The second guiding scenario plays out at the other end of the scale. No dig, no time pressure, no “no second attempt” — but a retailer with two hundred articles in her web shop and a photo studio with full control. Where the archaeologist works under the hardest conditions, here you work under the best. This part addresses both sides of the commission: shop operators who want to understand what is possible — and photographers who supply shops and find their new offering here.
The product page with six photos from six angles was always a compromise: it shows six opinions about a product, but not the product. The customer wants to turn it, tip it, step closer — the moves with which they decide in the shop within seconds. A twin gives them exactly that back.
And there is a second reason that weighs more heavily month by month: the net is filling with AI-generated product images — flawless, arbitrary, and ever more often without reference to a real object. In this flood the genuine becomes a value proposition. A digital twin is computed from photographs of an actually existing piece — its all-round consistency cannot be prompted into being. Exactly the technology that competes with the product photographer for the still-image commissions delivers him the antidote: an AI image can show anything — a twin can prove it.
AI image
Read the six needs from Part 3 once more with studio glasses on: parallax? You determine every position. Overlap? No one is rushing you. Sharpness? Continuous light and tripod as you please. Consistency? Nothing changes that you do not change. Light? Your softboxes, your light tent, your decision. The studio is the only place where all six needs are fully satisfiable — and whoever masters product photography already masters light, setup and repeatability. The path from product photo to product twin is not a change of profession. It is an extension by exactly the knowledge you are reading right now.
The product stands raised on a structured, matte underlay — and you move, not the product (the turntable fallacy from 5.7 applies in the studio just the same; whoever needs the turntable for working reasons creates the professional conditions named there: structureless background for cutting out, absolutely even all-round light). The standard run is the one from Part 3.2: one circuit in 10–15-degree steps at product height, a second from diagonally above, and for complex forms a third from slightly below. Light: diffuse and all around — the light tent is the workhorse for small to medium products, softboxes for anything larger. Camera per the checklist from Part 6, mode M, and the distance stays constant.
AI image
Not every category of goods is equally good-natured — and this honesty is your ally in the client conversation, for it distinguishes the professional from the over-promiser. The quick overview, with the reasons from Parts 3 and 7:
AI image
The difference between one twin and two hundred is not diligence but a system: the setup stays standing. Light, distance, plinth, camera settings — set up once, sent once through the import on a sample article (Motion Guard dress rehearsal, 6.7), then everything freezes. From then on every article is the same rhythm: place, circuit, upper ring, article number as the first image (the same principle as the find tag in 8.2 — the attribution travels in the series), clear away, next. One folder per article, named after the article number. The image check runs automatically on every import — it is your quality assurance in the batch, and what it reports you decide on screen rather than the next day at the rebuilt set.
The setup stands once — then the same four-step rhythm runs for every article: place, circuit, number, clear away.
Two paths lead from the finished twin to the product page, and they do not exclude one another. The first: the GLB export — the standard 3D format that common shop systems and 3D viewers display directly; the product then turns where today the image carousel stands. The second, which can do more: embed the Orbit presentation — with everything Part 8.5 described, plus two hotspot types that turn the showpiece into a salesperson: “Website” links out of the model (to the size chart, the care instructions, the accessories), and “Order” leads directly into the shopping cart. The customer turns the product, taps on the detail that convinced them — and buys on the spot.
What holds for the sherd in the museum (8.5) holds doubly for the shoe in the shop, for here findability decides revenue: the export carries its product information machine-readably within it — structured data, of the kind search engines reward. And the reckoning reaches further: more and more purchase decisions no longer begin in a list of results but in the answer of an AI assistant. A product that explains itself — what it is, what it costs, where it stands — is visible to this new layer of the net. Most shops have never heard of it. Yours has.
Here, deliberately, there are no invented percentages — only effects that follow logically from what has been shown and that you can justify in the client conversation: whoever has inspected a product from all sides orders the wrong thing less often — and what is ordered wrong less often comes back less often. Whoever can see authenticity trusts — especially next to competitors with generic render images. And a twin is a multi-purpose tool: the same file serves the product page, the trade-fair terminal (offline, 8.6), the in-store consultation and every future campaign. Captured once, used many times — that is the argument that carries the investment.
The business calculation at the end, because it carries the livelihood: DFX SplatCore runs on your hardware. No fee per model, no subscription per month, no cost per view. The investment is one-time (computer, light, time to learn — this work is the learning part), after which the margin scales with the unit count: the two-hundredth article costs only your working time at the well-rehearsed set. Add to that the argument that opens doors with brand and industrial clients: the product data does not leave the house. No upload of unreleased collections into someone else's cloud, no dependency on a service that changes its terms tomorrow. Your livelihood stands on your ground.
Two cost curves, one difference: the subscription line rises with every article, your own hardware stays almost flat after the one-time investment.
