Activision drops Call of Duty: Warzone map data for AI training
Activision just released an open-source dataset pulled from Call of Duty: Warzone’s Caldera map on GitHub. Formatted in OpenUSD, it includes near-complete map geometry and anonymized player movement traces to support research. It marks one of the largest production-validated gaming datasets ever made public, giving generative AI models rich 3D environments to learn from.
Activision has open-sourced the Caldera map dataset from Call of Duty: Warzone on GitHub, giving AI researchers and 3D developers production-grade assets to train world-generation models.
What is inside the dataset
The release is packaged in OpenUSD, the open 3D format backed by Apple and Pixar. It features the near-complete geometry of the massive Caldera map alongside anonymized time-series samples tracking how players moved across the terrain.
According to Activision software fellow Michael Vance, it stands as one of the largest production-validated gaming datasets ever made public in terms of geometry complexity and instance counts. Researchers and developers can grab the files directly from GitHub right now.
Why it matters
Getting access to AAA production assets has always been a massive bottleneck for AI researchers working on 3D synthesis and spatial intelligence. Most academic labs simply do not have the resources to build a sprawling island map from scratch.
We think Activision made a smart move by releasing standard OpenUSD data rather than keeping it locked behind proprietary engine walls. Even with ongoing industry anxiety around generative AI, real-world spatial and behavioral data at this scale will be pure fuel for the next wave of virtual world generators.
Sources (1)
- theverge.com(opens in a new tab)Originaltheverge.com