Turning Survey Scan Data Into a Client-Viewable 3D Scene

A crew spends a day on site with a terrestrial scanner and comes back with a point cloud accurate to a few millimetres. Three days later the client wants a PDF instead, because nothing on their laptop opens the file. That gap, turning survey scan data into a client-viewable 3D scene rather than handing over the raw registration, is where most of this actually goes wrong.

The scan was fine. The handoff wasn't.

The point cloud did its job. It's correctly registered, dimensionally sound, and holds everything a licensed surveyor needs to pull a measurement or check a clash six months from now. None of that is the problem. The problem is that reading it requires viewer software, a license, and enough practice navigating a cloud of individually plotted points to not get lost inside it, and the person who needs to look at it is a property manager, an insurance adjuster, or a general contractor pricing a renovation, not a survey technician.

So the deliverable that actually gets used isn't the scan. It's whatever got flattened out of it: a marked-up PDF that answers this week's question and nothing else, or a call where the surveyor shares their own screen and drives through the cloud on the client's behalf. Three months later a change order dispute comes up that the original scan could settle in thirty seconds, except nobody on the client side can open it, so someone drives back out to look at the wall in person.

The people this actually affects aren't hypothetical. An insurance adjuster reviewing storm damage needs to see the extent of it without scheduling a second inspection. A general contractor pricing a renovation against an existing building needs to check a clearance before the bid goes out, not after. A property manager signing off a tenant buildout needs to confirm what's actually there against what was promised. None of them are the audience E57 and LAS files for client review were ever really built for, and none of them are going to install point cloud software to check one dimension.

What turning survey scan data into a client-viewable 3D scene actually changes

Nothing about the scan itself changes. The E57 or LAS file stays exactly what it was: the survey-grade record a licensed professional pulls dimensions and clash checks from. What changes is that the same data gets reconstructed a second way, as a Gaussian Splat scene, so instead of a sparse field of measured points the surfaces read as continuous and photoreal. A point cloud stays a cloud of measured points, accurate but sparse and hard to read for anyone outside the survey team. Reconstructed as a splat scene, the same data reads as continuous surfaces, so a client sees the space rather than a diagram of it.

This isn't a new capture, and nothing about the scan itself changes to produce it. What comes out the other side is a walkable 3D scene from a point cloud that was already sitting on a drive, generated once and handed to people who were never going to open the point cloud file in the first place.

What it demands from the crew is nothing beyond the registration they already deliver. The reconstruction works from the measured points and whatever colour or imagery the scanner already captured alongside them, the same registered output that would otherwise go straight into CAD. Nobody needs to shoot extra photos or rerun the scan at a different setting to make the scene look right.

The workflow

What's already on the drive

If the firm is scanning with a terrestrial or mobile LiDAR unit, the raw material already exists: registered point clouds in E57, LAS, LAZ, or PLY. No new fieldwork, no second visit, no different scanner settings. The files a crew already delivers as part of a normal survey are the input.

Sending the data

Turning that into a viewable scene isn't a self-serve upload-and-wait tool the firm runs itself. It's a managed LiDAR to 3D service: a firm sends the scan data it already holds and gets back a hosted scene, without standing up its own reconstruction pipeline or licensing a new piece of software in-house. That matters because the alternative for most firms is a fragmented toolchain: one program to clean the registration, another to mesh it, a third to render something presentable, each with its own license and its own learning curve. Sending the file skips that stack entirely. Firms scanning on a regular schedule can also push files through the reconstruction API instead of a manual upload each time.

Processing

Reconstruction runs in the cloud, typically 10 to 120 minutes, with no GPU or install required on the firm's end. A single room sits toward the fast end of that range; a full building or an exterior site sits toward the slow end.

What comes back

A scene in a browser-native format such as .spz, .ply, .splat, or .ksplat, opening on a phone, tablet, desktop, or VR headset with nothing to install. For firms that want their own name on it rather than a third party's, the viewer and editor can be white labelled: the firm's brand, colours, and domain instead of a generic link.

Getting it in front of the client

Delivery is a link, an embed on a project page, or a QR code printed on the cover of a report. Inside the editor, a guided camera path keeps a first-time viewer from getting lost, and hotspots carrying text, an image, a video, or a link can point directly at the crack, the clearance, or the dimension note that the visit was actually about.

Who actually opens it

The link goes to whoever asked for the visit in the first place: an adjuster who needs to see the water line on a wall without another trip out, a contractor confirming a duct run before locking in a bid, a property manager walking a space during a lease negotiation from a different building entirely. None of them touch the point cloud. They open a browser, follow the camera path if one's been set, and click through the hotspots for the specific things the visit was actually about. It's a 3D deliverable for non-technical clients built on top of a technical one, not instead of it.

Objections

Is this still survey grade? No, and it isn't meant to be. The reconstructed scene is what a client looks at. Extraction, distances, and CAD work still come from the original E57 or LAS file, the same as before this existed. Nothing about where the measurements live has changed, and nobody signs off a number pulled from the scene instead of the scan.

Who on the team has to learn something new? Nobody, on either side. The crew scans and registers exactly as it always has. On the client's end there's a link to open in a browser, not a piece of software to install and learn.

What happens when the site changes before the client gets around to opening it? A construction site rarely holds still. For that, Multi Splat aligns a follow-up capture against the first one and renders the difference as a 4D timelapse, so the answer to "what changed since the last visit" is a rendered comparison instead of two people arguing over two separate PDFs.

Does this replace what we already deliver? No. It sits alongside the CAD file, the point cloud export, and the written report, as the one piece of the package a client with no survey background can actually open without help.

Laid side by side against the way a point cloud handoff usually goes today, the difference isn't in what the survey produced. It's in what the person on the other end can do with it.

FAQ

Can a client see the scene without installing anything?

Yes. It opens in a browser on a phone, tablet, desktop, or VR headset, with nothing to install on the viewer's device. The link, the embed, or the QR code is the entire delivery mechanism, and nothing else is required on the client's side.

Does converting a point cloud to a splat scene change the survey data?

No. The original E57 or LAS file stays intact as the record used for measurement and CAD extraction. The splat scene is a second, client-facing output built from the same scan, not a replacement for it or a new survey in its own right.

What formats can a firm send in?

E57, LAS, LAZ, and PLY point clouds, plus photos, video, 360 footage, or drone footage if a firm wants to combine sources for a richer scene. Whatever comes in, the output opens inside the viewer as .spz, .ply, .splat, or .ksplat.

How long does turnaround actually take once the scan is sent?

Processing itself runs 10 to 120 minutes in the cloud, but turnaround depends on when the file gets queued and reviewed on the way out. A single room or a small commercial space can realistically go out the same day. A full site or a multi-building survey sits toward the longer end, closer to the size of the scan than to any fixed schedule.

The LAS file that's been sitting on the drive since the site visit becomes a link, opened in a browser, by someone who was never going to install a point cloud viewer to check a wall dimension. Somewhere between ten minutes and two hours of unattended processing is the entire distance between a scan the survey team can read and a scene the client actually walks. Nothing about the survey changes to get there. The only thing that moves is who gets to look at it, and how.