Sketch-based knowledge-enhanced diffusion method for generating diverse and plausible 3D scenes.
Frankenstein introduces a tri-plane diffusion-based framework for generating semantic-compositional 3D scenes in a single pass.
Depth inpainting is crucial for generating geometrically consistent 3D scenes from a single image or text prompt.
SceneCraft is a novel method for generating high-quality 3D indoor scenes by combining user-specified textual descriptions and spatial layouts, overcoming limitations of previous approaches by supporting complex multi-room scenes and free camera trajectories.
SceneCraft는 사용자가 지정한 레이아웃과 텍스트 설명을 기반으로 사실적이고 복잡한 3D 실내 환경을 생성하는 혁신적인 방법을 제시합니다.
MagicDrive3D is a novel framework that leverages easily accessible driving datasets to generate highly realistic and controllable 3D street scenes, enabling diverse applications like autonomous driving simulation and enhancing 3D perception tasks.
SceneDreamer360은 텍스트에서 사실적이고 일관된 3D 장면을 생성하기 위해 파노라마 이미지 생성과 3D 가우시안 스플래팅을 결합한 새로운 텍스트 기반 3D 장면 생성 프레임워크입니다.
L3DG is a novel method for generating 3D scenes by representing them as 3D Gaussians and using a latent diffusion model trained on a compressed latent space learned by a VQ-VAE with a sparse convolutional architecture.