In the past, ShapeNet has enabled remarkable progress and benchmarking across 3D computer vision! But, it lacks visual diversity, realismโฆ
and scale! ๐ชจ Objaverse is more than an order of magnitude larger and has ~400x more categories! ๐
Objaverse is incredibly diverse and includes a vast amount of visual styles! โ Photogrammetry Scanned Objects โ 3D Modeled Objects โ Interiors โ Exteriors โ Characters โ High-Poly โ Low-Poly โ Cartoon And More!
Many objects are also animated! ๐ฆ
We began to show the potential of Objaverse by generating 3D objects! ๐๐๐๐ Here we use the amazing work of GET3D from
@JunGao33210520et al. to generate textured meshes!
Weโre also thrilled by the potential for Objaverse to impact robotics! ๐ค We introduce a task where the agentโs goal is to navigate to objects as described by language. ๐ Hereโs a house we set up purely with objects in Objaverse:
Objaverse is hosted on ๐ค
@huggingface! Weโre thrilled to have worked with their incredible team to make it easily downloadable! ๐ Individual objects are available under a Creative Commons license. ๐ค Dataset:
Rendering images of the 3D objects is very fast with Python in
@Blender! All objects are standardized GLB files. ๐ GitHub Rendering Scripts:
Weโre thrilled to see what you build! ๐คฉ Just this week,
@ruoshi_liuet al. released Zero-1-to-3, incredible work that combines a generative image model and Objaverse, for learning a shape prior, to render novel viewpoints of an object!
Text2Tex is an exciting paper using Objaverse that also came out just this week by
@davech2yet al. for texturing 3D meshes from a text description!
And even earlier this month, Anchit Gupta,
@XiongWenhan,
@EasonNie, Ian Jones, Barlas Oฤuz at
@MetaAIused Objaverse in work on 3DGen for triplane textured mesh generation!