Category: photogrammetry

Today, Mapillary is launching NeRFs, a new feature that will allow you to explore landmarks and popular sites in detailed 3D views – all reconstructed from 2D images uploaded to Mapillary.

https://blog.mapillary.com/update/2024/03/11/Mapillary-NeRF.html

https://www.mapillary.com/app/?lat=17.751177534360437&lng=0&z=1.5

https://medium.com/@patriciogv/the-state-of-the-art-of-depth-estimation-from-single-images-9e245d51a315

https://www.reshot.ai/3d-gaussian-splatting

 what are 3D Gaussians? They are a generalization of 1D Gaussians (the bell curve) to 3D. Essentially they are ellipsoids in 3D space, with a center, a scale, a rotation, and “softened edges”.

Each 3D Gaussian is optimized along with a (viewdependant) color and opacity. When blended together, here’s the visualization of the full model, rendered from ANY angle. As you can see, 3D Gaussian Splatting captures extremely well the fuzzy and soft nature of the plush toy, something that photogrammetry-based methods struggle to do.

https://towardsdatascience.com/a-comprehensive-overview-of-gaussian-splatting-e7d570081362

https://neuralradiancefields.io/luma-interactive-scenes-announced/

“…these are in fact Gaussian Splats that are being run and it’s a proprietary iteration of the original Inria paper. They hybridize the performance gain of realtime rendering with Gaussian Splatting with robust cloud based rendering that’s already widely being used in commercial applications. This has been in the works for a while over at Luma and I had the opportunity to try out some of my datasets on their new method.”

MICHAEL RUBLOFF

(more…)