top of page
SYNTEHTIC LONDON

Scale of Production # 2: Architectural Form
This project delivered the proverbial magic trick before Dall-E or mid-journey made their advent into our lives. Through the employ of machine learning models, we explored highly detailed hyper-realistic synthetic world-building. GPU-based real-time raytracing of then young Unreal Engine channeled to produce detailed digital assets, predictive synthetic satellite maps. Integrated assets derived from photogrammetry and laser scans were interpolated to elude realism.
Advanced lighting
Texture mapping
HDRI maps
Built with a keen understanding of LOD
unsupervised deep learning
Â
DCGAN
synthetic panorama [london]
Â
supervised machine learning overlay
Â
vgg-19 (style transfer)
style weight study
Â


bottom of page