Researchers have introduced WorldCrafter, a video world model designed to let users interactively explore dynamic environments while respecting what has already been observed over long horizons and across changing viewpoints.
The core idea is that the requested camera viewpoint should shape how multi-view evidence is compressed into the video generator's limited token budget. A memory encoder and a pose-conditioned readout module are trained jointly with the video generator, integrating historical observations into a fixed set of target view-specific tokens before denoising — without relying on explicit depth-based correspondences.
By combining this implicit 3D-aware memory with recent temporal context and few-step distillation, WorldCrafter supports streaming scene exploration from a single input image or text prompt.
Experiments across static and dynamic scenes reportedly show substantial gains in long-horizon consistency and camera-control accuracy while preserving visual quality during minute-scale exploration.
The project page and code are available via the authors' public links.