Help Center

Straight Answers

The questions I hear most, answered the same way I'd answer them on a call.

Getting Started

What is the 3D Geodata Academy?

The Academy is where I teach the full 3D AI chain. Point clouds, photogrammetry, neural rendering, spatial AI. Structured paths, real projects, and a community where I’m active every week.

Which learning path should I choose?

Start with the free mission. It takes you from a raw LiDAR scan to a classified point cloud in 15 minutes. After that you’ll know which path fits you, and I’ll point you to the right track.

Do I need prior programming experience?

Intermediate Python helps. If you’re brand new to Python I include the basics you need for 3D processing. The path adapts to where you actually are.

How long does it take to complete a learning track?

You get clarity from episode 1. Your first working artifact ships in 15 minutes. A focused week full-time covers the accelerator. At 10 hours a week it runs under 30 days. Lifetime access, so you come back whenever you need to.

Which course should a surveyor start with?

Point Cloud Intelligence. It runs from raw LiDAR to automated classification and 3D modelling across five modules on real datasets, and it is written for people who already handle scan data professionally rather than as an introduction to 3D. If you have never written Python, start with Python and AI Programming for 3D Applications first, then come back to Point Cloud Intelligence.

Do I need deep learning to classify a point cloud?

No. Plenty of production classification is geometric: RANSAC plane fitting, DBSCAN clustering and region growing solve a lot of real jobs without a single trained model, and they are cheaper to run and easier to defend to a client. Deep learning earns its place when the classes are semantic rather than geometric. Start with the segmentation and clustering guide, then move to the 3D Segmentor OS when you need learned labels.

Programs & Pricing

What is included in the 3D AI Architect Program?

The 3D AI Architect Program has three tiers, all lifetime access. Foundation (EUR 1,997) is the full path at your own pace: the 17-episode Spatial Accelerator, the whole course Library, and the Neurones 3D Software Suite. Professional (EUR 2,997) adds the four deep-dive production tracks, the Forge apps, the Neurones 3D (Science) part, and monthly live Q&A. Architect (EUR 12,997) adds one-to-one advisory, a private channel, and a custom solution we build together; that cohort is currently full and applications are open for the next intake. The optional annual renewal for new content is EUR 697 a year on Foundation and EUR 497 on Professional and Architect.

Can I purchase courses individually?

Yes. Every course in the Library can be bought individually. If you want more than two or three, the Library subscription or the full Program is cheaper.

Is there a free trial available?

Yes. The free mission gives you 4 full episodes, real datasets, a pre-configured stack and the first chapters of my ebook. No credit card. That’s the best way to see how I actually teach.

What's your refund policy?

14 days on single courses, the Library and Build Camp. Write to me within 14 days of purchase and I refund in full. The 3D AI Architect tiers carry a separate 90-day transformation guarantee, under the conditions in section 12 of the Terms of Use.

Can I try the automation before buying anything?

Yes, two ways, both free and neither behind a card. The free mission gives you four full episodes with real datasets and a pre-configured stack. Scan to Knowledge is the Neurones 3D desktop engine, one 15 MB file that turns raw point clouds into classified scenes on your own machine, with no installer, no account and no cloud.

Technical Requirements

What software and tools do I need?

Python 3.10+, VS Code, Open3D, CloudCompare, a handful of packages I pin in the stack. Everything is open source. Every course lists the exact setup up front.

What are the hardware requirements?

16GB RAM is a sane minimum today. For deep learning episodes a CUDA GPU helps but isn’t mandatory. I provide Colab notebooks (free GPU) for everything heavy.

Can I use Mac or Linux?

Yes. Windows, macOS and Linux all work. Platform-specific notes where they matter.

Do you provide datasets for practice?

Yes. Every course comes with real datasets. LiDAR, photogrammetry, and the kind of noisy scans you actually meet in the field. No toy data.

What software and file formats does a point cloud pipeline use?

Python with Open3D, laspy and NumPy, plus CloudCompare for inspection, reading LAS and LAZ for LiDAR and PLY and OBJ for meshes. Everything I teach is open source and pinned in a setup each course hands you. The ingestion guide covers the parts that bite in production, probing LAS headers with laspy and resolving coordinate reference systems and datums with pyproj, and the LiDAR processing guide runs the whole chain.

What does a team need to start automating point cloud deliverables?

