World Labs Atlas is a 3D Gaussian splatting tool for creating navigable locations and reference back plates from text, vision, video, and other multimodal inputs.
Why it matters
Investor A, Investor B
API Docs
API Available
overview
World Labs Atlas is an AI-powered 3D scene creation tool that enables developers, game designers, and architects to create navigable locations and reference back plates. It uses Gaussian splatting and accepts text, vision, video, and multimodal inputs. The product supports interactive editing and expanding or combining worlds.
features
World Labs Atlas centers on generating and manipulating 3D Gaussian-splat scenes. Its stated inputs and editing functions support scene creation and iteration; API usage is documented for beta access.
use cases
The listed audiences are developers, game designers, and architects. The product’s stated applications include game development, virtual reality, and architectural visualization.
how to use
Use the web product to create and edit worlds, or consult the API documentation to submit tasks programmatically. The beta API uses WLT-Api-Key authentication, and documented task and concurrency limits are still to be announced.
pricing
The supplied pricing information lists a free Basic tier with 100 worlds per month, a Pro subscription at $29 per month, and usage pricing of $0.05 per world creation. The relationship between plan allowances and the per-world charge is not specified, so confirm current billing terms with World Labs before estimating total cost.
Enjoying this? Get one like it in your inbox each morning.
one email a day · unsubscribe in two clicks · no third-party tracking
Similar Tools
World Labs Atlas is described as generating worlds from multimodal inputs and supporting world expansion and combination. The supplied comparisons characterize Nerfstudio, Scaniverse, and Polycam primarily as tools for locally trained or captured and reconstructed scenes; their workflows differ from Atlas’s stated generative world creation.
Provides an end-to-end, modular open-source pipeline for training and visualizing splats (via Splatfacto) completely on local hardware without cloud vendor lock-in.
You trade Atlas's cloud processing and automated generative scene extensions for full local control and zero hosting fees. It requires a modern Nvidia GPU and comfort working in the terminal or Python.
Allows complete, real-time on-device Gaussian splat capture and processing directly from a smartphone with export capabilities to SPZ and PLY.
Scaniverse relies primarily on physical video and LiDAR capture from a mobile phone or 360 camera rather than generative image-to-world creation. It excels at real-world spaces but cannot hallucinate uncaptured viewpoints like an AI foundation model.
Provides a turnkey cloud-processing platform that accepts drone footage, video, and image sets to output Gaussian splats and textured meshes with built-in measurement tools.
While Polycam offers ready-to-share web viewers and simple multi-format exports, it reconstructs physical footage rather than generating persistent 3D worlds from sparse prompt conditioning. Gaussian splat exports also require a paid tier.
More on Stork
Other tools in this category, matched by shared tags
