diff --git a/_posts/paper/2024-03-15-GALA3D.md b/_posts/paper/2024-03-15-GALA3D.md index 3c7f944..f59a543 100644 --- a/_posts/paper/2024-03-15-GALA3D.md +++ b/_posts/paper/2024-03-15-GALA3D.md @@ -10,10 +10,10 @@ venue: ICML 2024 year: 2024 # paper year, number month: March # paper month, full name projectPage: None # If has project page, link here, otherwise None -supplemental: https://arxiv.org/abs/2402.07207 +supplemental: https://gala3d.github.io/ data: None # If has data, post data link here, otherwise None code: https://github.com/VDIGPKU/GALA3D # If has data, post code link here, otherwise None -paperLink: https://arxiv.org/abs/2402.07207 # post paper pdf link here +paperLink: https://gala3d.github.io/ # post paper pdf link here --- We present GALA3D, generative 3D GAussians with LAyout-guided control, for effective compositional text-to-3D generation. We first utilize large language models (LLMs) to generate the initial layout and introduce a layout-guided 3D Gaussian representation for 3D content generation with adaptive geometric constraints. We then propose an instance-scene compositional optimization mechanism with conditioned diffusion to collaboratively generate realistic 3D scenes with consistent geometry, texture, scale, and accurate interactions among multiple objects while simultaneously adjusting the coarse layout priors extracted from the LLMs to align with the generated scene. Experiments show that GALA3D is a user-friendly, end-to-end framework for state-of-the-art scene-level 3D content generation and controllable editing while ensuring the high fidelity of object-level entities within the scene. The source codes and models will be available at gala3d.github.io.