# @radiancefields on YouTube

- **Type:** Video
- **Original URL:** https://youtube.com/watch?v=dNnbTHvOu7I
- **Gondola URL:** https://gondola.cc/posts/68998142-radiancefields-youtube
- **Thumbnail:** https://img.gondola.cc/tr:w-,h-,fo-auto/postThumbnails/d00a2fbbcc.jpg
- **Posted:** 2026-08-11T16:02:10.000+00:00
- **Account Owner:** Michael Rubloff (@RadianceFields) — https://gondola.cc/RadianceFields

## Caption

Esri has shipped production Gaussian splatting across ArcGIS — at city, state, and national scale. At Esri's User Conference in San Diego, I sat down with Arkadiusz Szadkowski, who leads reality mapping business development globally at Esri, to talk about how a technique that was GitHub demos three years ago became a delivered layer in national mapping programs.

We get into why Gaussian splatting solves two problems meshes have never handled well, thin structures and reflective surfaces, and why that opens up the infrastructure market specifically power lines, substations, telecom towers, linear assets, open-pit mining.

We also cover analysis beyond rendering (line of sight, viewshed, measurement, snapping, segmentation), why file format standardization through Khronos and glTF matters for procurement, and what geospatial AI applied on top of splat texture information looks like, including a tree inventory that mapped 7,000 trees, with canopy size and height, in six minutes.

CHAPTERS
0:00 Introduction
0:14 Who is Arek — reality mapping at Esri
0:52 Thirty years from pen-and-paper survey to photogrammetry
2:00 The mission: bridging GIS and modern geospatial capture
3:03 How new reality mapping still is at Esri
3:49 What Gaussian splatting actually adds to a mapping stack
4:30 Discovering splats on a family holiday
5:02 Pitching it internally to Konrad and Stuttgart
5:19 Same SfM input, same engine, a new output layer
5:57 Why it's a game changer: thin structures
6:50 Reflective surfaces and the infrastructure opportunity
7:11 Papers to production at scale in three years
7:28 National programs and government-scale capture
8:12 "It looks great, but what can we analyze?"
9:16 Line of sight, viewshed, measurement, segmentation
9:45 Classification and geospatial AI on splat texture data
10:10 Interoperability, Khronos, and why glTF matters
10:50 The OBJ cautionary tale
12:21 Co-opetition: why competitors standardizing helps end users
13:02 Who the early adopters are
13:22 The resolution ceiling — why 3 inch / 7.5 cm is the line
14:17 Heavy infrastructure, drones, AEC, mining
15:34 Splats as the free extra layer that builds trust
16:31 What the next era of mapping looks like
16:43 Compressing a year-long delivery cycle
18:14 From maps to insights
18:31 The real unlock: temporal resolution
19:07 7,000 trees in 6 minutes
19:56 Geospatial AI needs training, not perfection on day one
20:40 What Arek hopes attendees take away
21:29 "Advertising is the cost of being unremarkable"
22:33 How to actually get started with Esri reality mapping
23:30 Wrap

MENTIONED
ArcGIS Reality Studio, Site Scan for ArcGIS, Drone2Map, ArcGIS Reality for ArcGIS Pro,
Reality Mapping User Type Extension, Khronos Group, glTF

Data Attribution
Data courtesy of: PhaseOne, Flycom Technologies, Mapping Authority of Slovenia​. Captured with PhaseOne PAS 880​. Processed with ArcGIS Reality & ArcGIS Pro.
Data courtesy of: Geofly, LGL Baden-Württemberg.
Data courtesy of: IGI, City of Phoenix.

—

Radiance Fields is the independent publication covering Gaussian Splatting, NeRFs, and radiance fields.

#GaussianSplatting #RadianceFields #Esri #ArcGIS #GIS #Geospatial #RealityCapture
#Photogrammetry #3DReconstruction

## Stats

- **Views:** 693
- **Likes:** 32
- **Shares:** 0
- **Comments:** 2

## Tags

photogrammetry, realitycapture, gis, geospatial, gaussiansplatting, arcgis, 3dreconstruction, radiancefields, esri

---
Copyright (c) Gondola