3D Gaussian Splatting explained in a non technical way

Public Speaking for Geospatials: Storytelling, Networking & Presentations - Juraj Holub

Public speaking, business storytelling, networking, and technical presentation skills are no longer “soft skills” — they are survival skills for geospatial, reality capture, 3D, digital twin, and technology professionals.In this episode, Michal talks with Juraj Holub, international moderator and public speaking coach, about how technical experts can stop sounding boring, explain complex ideas clearly, pitch clients through stories, network without awkward small talk, and deliver presentations people actually remember.Juraj breaks down why public speaking is a meta skill for business, how storytelling turns technical content into client value, why audience research matters, how to use analogies without losing people, and why AI-generated presentations can make speakers sound generic if they outsource the thinking.You’ll learn practical frameworks for better presentations, stronger conference networking, client storytelling, elevator pitches, and authentic public speaking in the age of ChatGPT.00:00:00 Why speaking is the most important business skill00:02:05 Becoming a geospatial event moderator00:03:35 Where public speaking talent comes from00:05:20 Great communication means showing, not telling00:07:03 Passion, likability, and authenticity00:10:03 Storytelling for technical professionals00:10:54 Why stories are hardwired into humans00:13:21 Why smart technical people sound boring00:16:37 How to make any story sticky00:21:38 The 5 elements of a strong story00:24:08 Know your audience before telling stories00:28:06 How to prepare for a technical presentation00:31:14 Using analogies to explain complex ideas00:33:27 When pop culture references fail00:38:22 Storytelling for geospatial client pitches00:44:22 Before-after-bridge for sales stories00:45:49 When to use storytelling with clients00:48:29 How beginners collect business stories00:51:28 Networking as a business skill00:55:12 Fear of rejection at events00:58:06 Moderated vs freestyle networking00:59:32 The elevator pitch formula01:01:20 Small talk for people who hate small talk01:07:18 Structuring business presentations01:10:17 How to give the worst presentation ever01:11:45 Start with the punchline01:13:23 Why generic AI presentations fail01:17:03 Stop reading slides and AI scripts01:24:56 How to deliver a great presentation01:29:40 Humor, pauses, and human delivery01:34:36 Public speaking workshops and final thoughts

From 3D Scans to Indoor Navigation and Tracking

I am speaking to Nikhil Sawlani, founder of MultisetAI.Instead of delivering only a static point cloud or 3D mesh, reality capture professionals can reuse already-scanned spaces to create live digital twin services: indoor positioning, 6DoF tracking, asset navigation, connected worker workflows, smart glasses guidance, no-code AR instructions and robot-ready spatial maps.We cover how Multiset AI works, why GPS and Bluetooth beacons fail indoors, what scan data is required, how confidence scoring and map versioning handle changing environments, and where the clearest upsell opportunities are for scan-to-BIM, industrial facilities, shopping malls, factories and enterprise digital twin teams.Chapters00:00:00 Are reality capture companies leaving money on the table?00:01:00 Point clouds vs real digital twins00:02:38 Upselling already captured scan data00:04:05 360 video, SLAM, LiDAR and scan quality00:07:03 What Multiset AI actually does00:09:10 Why GPS and GNSS fail indoors00:11:57 Bluetooth beacons, Wi-Fi and UWB limitations00:17:23 Hardware-free visual positioning systems00:19:18 5–10 cm indoor accuracy and 6DoF tracking00:23:40 How VPS matches camera views to 3D scans00:27:01 Turning massive point clouds into machine-readable maps00:30:46 Handling changing environments and crowds00:34:01 Confidence scores, sensor fusion and accuracy control00:41:54 AR navigation, overlays and device tracking00:44:39 Supported scanners, maps and file formats00:48:30 Panoramas, point clouds and E57 requirements00:53:37 Business case: upselling navigation from existing scans00:57:53 AR work instructions and connected worker use cases00:59:54 AI agents for factory asset navigation01:03:18 Mapping with iPhone LiDAR, Matterport, NavVis, Leica and XGRIDS01:07:07 Gaussian splats, metric scale and VPS localization01:13:11 Robotics, physical AI and one spatial source of truth01:15:36 Smart glasses navigation with Meta Ray-Bans01:17:32 Onboarding, free tier, SDKs and no-code options01:19:50 Final advice for reality capture professionals

