SceneGen Generates 3D Scenes from a Single Image in One Feedforward Pass

tom_doerr · x · 2026-08-10

SceneGen, accepted to 3DV 2026, introduces a novel architecture capable of generating complete 3D scenes from a single image in just one feedforward pass.

The project's training code, inference scripts, data processing toolkits, and pretrained models have been fully open-sourced on GitHub. Developers can deploy and reproduce the results locally with supported hardware (e.g., NVIDIA GPUs).

Original post →

More from Multimodal

Multimodal channel →