Orbis Tabula

We're still in the very early days of Radiance Field exploration, but the community building the technology to make this viable is growing, and very excited about this field of research. #nerfstudio #radiancefields #opensource

Absolutely incredible breakdown of Dylan's work on his Environmental Cinematic.

Clients are going to LOVE this...

HO HO HO! 🎅 We are making Atlux λ for Unreal Engine available to everyone, giving one month for free to celebrate the end of the year and help with the January struggle! Follow the instructions here to activate your license and get the installer: https://atlux.ai/christmas CODE: ATLUXλGIFT All the best for the holidays, and I hope you enjoy Atlux λ! ❤️

“Victory over Google! After 4 weeks of detailed court testimony, the California jury found against the Google Play monopoly on all counts. The Court’s work on remedies will start in January. Thanks for everyone’s support and faith! Free Fortnite!” - Tim Sweeney This is a massive victory! Google has already stated they will appeal, but that's a larger hurdle coming from a jury verdict.

In addition to reconstruction artifacts due to poor capture or camera pose estimation, aliasing and scale are issues that researchers are working to overcome. This work provides benefits in anti-aliasing, similar to Tri-MipRF and Mip-VoG, as well as multisampling as in Zip-NeRF.

❓Do you know how to make a turntable animation render for any open project in RealityCapture using Python? Check out the step-by-step process at https://lnkd.in/gtsUY7pm #RealityCapture #photogrammetry

🤪 One of the greatest limitations in getting good results from NeRF or Gaussian Splatting is the camera pose estimation. COLMAP is the default for many implementations, but it's not ideal as it takes a good bit of time and it's not the most accurate. BARF and CAMP were good upgrades to this process, and yet Ant Group & Tencent AI Lab researchers have come out swinging with this gem. Abstract: "Neural Radiance Fields (NeRF) have achieved photorealistic novel views synthesis; however, the requirement of accurate camera poses limits its application. Despite analysis-by-synthesis extensions for jointly learning neural 3D representations and registering camera frames exist, they are susceptible to suboptimal solutions if poorly initialized. We propose L2G-NeRF, a Local-to-Global registration method for bundle-adjusting Neural Radiance Fields: first, a pixel-wise flexible alignment, followed by a framewise constrained parametric alignment. Pixel-wise local alignment is learned in an unsupervised way via a deep network which optimizes photometric reconstruction errors. Frame-wise global alignment is performed using differentiable parameter estimation solvers on the pixel-wise correspondences to find a global transformation. Experiments on synthetic and real-world data show that our method outperforms the current state-of-the-art in terms of high-fidelity reconstruction and resolving large camera pose misalignment. Our module is an easy-to-use plugin that can be applied to NeRF variants and other neural field applications." Project Page: https://lnkd.in/efMr2H7F arXiv: https://lnkd.in/evwdxsrz GitHub: https://lnkd.in/e_YCYAe3 License (MIT): https://lnkd.in/emK8uWqn For more like this ⤵ 👉 Follow Orbis Tabula #NeRF #cameraposeestimation #cameraalignment

🤯 When you understand the ceiling potential, you can ignore the negative naysayers... "I can't use it if it doesn't mesh." "I can't use it, the mesh isn't good enough." "NeRFs are pointless, but have you seen Gaussian Splatting?" 🙄 Consider this development from NVIDIA's Toronto Research Lab. NeRF. Meshing. Performant. Now all we need is the code and a viable license. Abstract: "Neural radiance fields achieve unprecedented quality for novel view synthesis, but their volumetric formulation remains expensive, requiring a huge number of samples to render high-resolution images. Volumetric encodings are essential to represent fuzzy geometry such as foliage and hair, and they are well-suited for stochastic optimization. Yet, many scenes ultimately consist largely of solid surfaces which can be accurately rendered by a single sample per pixel. Based on this insight, we propose a neural radiance formulation that smoothly transitions between volumetric- and surface-based rendering, greatly accelerating rendering speed and even improving visual fidelity. Our method constructs an explicit mesh envelope which spatially bounds a neural volumetric representation. In solid regions, the envelope nearly converges to a surface and can often be rendered with a single sample. To this end, we generalize the NeuS formulation with a learned spatially-varying kernel size which encodes the spread of the density, fitting a wide kernel to volume-like regions and a tight kernel to surface-like regions. We then extract an explicit mesh of a narrow band around the surface, with width determined by the kernel size, and fine-tune the radiance field within this band. At inference time, we cast rays against the mesh and evaluate the radiance field only within the enclosed region, greatly reducing the number of samples required. Experiments show that our approach enables efficient rendering at very high fidelity. We also demonstrate that the extracted envelope enables downstream applications such as animation and simulation." Project Page: https://lnkd.in/eRCdAPxD arXiv: https://lnkd.in/e_J3mE2R Supplemental: https://lnkd.in/efXEh8Uz For more like this ⤵ 👉 Follow Orbis Tabula #NeRF #neuralnetworks #geometry

#QGIS v3.34 now with Cesium/The Open Geospatial Consortium (OGC) #3DTiles