A few years ago I held a mango in my hand. Nothing special — a fruit like any other in any bowl. But as I turned it, a thought came to me that has never let go of me since: How do I get exactly this thing — with its form, its colour gradients, its little bruises — onto the internet as a digital twin?
The longer I thought about it, the clearer something else became: the internet has been flat since its earliest days. Texts, images, videos — all two-dimensional. Three-dimensional things existed, but they were reserved for the specialists: expensive scanners, 3D programs with years-long learning curves, programming. Anyone who wanted to show a mango, an heirloom or their product in 3D needed what amounted to a degree in computing.
That is what I wanted to change. 3D for everyone — for anyone who can operate a camera. DFX SplatCore grew out of that thought. And the pages you are reading right now are the part of it that belongs entirely to you: the knowledge of how to capture. Because, let me give this much away — the software today can compensate for a surprising amount. But the most important moment remains the instant in which you press the shutter.
Ralf HeitzigStudio Heitzig
AI image
Since the first web pages of the 1990s the net has changed enormously — and in one respect it has stayed the same: it shows the world in two dimensions. The product page shows six photos from six angles, because it cannot show the product, only depict it. The museum puts a photo series online, because the display case does not fit through the cable. We have grown so used to this that we barely notice what gets lost along the way: the turning, the stepping closer, the grasping in the literal sense.
Three-dimensional representation on the net did exist — but every time with a high barrier to entry. Either someone had to rebuild the object on the computer (handwork, hours to weeks), or scan it with special hardware (laser scanners, structured-light scanners — devices beyond the price bracket of a photo studio), and in both cases publishing it required programming skills. These three thresholds — modelling, scanning, programming — kept 3D a specialist matter for decades.
Two developments recently tore these thresholds down. First: ordinary photos can today be turned into photorealistic 3D representations — the process behind it (Gaussian Splatting, whose story Part 2 tells) freed the capture side from special hardware. The camera you already own is enough. Second: publishing no longer requires programming — which DFX SplatCore handles with its Orbit module. What remains is exactly one skill that no software can replace: capturing well. It is precisely to this skill that the following chapters are devoted.
Many people meet everything called “AI” with a certain unease — and this unease is more understandable than the tech industry likes to admit. It arises not from technology as such, but from technology that one is not allowed to see through: a black box that somehow produces results without anyone explaining how and why. No advertising promise helps against that. Only one thing does: understanding.
That is why you get four commitments here, against which you may hold every chapter:
Once you have understood that behind 3D reconstruction lies two-thousand-year-old surveying geometry (Part 1), that the image check is a verifiable calculation and not an oracle, and that the limits of the process have known physical reasons — you are no longer afraid of this technology. You use it.
For photographers — and quite especially for those who are right now watching AI-generated product images compete for their commissions. To them one thought is offered, one that runs through every chapter: the threat and the opportunity come from the same direction. An AI image generator can invent an arbitrarily pretty product photo — but it cannot produce a real, rotatable, verifiable image of an actual object. That is exactly what the digital twin is. Anyone who can photograph, knows clients and masters light already brings almost everything to this new craft. The tool that threatens becomes the tool that sustains.
And for everyone else with a camera — the collector who wants to document her pieces; the craftsman who wants to show what he builds; the museum team with no digitisation budget; the retailer who wants to set his shop apart from the competition. Basic photographic knowledge is assumed (you know what aperture and shutter speed are) — but expressly no 3D knowledge and no programming.
A principle of DFX SplatCore that you will meet often, because it has practical consequences: everything runs on your own computer. No subscription, no fee per model, no cloud into which your images have to be uploaded.
This is more than a question of cost. Anyone who works for industrial clients under confidentiality, who digitises museum holdings, who documents evidence — cannot simply hand their material over to someone else's data centre. And anyone building a livelihood should not build it on the ground of a service that can change its prices or be shut down. A finished digital twin from DFX SplatCore is a collection of files that belongs to you — it runs offline, at a trade fair, in a museum, on a tiny computer, and it does not vanish when some provider vanishes.
The structure follows one idea: first understand, then do. But no one has to read from front to back:
About this edition You are reading the complete first edition: all eleven parts and the appendices. The work continues to be maintained — a dedicated deep dive into the walkable space (Part 7.6) has already been announced. The guide to image checking (Motion Guard) exists as its own manual in this library; Appendix C points the way.
