[ad_1]
Bill Wurtz-like approach to explaining 3D Gaussian Splatting
[ad_2]
View Reddit by loztriforce – View Source
[ad_1]
Bill Wurtz-like approach to explaining 3D Gaussian Splatting
[ad_2]
View Reddit by loztriforce – View Source
Damn, two-minute papers going back to being twoish minutes long is pretty refreshing!
I love this
Wow.
This might be good for people who want to be “wowed” or people who know the industry, but its pretty arcane to me. Multivariate? Alpha? “It looks like this:” *insert wave form*. Uh…ok.
All of the videos about this keep talking about its potential use for gaming or rendering, but as far as I can understand this fundamentally is only for static scenes (or as a high quality background with traditional CG overlaid) as the whole concept of the compression it achieves relies on a single “frame” with lighting being baked into the image and objects not existing as distinct geometry (like the lighting they are baked into an image as a whole). Can someone explain how this could work with real time lighting or movement of objects within the scene if even possible?
Besides being very annoying and not very educational, this video is also wrong. Gaussian splatting is not new, see for instance [this paper from 2001](https://www.cs.umd.edu/~zwicker/publications/EWASplatting-TVCG02.pdf). This is referenced in[ the paper](https://repo-sam.inria.fr/fungraph/3d-gaussian-splatting/3d_gaussian_splatting_low.pdf) this video tries to explain:
> Point-based methods efficiently render disconnected and unstruc-
tured geometry samples (i.e., point clouds) [Gross and Pfister 2011].
In its simplest form, point sample rendering [Grossman and Dally
1998] rasterizes an unstructured set of points with a fixed size, for
which it may exploit natively supported point types of graphics APIs
[Sainz and Pajarola 2004] or parallel software rasterization on the
GPU [Laine and Karras 2011; Schütz et al . 2022]. While true to the
underlying data, point sample rendering suffers from holes, causes
aliasing, and is strictly discontinuous. Seminal work on high-quality
point-based rendering addresses these issues by “splatting” point
primitives with an extent larger than a pixel, e.g., circular or elliptic
discs, ellipsoids, or surfels [Botsch et al . 2005; Pfister et al . 2000; Ren
et al. 2002; Zwicker et al. 2001b]
His comparison with photogrammetry is also strange, photogrammetry has nothing to do with path tracing, ray tracing, or diffusion. It’s used to produce a mesh model that can later be rendered, just as this is used to produce a 3D representation of a scene that is later rendered in real time.