achieving state-of-the-art performance across diverse tasks.

Game development could revolutionize how we generate high-quality reward signals for spatial world models, moving beyond fuzzy proxies to executable environments.

LLM agents can dramatically improve their long-horizon decision-making by strategically managing memory, with performance varying significantly based on memory structure.

PhysEditWorld reveals that explicit control over physical parameters can transform how game world models interact with their environments, leading to more realistic and manipulable simulations.

Runtime pass rates plummet from 80.4% to 5.7% as project complexity increases, exposing critical architectural flaws in code generation models.

COM-based execution outperforms GUI interactions, achieving state-of-the-art results in professional software manipulation tasks where traditional methods fail.