What is the primary difference between Waypoint-1.5 and its predecessor, Waypoint-1?
Waypoint-1.5 significantly improves visual fidelity and dramatically expands hardware accessibility by introducing dual model tiers (720p and 360p). It was also trained on nearly 100 times more data, leading to more coherent environments and consistent motion, and incorporates more efficient video modeling techniques for real-time performance.
How does Waypoint-1.5 address the 'immersion gap' in generative world models?
The 'immersion gap' refers to the difference between watching a scene and feeling inside it. Waypoint-1.5 addresses this by focusing on low latency, responsiveness, and controllability, ensuring environments react instantly to player input, evolve, and remain coherent as users explore them, all while running locally on consumer hardware.
What are the hardware requirements for running Waypoint-1.5 locally?
Waypoint-1.5 is designed to run on a wide range of consumer hardware. The 720p model targets higher-performance systems, while the 360p model is optimized for a broader variety of gaming PCs, including modern NVIDIA RTX GPUs. Support for Apple Silicon Macs is also planned for the future.
Can users interact with the generated worlds, or is it primarily for viewing?
Waypoint-1.5 is explicitly designed for interactive experiences. The goal is to build worlds people can get lost in, allowing users to explore, modify, and play within the generative environments, with the worlds responding instantly to input and evolving coherently.
What are the two main ways to access and use Waypoint-1.5?
Users can experience Waypoint-1.5 in two ways: either by running it locally through the Overworld Biome runtime, which includes a simple EXE installer for quick setup, or by instantly accessing the model in their browser via the Overworld.stream service without any local installation required.