One engineer who can write Python, a machine with 16GB of RAM, and a real dataset of your own. A CUDA GPU helps on the deep learning stages but is not required, and I provide Colab notebooks for anything heavy. The thing teams underestimate is having someone who owns the pipeline rather than treating it as a side task, which is why enterprise engagements are scoped on a call instead of sold per seat.

Learning Experience

How are the courses structured?

Each course is videos plus notebooks plus a real project. Theory lands only when it turns into code. I track your progress so you always know the next step.

Can I get help when I'm stuck?

Yes. The Spatialetics community is where I answer questions every week. Program members get priority. Library students get forum access and docs. Architect tier gets a private channel with me.

Are there certifications?

Yes, completion certificates. Share them on LinkedIn if you want. Honestly, the artifacts you build matter more than any badge.

How often is content updated?

Updates land as the field moves. Free for existing students, forever. Major new modules roughly every quarter.

How long does it take to build a working 3D pipeline?

A first working prototype takes minutes, a production pipeline takes the length of a course. The first mission of the free mission produces an intelligent point cloud in about 15 minutes. A deployable pipeline is a different scale of work: the Operating Systems run to five production modules each, and seven for 3D Deep Learning OS, because deployment rather than the algorithm is where the time goes. 3D Spatial OS is the one that ends in a deployed application.

What accuracy can I realistically expect from an automated pipeline?

No pipeline is more accurate than the scan and the control behind it, so accuracy is set upstream of the automation. Registration error, point spacing, sensor choice and the coordinate reference system decide your ceiling, and a classifier cannot recover what the capture never measured. I teach you to measure that ceiling on your own data instead of quoting a headline figure, because a number from someone else’s site is not a number you can defend to a client. The 3D sensors guide is where that starts.

Community & Support

How can I connect with other learners?

The Spatialetics community is where students share work, ask questions, and hire each other. I’m active there. Program members get study groups and occasional live sessions.

Can I showcase my projects?

Yes. The Use Cases page is built from student work. Submit yours in the community. The best ones become full case studies.

Do you offer corporate training?

Yes. See the Enterprise page. I run intensive sprints, team programs and multi-month capability builds. Email me and we’ll scope it.

How do I stay updated with new content?

Sunday newsletter for the weekly note. Blog for deep dives. Spatialetics community for real-time conversation.

Can you automate point cloud to CAD conversion?

Partly, and the honest split matters: the geometry extraction automates well, the drafting conventions do not. Plane fitting, shape detection, footprint extraction and vectorisation are all scriptable in Python today, and that is the part I teach. Deciding what counts as a wall in your client’s drawing standard still needs a human in the loop. See the vectorisation tutorial, then Point Cloud Intelligence for the full chain.

Do you teach scan to BIM automation?

Yes, as a pipeline you build rather than software you buy. The scan-to-BIM chain is capture, registration, segmentation, geometric fitting, then export to your BIM authoring tool, and I cover the middle of that chain in Python: classifying a scan, detecting planar and structural elements, and turning them into measurable geometry. The indoor shape detection tutorial is the concrete version, and the 3D AI Architect program runs it end to end through to CAD, BIM and GIS export.

How do I automate deliverables from 3D scan data?

You replace the manual steps with a scripted pipeline: ingest the scan, classify it, extract the features the deliverable needs, then export. That is the spine everything here is built on, Capture, Understand, Deliver. The fastest way to see it working is the free mission, and 3D Forge: Build Camp is where you build four working tools end to end, including video to floor plan and point cloud to digital twin.

Book & Resources

What's in the '3D Data Science with Python' book?

3D Data Science with Python, O’Reilly. 18 chapters, 690 pages, 200+ code examples, real case studies. The full chain from raw point cloud to deployed AI system.

Is the book included with course purchases?

The book stands on its own. It pairs well with the Program, but you don’t need the Program to get full value out of it.

Can I access the book digitally?

PDF, EPUB, and paper. Whatever fits how you read.

How is the Library different from the courses?

The Library is a subscription to the whole catalogue, while a single course is a one-time purchase of one. Every standalone course I publish stays open for as long as your subscription runs, and it comes with the complete code repository for each course, the video and written lessons, a commercial licence for all the code, and community forum access. It runs on a three-month minimum and you cancel when you want. Buy a single course instead and it is yours for life with one payment. The four Operating Systems sit outside the Library and are bought on their own. Full details on the Library page.

Your Journey