Backend
Map2Video: Guiding Real-World-Grounded AI Video Generation
arXiv cs.HC0 views
arXiv:2512.17883v2 Announce Type: replace Abstract: AI video generation lowers barriers to video creation, but existing tools provide limited support for grounding content in real-world locations, restricting their use in applications such as film previsualization and scenario generation. In practice, users rely on verbose textual descriptions or reference images using on-site photos or street view imagery (SVI), which are cumbersome and unreliable across shots. Our formative study with five filmmakers revealed challenges in shot composition, character motion, and camera control when using SVI with an image-to-video tool. We present Map2Video, a system for guiding SVI-grounded AI video generation through map-based character trajectory annotation and direct manipulation of cameras and character masks defined in geodetic coordinates and projected onto panoramic screen coordinates. In an evaluation with twelve filmmakers, Map2Video provided stronger controllability for both scene replication and open-ended creative exploration, reduced cognitive effort, and improved spatial consistency across multiple shots compared to an image-to-video baseline.
Source:https://arxiv.org/abs/2512.17883PreviousCoMPASS: A Multi-Turn Benchmark for Measuring LLM Reinforcement of Parasocial BehaviorNextSycamore: Grounded Synthetic Evaluation for Visualization Systems
Related posts
Splunk's Empty Cluster Secret Turns a Standalone Default into an Authentication Bypass
0dev.to
Pragmatic zero-trust: what to ship in the first 90 days
0dev.to
The EKS cost playbook: 7 levers we pull on every audit
0dev.to
Building a hearing test in the browser with the Web Audio API (and why we refuse to call it an audiogram)
0dev.to
How we score French speaking with AI: an honest look at limits
0dev.to
What a Four-Letter Word Ladder Taught Me About Planning Under Constraints
0dev.to