Abstract
Digital map interactions such as zooming in/out or panning can be hampered by the latency for fetching granular (for zooming in), wide field-of-view (for zooming out), or adjacent (for panning) data before the selected region of the digital map can be rendered. This is a suboptimal user experience. This disclosure describes techniques that leverage generative interpolation to provide visually smooth transitions between zoom levels. When a user of a digital map app changes the region being viewed, a generative artificial intelligence (AI) model is applied to perform outpainting or inpainting on the current view to create a series of interpolating images that progress towards the target view. The interpolated images are displayed in succession on the user interface screen as an initial updated rendering. When spatial data corresponding to the target region becomes available, the model is conditioned on the target view such that it interpolates towards it. The transition is smooth and ends at a final rendering of the target region of the map based on actual spatial data.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Shin, D, "Smooth Continuous Zoom In/Out via Generative Map Interpolation", Technical Disclosure Commons, (July 31, 2024)
https://www.tdcommons.org/dpubs_series/7251