Xiaoyu Xiang
@mukosame
Followers
446
Following
1K
Media
60
Statuses
591
Research Scientist @RealityLabs. Working on generative models for 3D, and computational photography. Previous @Purdue @bytedanceTalk @Tsinghua_Uni
Bay Area, CA
Joined October 2017
This work is accepted to #WACV 2025. Congratulations to Liat, Yuchen, Omri, @fiandola , Yilei, Ita, Rakesh!.#texturesynthesis #3dmodeling.
0
0
0
โจ Excited to share our latest work: Make-A-Texture! ๐จ.โก๏ธ It generates high-res texture maps from text prompts for 3D models in just 3.07 seconds with comparable quality to existing SOTAs. Check it out! ๐.๐ Paper: ๐ Website:
2
0
6
RT @bingyikang: Want to use Depth Anything, but need metric depth rather than relative depth?. Thrilled to introduce Prompt Depth Anything,โฆ.
0
77
0
Amazing generative physics engine powered by Taichi โฏ๏ธ.
Everything you love about generative models โ now powered by real physics!. Announcing the Genesis project โ after a 24-month large-scale research collaboration involving over 20 research labs โ a generative physics engine able to generate 4D dynamical worlds powered by a physics
0
0
5
RT @tkexpress11: [New] @a16zgames AI x Game Dev Survey 2024. We surveyed 650+ game devs & the results are fascinating:.- 73% of studios areโฆ.
0
24
0
SpiRobs morphologically replicate the logarithmic spiral that is ubiquitous in natural organisms. They are easy and fast to build across scales via 3D printing. They are actuated by cables, which allow for fast and life-like movements.
0
0
0
Sharing an interesting work about spiral robots with simple tendon drives, replicating biological flexibility for diverse gripping tasks. Paper:
2
0
1
RT @_BiaoZhang: ๐ข๐๐๐จ๐ฆ๐๐ญ๐ซ๐ฒ ๐๐ข๐ฌ๐ญ๐ซ๐ข๐๐ฎ๐ญ๐ข๐จ๐ง๐ฌ.We propose to represent general geometries as distributions using diffusion models. project site:โฆ.
0
31
0
RT @sarafianosn: A New Approach for 3D Mesh Editing. Adding rabbit ears to a bird in 3D is a pretty challenging task but it's now possibleโฆ.
0
25
0
๐Thrilled to share our work EfficientTAM, a small, fast while high-performance model.๐ฑruns >10FPS on iPhone 15 ProMax, >2x faster than SAM2.
[1/n] ๐๐ณ๐ณ๐ถ๐ฐ๐ถ๐ฒ๐ป๐ ๐ง๐ฟ๐ฎ๐ฐ๐ธ ๐๐ป๐๐๐ต๐ถ๐ป๐ด from @Meta: interactive video segmentation and tracking on an iPhone!
0
0
11
RT @YoungXiong1: ๐Excited to share our Efficient Track Anything. It is small but mighty, >2x faster than SAM2 on A100 and runs > 10 FPS onโฆ.
0
37
0
RT @YoungXiong1: ๐จVideoLLM from Meta!๐จ.LongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding. ๐Paper: https://tโฆ.
0
73
0
MIT News covers our #CVPR2024 award-candidate paper PlatoNeRF ๐ congrats to @tklinghoffer! . Check our oral && poster sessions on Thursday!.๐๏ธOral: Summit Flex Hall AB, 1:54-2:12pm .๐ฐPoster: Arch 4A-E Poster#2, 5:00-6:30 pm.
news.mit.edu
A new technique can model an entire 3D scene, including areas hidden from view, from just one camera image. The method, developed by MIT and Meta researchers, relies on image shadows, which provide...
0
3
13
PlatoNeRF is selected as oral at #CVPR2024 . Congratulations to all authors! @tklinghoffer @sidartsoma Yuchen Fan, @c_richardt @raskarmit and Rakesh Ranjan.
Meta announces PlatoNeRF. 3D Reconstruction in Plato's Cave via Single-View Two-Bounce Lidar. paper page: propose PlatoNeRF: a method to recover scene geometry from a single view using two-bounce signals captured by a single-photon lidar
1
4
37
RT @sarafianosn: We introduce a new method to stylize 3D garments!. We generate new garments by stylizing the geometry & texture given textโฆ.
0
12
0