Fuji Tsang Clément

177 posts

Fuji Tsang Clément

Fuji Tsang Clément

@Caenorst

AI Enthusiast, Research Scientist at NVIDIA

US / FR / CA Beigetreten Ağustos 2012
175 Folgt148 Follower
Fuji Tsang Clément retweetet
NVIDIA AI Developer
NVIDIA AI Developer@NVIDIAAIDev·
Most 3DGS segmentation tools either pre‑train per scene or lock errors into a feature field you can’t undo. ArtisanGS instead turns a few 2D masks into editable 3D object selections via Cutie tracking + black‑box splat aggregation, then lets you iteratively correct mistakes with consistent 2D/3D selection modes. 📗 #NVIDIAResearch paper: arxiv.org/abs/2602.10173
English
7
89
425
41.5K
Fuji Tsang Clément
Fuji Tsang Clément@Caenorst·
@VFSGlobal @vfsglobalcare Can you please try to fix your website? I've been trying to apply for Visa for France for my relatives to attend my wedding for a month now and it's always crashing...
English
1
0
0
45
Fuji Tsang Clément
Fuji Tsang Clément@Caenorst·
So proud of the Kaolin team with our new release, join our Hands-on lab at #SIGGRAPH2024 to learn how Kaolin can improve your 3D Deep Learning workflow and enable interactive physics in your Jupyter notebook! s2024.conference-program.org/presentation/?…
NVIDIA AI Developer@NVIDIAAIDev

✨Just announced: Representation agnostics physics simulation. ➡️ nvda.ws/3SuHUXZ This new framework is now available in the Kaolin Library. Simulate: ✅Gaussian Splats ✅Neural Radiance Fields ✅Signed Distance Functions and more…

English
0
0
3
152
Fuji Tsang Clément retweetet
NVIDIA AI Developer
NVIDIA AI Developer@NVIDIAAIDev·
✨Just announced: Representation agnostics physics simulation. ➡️ nvda.ws/3SuHUXZ This new framework is now available in the Kaolin Library. Simulate: ✅Gaussian Splats ✅Neural Radiance Fields ✅Signed Distance Functions and more…
GIF
English
6
26
101
30.2K
Jonathan Lorraine
Jonathan Lorraine@jonLorraine9·
Thrilled to announce I've completed my PhD! 🎉 My advisor, @DavidDuvenaud's support and guidance made the journey enjoyable. You showed me that research can be fun, and I will always cherish our brainstorming sessions on the whiteboard. Thank you for taking a chance on me and for your never-ending support, even during the more disappointing paper rejections. I'll always fondly look back on my times in grad school.
Jonathan Lorraine tweet media
English
24
3
275
25.1K
Fuji Tsang Clément retweetet
NVIDIA AI Developer
NVIDIA AI Developer@NVIDIAAIDev·
Exciting news for #PyTorch enthusiasts: Our NVIDIA Kaolin library introduces SurfaceMesh class to simplify managing mesh attributes including consistency, auto-compute normals, & streamline indexing. 👀 See the video tutorial from #NVIDIAResearch ➡️ nvda.ws/4aRCi1D
NVIDIA AI Developer tweet media
English
1
17
102
11.1K
Fuji Tsang Clément
Fuji Tsang Clément@Caenorst·
@JulieChangRE The article says "the middle 20 percent of income" and the bottom income is not the same as the top income of the bottom 20%. So there is a whole range of income missing (between $56,000 and $617,900 !!! and between $23,200 and $35,800)
English
0
0
0
14
Julie Chang
Julie Chang@JulieChangRE·
“Texans actually pay more in taxes than Californians do, unless those Texans are in the top one percent of all earners. “ chron.com/news/houston-t…
Julie Chang tweet media
English
519
3K
11.2K
1.8M
Fuji Tsang Clément retweetet
Masha Shugrina
Masha Shugrina@_shumash·
If you want to visualize 3D Gaussian Splatting radiance fields *interactively* in a Jupyter Notebook, here's an easy recipe using Kaolin Library: youtube.com/watch?v=OcvA7f…
YouTube video
YouTube
Masha Shugrina tweet media
English
1
15
86
7.1K
Fuji Tsang Clément
Fuji Tsang Clément@Caenorst·
Excited about this new release, the interactive visualizer and SurfaceMesh class are strong improvements of QoL when debugging a renderer or a model. Great job from the team @_shumash @OrPerel @FidlerSanja @NVIDIAAI
NVIDIA AI Developer@NVIDIAAIDev

