This was part of an ongoing exploration of how AI video models interpret glass and the way it breaks light and the view behind them. Making this video (in early 2026) I was very impressed with the capabilities. Images were created in Midjourney, video in Veo, music in Suno, manually edited.