Super Resolution Satellite Imagery
See the difference 4x resolution makes. Interactive comparison of Sentinel-2 at 10m with AI-upscaled 2.5m imagery over Melbourne, streamed as Cloud Optimized GeoTIFF.
Platform
afterrealism.com/project/super-resolution
Super Resolution
AI-Enhanced Satellite Imagery
Key Results
Resolution from 10m to 2.5m ground sample distance
Streaming on-demand tile loading via HTTP Range requests
Per-layer visibility toggle and opacity controls
Feature Deep Dive
From on-demand COG streaming to AI-powered resolution enhancement, every component is designed for efficient satellite imagery comparison.
COG Tile Streaming
Cloud Optimized GeoTIFF served from R2 with HTTP Range requests for on-demand loading. Only the visible tiles are fetched, keeping bandwidth usage minimal even for large imagery datasets.
Layer Controls
Independent visibility toggle and opacity slider per layer. Compare original Sentinel-2 and super-resolved output side by side, or blend them to spot differences at any opacity level.
Dark Basemap
CartoDB dark tiles optimized for satellite imagery contrast. The dark background makes imagery colors pop and reduces visual noise when comparing resolution differences.
Deep Learning
AI super resolution model produces 4x uplift from Sentinel-2 input. The model enhances spatial detail from 10m to 2.5m ground sample distance while preserving spectral fidelity.
See the difference 4x resolution makes
Compare original Sentinel-2 imagery with AI-enhanced super resolution output. Toggle layers, adjust opacity, and explore Melbourne at 2.5m detail.
View Demo