Cedric Rousteau
@errno_h
Senior Research Engineer at @unity3d , Former Head of Research at @PiXYZSoft , CAD, Mesh, Geometry Processing, Rendering, C++
Bạn có thể thích
Check out our latest talk at #UniteCopenhagen ! You’ll learn more about @PiXYZSoft algorithms and how they help to turn large #CAD assemblies into assets ready for realtime experiences in @unity3d youtu.be/BNUZLnhCR2Q
youtube.com
YouTube
Turning large CAD assemblies into real-time 3D visualizations - Unite...
“Everyone knows” what an autoencoder is… but there's an important complementary picture missing from most introductory material. In short: we emphasize how autoencoders are implemented—but not always what they represent (and some of the implications of that representation).🧵
New Anthropic research: Tracing the thoughts of a large language model. We built a "microscope" to inspect what happens inside AI models and use it to understand Claude’s (often complex and surprising) internal mechanisms.
📢Sonata: Self-Supervised Learning of Reliable Point Representations📢 Meet Sonata, our"3D-DINO" pre-trained with Point Transformer V3, accepted at #CVPR2025! 🌍: xywu.me/sonata 📦: github.com/facebookresear… 🚀: github.com/Pointcept/Poin… 🔹Semantic-aware and spatial…
Excited about the M3 release! 🚀 It introduces Gaussian Memory Attention for efficient alignment between 3D Gaussians and 2D foundation models. 🔥 Only 16 degrees inside one Gaussian primitive for one LMM! Looking forward to more applications! 🤖✨
[1/n] We are releasing M3 (#ICLR2025): a Gaussian Splatting method that builds LMM memories for arbitrary scenes. 🔥 [Efficient] 16 degrees in each Gaussian primitive for one LMM. 🔥 [Alignment] The rendered features are directly in the source LMM embedding space.
having a bit of an Intelligence aid from Gemini 2.0 👌⚙️⛵ huggingface.co/spaces/tinazon…
wow.. Sesame just open sourced its 1B base AI model CSM (Conversational Speech Model) - trained on 1 MILLION hours of data - voice cloning - based on llama architecture - real-time synthesis - free to try and download now Link in comments.
All the LLMs and other deep learning models are based on neural networks. We can think of them as mathematical functions with hundreds of billions of parameters. Those parameters (weights) are determined during training and we train these networks with trillions of tokens (text,…
Introducing Genie 2: our AI model that can create an endless variety of playable 3D worlds - all from a single image. 🖼️ These types of large-scale foundation world models could enable future agents to be trained and evaluated in an endless number of virtual environments. →…
"Half-edge data structure considered harmful", by @logicalerror sandervanrossen.blogspot.com/2017/09/half-e…
The quest for the best SSAO continues!💪 SSRT3 raw.githack.com/zalo/three.js/… GTVBAO raw.githack.com/zalo/three.js/… github.com/mrdoob/three.j…
First Time Ghost scenes released! ✨ We’ve just published two projects: showcasing the character and the grassy environment scene. Have a look at the hair simulation…
C++ code is out! The Signed Heat Method has been added to geometry-central. In 3D, compute generalized SDFs to point clouds, triangle meshes, and polygon meshes. The method should "just work" on geometry with holes, intersections, nonmanifold-ness, etc. Links below 🔽
Signed distance functions (SDFs) are fundamental tools in graphics, vision, and physics simulation. But how do you get a high-quality SDF from messy, real-world input? At #SIGGRAPH2024, we introduced a simple method for turning "broken" geometry into a well-behaved SDF. <🧵>
Holomorphic functions (differentiable over complex numbers) define conformal maps, which are angle-preserving. en.wikipedia.org/wiki/Conformal… en.wikipedia.org/wiki/Holomorph…
I wrote my entire NES game in C, using the 8bitworkshop.com IDE. The demo was just published a few days ago and you can play it online right now!! 🌴x.com/ninjadynamics/…
What is happening inside these generative AI companies?!? There are multiple open-source mesh-processing libraries *with sample code* that can do a better job at meshing than this!
Comparison of Direct, ZIgnore, Cube Spherical, Fibonacci, Spheremap and Octahedral at 8 bits, 12 bits 16 bits, 24 bits and 32 bits per normal: shadertoy.com/view/llfcRl
a guide to why you should hire an artist and never pay for generative 3D services. on the left is an AI generated Squirtle model posted earlier today. on the right is a Squirtle I modeled tonight using only that one reference image. my sculpt took an hour and a half.
I took an image from the original Pokémon artists to see how far cube.csm.ai can go: 3d.csm.ai/detail/SESSION…
United States Xu hướng
- 1. #DWTS 24.4K posts
- 2. #WWENXT 12.5K posts
- 3. Robert 103K posts
- 4. #DWCS 3,866 posts
- 5. Carrie Ann 2,559 posts
- 6. #RHOSLC 3,324 posts
- 7. Elaine 49.6K posts
- 8. Whitney 13.7K posts
- 9. Max Muncy 1,857 posts
- 10. Dylan 31.8K posts
- 11. Young Republicans 44.9K posts
- 12. Roldan 1,739 posts
- 13. #USMNT N/A
- 14. Haji Wright N/A
- 15. Politico 254K posts
- 16. Peralta 5,535 posts
- 17. The Rangers 8,494 posts
- 18. D’Angelo 399K posts
- 19. Rahiki 1,217 posts
- 20. Charlie 787K posts
Something went wrong.
Something went wrong.