English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Recent advances in generative AI have unveiled significant potential for thecreation of 3D content. However, current methods either apply a pre-trained 2Ddiffusion model with the time-consuming score distillation sampling (SDS), or adirect 3D diffusion model trained on limited 3D data losing generationdiversity. In this work, we approach the problem by employing a multi-view 2.5Ddiffusion fine-tuned from a pre-trained 2D diffusion model. The multi-view 2.5Ddiffusion directly models the structural distribution of 3D data, while stillmaintaining the strong generalization ability of the original 2D diffusionmodel, filling the gap between 2D diffusion-based and direct 3D diffusion-basedmethods for 3D content generation. During inference, multi-view normal maps aregenerated using the 2.5D diffusion, and a novel differentiable rasterizationscheme is introduced to fuse the almost consistent multi-view normal maps intoa consistent 3D model. We further design a normal-conditioned multi-view imagegeneration module for fast appearance generation given the 3D geometry. Ourmethod is a one-pass diffusion process and does not require any SDSoptimization as post-processing. We demonstrate through extensive experimentsthat, our direct 2.5D generation with the specially-designed fusion scheme canachieve diverse, mode-seeking-free, and high-fidelity 3D content generation inonly 10 seconds. Project page: https://nju-3dv.github.io/projects/direct25.

Direct2.5: Diverse Text-to-3D Generation via Multi-view 2.5D Diffusion