📣 Just released: #NVIDIAKaolin v0.14.0 Accelerate 3D deep learning research, debug your custom renderer interactively in a Jupyter notebook, and manage batched mesh attributes with a SurfaceMesh container, within the Kaolin #Pytorch library. Github: nvda.ws/3QeuDCr

English
0
1
8
583
Fuji Tsang Clément
Fuji Tsang Clément@Caenorst·
Super excited about this release, Kaolin is now installable with a single command line! Check out our tutorials on lighting: Diffuse: github.com/NVIDIAGameWork… Specular: github.com/NVIDIAGameWork… Documentation: kaolin.readthedocs.io/en/latest/modu…
NVIDIA AI Developer@NVIDIAAIDev

📣 Just released: #NVIDIAKaolin v0.13.0 With new lighting features enabling #3D DL works like DIBR++ or NeRD, check out our tutorials on diffuse and specular reflectance. Kaolin is now pip installable. Installation: nvda.ws/3YN40FY Release notes: nvda.ws/3EbEYZ2

English
0
8
28
4.8K
Fuji Tsang Clément
Fuji Tsang Clément@Caenorst·
@RanaHanocka That can actually be very useful to generate test cases for 3D functions! I’ll definitely give it a try.
English
1
0
3
257
Rana Hanocka
Rana Hanocka@RanaHanocka·
Here is the chatgpt response. Chatgpt seems to be able to generate .objs for simple stuff, but the geometry starts to break down for more complicated requests. Still, I think this shows promise for a 3D model creation through a natural conversation! 2/
Rana Hanocka tweet mediaRana Hanocka tweet media
English
6
7
132
14.8K
Rana Hanocka
Rana Hanocka@RanaHanocka·
I asked chatgpt to generate an .obj file of a tetrahedron. It generated a list of vertices/faces in the .obj format. If you paste that into an empty file, it will open up in a 3D viewer, like meshlab! (It's even topologically correct: manifold, correct face normals, etc) 1/
English
30
109
1.6K
273.6K
Fuji Tsang Clément
Fuji Tsang Clément@Caenorst·
@srush_nlp @davidweichiang @boazbaraktcs One of the big advantage of named tensor is to allow flexibility for the compiler / JIT to make optimization in the backend without having any intervention from user, an example is the painful transition from NCHW to NHWC for convolution on recent GPU
English
1
0
12
3K
Fuji Tsang Clément
Fuji Tsang Clément@Caenorst·
@fhuszar Hehe, good espresso at home take some time and skills (and money), especially if you like those light roast goodness
English
0
0
3
0
Ferenc Huszár
Ferenc Huszár@fhuszar·
How it started — how it’s going.
Ferenc Huszár tweet mediaFerenc Huszár tweet media
English
8
2
161
0
Fuji Tsang Clément
Fuji Tsang Clément@Caenorst·
@wojczarnecki @francoisfleuret @PyTorch Well, broadcasting operations are way more efficient than the element-wise alternative. You can "expand" to have a similar effect to broadcasting (unless you use "contiguous()") but lots of researchers would use "repeat" without knowing that it doesn't behave the same.
English
1
1
1
0
Wojciech Czarnecki
Wojciech Czarnecki@wojczarnecki·
@francoisfleuret @PyTorch I have the same feeling about forbidding all the DL libraries from implementing auto broadcasting. Just... Don't. Vector plus vector should never result in a matrix......
English
2
1
8
0
François Fleuret
François Fleuret@francoisfleuret·
If @PyTorch was printing the sizes, dtype and devices of all the tensors involved in an operation that failed, we would be getting AGI ten years earlier.
English
16
83
1.2K
0
Fuji Tsang Clément retweetet
Jun Gao
Jun Gao@JunGao33210520·
Excited to share our #NeurIPS2022 @NVIDIAAI work GET3D, a generative model that directly produces explicit textured 3D meshes with complex topology, rich geometric details, and high fidelity textures. #3D Project page: nv-tlabs.github.io/GET3D/
English
38
548
2.7K
0

Entdecken

@VFSGlobal @vfsglobalcare @jonLorraine9 @DavidDuvenaud @JulieChangRE @_shumash @OrPerel @FidlerSanja