Single-Image 2.5D Reconstruction Built Inside ChatGPT Work Mode With Depth Anything V2
paw_lean · x · 2026-09-08
A developer used ChatGPT mobile's Work mode with Depth Anything V2 and Three.js to build synchronized virtual RGB and depth views in the browser, achieving single-image 2.5D reconstruction with an assumed scale — demo shown at 2x speed.
More from Multimodal
- Reconstructing a house in Blender from $500 DJI drone telemetry and free map data — doodlestein · 2026-09-08
- Turning old photos into films with Google Flow: a hands-on Veo tutorial with prompt tricks — InformalAnywhere2511 · 2026-09-08
- Building a 3.5-minute music video for an AI cohost from 27 H3-generated clips — VisualRecording4960 · 2026-09-08
- Modular MiniMax H3 workflow enables seamless ultra-long video with segment regeneration — wjc_5 · 2026-09-08
- 4D reconstruction of Amazon Prime Air cargo plane runway overrun at Miami airport — bilawalsidhu · 2026-09-08
- Beeple amplifies viral take: artists don't need studios in the AI era — soleio · 2026-09-08