Most Digital Twins Are Just 3D Postcards — w/ Barry Bassnett

Digital twins, 3D Gaussian splatting, OpenUSD, point clouds, 3D meshes, and reality capture are transforming how facilities, factories, surveyors, and asset managers work — but Barry Bassnett argues most “digital twins” fail because they solve the wrong problem.In this episode, Barry explains why a point cloud or 3D mesh is usually not a digital twin, why “visual twins” still matter, and how digital twins become valuable only when they make information findable, actionable, updated, and useful to real humans.We cover the difference between digital twins and visual twins, the “LOD zero” idea for 3D replicas, why digital twin projects fail, how to sell digital twins through ROI and “search tax,” why Gaussian splatting could become the ultimate digital twin interface, and why OpenUSD may change data interoperability across BIM, AEC, reality capture, Omniverse, Rhino, point clouds, meshes, and 360 imagery.Barry also breaks down why facilities managers often do not need millimeter accuracy, why surveyors may be missing a massive market, how AI can help with tagging and inspection, and why the future of digital twins depends less on beautiful 3D visuals and more on solving one painful operational problem.Link to Barry's e-book: https://payhip.com/b/gEptoSponsor of the episode: www.twinzo.com A digital twin platform for logistics optimization.Chapters:00:00:00 Are most digital twins useless?00:02:11 Why a point cloud is not a digital twin00:04:00 The “LOD zero” idea for 3D replicas00:05:18 The atlas metaphor for digital twins00:07:22 Digital twins are a spectrum, not a binary00:09:07 Why humans struggle with point clouds00:10:52 Why 3D looks impressive but often fails operationally00:11:41 3D asset management and digital twin interfaces00:14:42 Findability: the missing layer in most digital twins00:16:33 Sponsor: Twinzo and internal logistics visibility00:17:22 Simple digital twins that work for CEOs and ground teams00:18:48 The cartography principle: leave out what does not matter00:20:10 Paper mill case study: grease points and real ROI00:22:36 Why digital twins should start as one problem, not everything00:23:36 Why “we digitize everything” is a dangerous sales pitch00:25:58 Digital twins must be updated or they become historic documents00:27:20 Search tax: selling digital twins as time savings00:29:32 Knowledge management, SharePoint, and spatial search00:32:55 Why a car is a real-world digital twin00:33:44 Human memory, spatial context, and digital twin UX00:35:50 Audience poll: is a point cloud a digital twin?00:38:49 Why Barry Bassnett is qualified to challenge digital twin hype00:41:58 How Barry actually builds digital twins00:45:20 Choosing the right platform for the right user00:47:53 OpenUSD, Omniverse, Rhino, and data portability00:50:37 Why OpenUSD adoption is still early00:52:36 OpenUSD explained through Pixar and HTML for 3D00:57:56 AI ethics in reality capture and digital twins00:59:06 When AI reconstruction is acceptable — and when it is not01:00:48 AI for crack detection, inspection, and object recognition01:03:36 Why AI will create better work, not just remove jobs01:05:48 AI understanding vs AI knowing everything01:07:48 Barry’s view on 3D Gaussian splatting01:09:30 Why bad imagery creates bad Gaussian splats01:10:15 Quality in, quality out: capture discipline still matters01:12:36 Gaussian splatting as another layer of reality capture01:13:01 Why surveyors should add Gaussian splats to deliverables01:14:29 Digital twins, accuracy, and when precision matters01:16:25 Gaussian splatting, OpenUSD, AR, and measurable spatial interfaces01:18:48 Photogrammetry skills that still matter01:22:18 Image processing, tagging, and making captured data useful01:24:17 Low-hanging fruit for upselling scans into digital twins01:26:13 Barry’s ebook: Tower of Twins01:29:06 Coaching, mentoring, and sharing digital twin mistakes01:31:39 Closing thoughts and how to contact Barry

DJI Ban, AI Drones & LiDAR Mapping: Dylan Gorman on the Future of Drone Reality Capture

In this episode, we go deep into the commercial drone industry: DJI vs. U.S. drone manufacturers, Skydio and lobbying, FCC authorization, drone data accuracy, stockpile measurements, construction mapping, LiDAR workflows, autonomous drone flight, AI-assisted outputs, 3D Gaussian splatting, RealityScan workflows, and how beginners can enter the drone mapping market with simple tools. Dylan explains why a complete DJI shutdown would disrupt the U.S. drone industry, why existing FCC-authorized drones remain a different issue than future authorizations, and why commercial drone operators must understand the difference between “cool data” and useful deliverables. Current FCC Covered List rules affect new authorizations for certain foreign-made drones, while previously authorized hardware remains a separate category under the regulatory discussion. You’ll learn why drone mapping is not just “flying a drone,” how one stockpile measurement saved a client $60,000, why not every project needs survey-grade accuracy, how AI is changing flight planning and data processing, and why the best business strategy is often the KISS method: keep it simple, solve one real problem, and deliver reliable data. Guest:  ⁨@DylanGorman⁩  [https://studio.youtube.com/channel/UCcv7q4TeTBlso7Mvc0aIAAA] Company:  ⁨@pilotbyte-labs⁩  [https://studio.youtube.com/channel/UC3KlXwahUb-xDi9L9U9SM8Q] Partner of the episode:  ⁨@NUBIGON⁩  [https://studio.youtube.com/channel/UCmMIsCGy1c2fAEC6gIhdCvw] - Showcase Your Reality Capture & Scan-to-BIM Projects with Animations and Renders

3D Gaussian Splatting explained in a non technical way

With Olli Huttunen 3D Gaussian Splatting workflow, 360 camera capture, Postshot processing, Blender tooling, PlayCanvas publishing, and synthetic Gaussian splats — this episode with Olli Huttunen breaks down what actually works in production. Olli explains 3DGS in plain language, shows why capture quality still matters, shares his Magic Wand multi-camera rig, and walks through his 360 Extractor plugin and Dioramics presentation system for interactive 3D storytelling. Inside the conversation, Olli explains 3D Gaussian Splatting like 3D pointillism, compares 3DGS with photogrammetry and NeRFs, and details where Gaussian splats shine: fast, photoreal scene rendering, immersive VR viewing, browser delivery, and lightweight presentation workflows. He also covers the tradeoffs: soft edges up close, weak geometry for mesh conversion, lighting inconsistency, floaters, and why fewer images is not the same as better capture. You’ll also hear Olli’s practical 360 scanning workflow: mixing camera heights, extracting usable views from equirectangular footage in Blender, removing operator-visible angles, combining ground and drone captures, and optimizing datasets for better training on limited VRAM. He shares why Postshot handled mixed-camera datasets better than Luma in his tests, why overcast conditions still matter, and how geospatial, VR, and synthetic 3DGS workflows are evolving. The episode also goes deep on browser-native 3DGS tools. Olli discusses PlayCanvas SuperSplat, SOG compression for smaller splat files, Babylon.js-based Dioramics, iframe embeds, offline-sensitive use cases, and the emerging opportunity to turn Gaussian splats into interactive presentations instead of static showcases.