Better Evidence for Disputed Insurance Claims: Before-and-After in 3D
Most claim disputes turn on two questions: what was there before, and how big the damage really is. Better evidence for disputed insurance claims starts with two dated 3D captures of the same property, aligned so the change shows in place and measured once the scene is scaled.
The ceiling stain is real. Nobody argues with that. The argument is about when it started. The policyholder says the leak came with last month's storm; the insurer's notes say the staining looks older, and the only record of the room before the loss is a listing photo from four years ago. Most evidence for disputed insurance claims fails at exactly this point: it shows the damage, and says almost nothing about what was there before it.
Why claims stall on "what was there before"
A claim gets disputed on two questions far more often than on anything else. Was this damage already there, or was it caused by the insured event? And if it was caused by the event, how much of it is there? The first is the pre-existing damage argument: wear and tear, gradual deterioration, an old repair that has failed. The second is the scope argument, where both sides agree on the cause and then disagree by a factor of two on the area.
The Financial Ombudsman Service publishes the shape of this. In August 2024 it reported 2,001 buildings insurance complaints in a single quarter, a ten-year high, with claim decline the cause of 41% of the complaints it looked at and claim value another 8%. Two years on, its quarterly data for April to June 2026 still showed 1,766 new buildings insurance complaints, a third of them upheld. Behind each one is usually a file that could not show, to everyone's satisfaction, what the property looked like on a particular date.
Photographs are the usual record, and they answer only the questions the person holding the phone thought to ask. A close-up of a crack does not show the wall around it. A wide shot does not show the crack. Three months later, at reinspection or review, the person who needs the answer is rarely the person who took the photos.
What a dated before-and-after 3D record changes
Instead of a set of chosen frames, a capture of the property becomes a Gaussian Splat scene: continuous, photoreal surfaces reconstructed from ordinary video, photos, 360 footage or a LiDAR scan, opened in a browser. A reviewer moves the camera to whatever angle answers the question in front of them, not just the angles someone anticipated on the day.
On its own, that is a better condition record. The change comes from having two of them. A capture taken before the loss and a capture taken after it, each dated, can be laid onto one shared coordinate frame. Then the comparison is not two people holding up two photos from different positions. It is one scene, one camera, and a seam you drag across the room to see which surfaces changed and which were already like that. That pairing is what turns a condition record into evidence for disputed insurance claims, rather than one more picture of the damage.
Scale the scene against something of known size and the second question gets an answer too. The damaged area becomes a number read off the capture rather than an estimate argued over the phone.
The workflow
None of this needs specialist hardware. A phone that records steady video covers most rooms, roofs seen from the ground, and building exteriors. A 360 camera is quicker in tight spaces such as loft voids and plant rooms, and firms that already hold LiDAR scans in E57, LAS, LAZ or PLY can use those.
Capture the baseline
The "before" capture is the hard part, because it has to exist before anyone knows it is needed. It is also the half of the record that a photo set almost never supplies. The practical moments to take one are the ones a claims or underwriting team already controls: a risk survey at inception or renewal, a pre-works condition survey before a contractor starts next door, the start of a tenancy, or the first visit after a loss, which becomes the baseline for any later claim reinspection. Walk the space slowly, cover every wall and the ceiling, and include something whose true size you know, such as a standard door, a tape laid on the floor or a printed marker.
Capture after the loss
Repeat the walk on the same route, at a similar pace, ideally with the same reference object in shot. Video goes up as an ordinary MP4 or MOV and processes in the cloud in 10 to 120 minutes, with no software installed on the adjuster's laptop. The output is a scene in .spz, .ply, .splat or .ksplat that opens on a phone, tablet or desktop from a link.
A capture only records what was in front of the camera. A wall skipped on the baseline walk cannot be compared later, however good the reconstruction looks. Cover the whole room, not just the area that looks damaged today.
Align the two captures in Multi Splat
Multi Splat is the free browser tool that does the before and after comparison. Drop both captures onto the page. In Align mode, pick the baseline as the reference, then click the same physical feature on each capture to place a matched pair of control points: a window corner, a socket, the edge of a door frame. Three pairs is the minimum, and more pairs spread widely across the room give a tighter fit. Press Align and the tool solves for the rotation, translation and uniform scale that bring the second capture onto the first. Points can be nudged or deleted, and every edit can be undone.
Then switch to Compare. The two captures sit either side of a seam you drag across the viewport, sharing one camera so the views cannot drift apart. Sweep it over the stained ceiling and the question of whether the mark was there before is answered by looking, in place, from any angle.
Files dropped into Multi Splat are read in your own browser and are not uploaded. Aligned captures can be exported back out as .spz files with the new position baked in, so the alignment travels with the file.
Measure the changed area
In the editor, set the scene's scale from the reference object in the capture. Once it is scaled, click points directly on the surface. Two points give a length; three or more trace an outline and report its area and perimeter. Trace the stained section of ceiling, the cracked run of render or the missing tiles, and the scope stops being a description.
Measure the same feature in both captures if it was partly there before. The difference between the two outlines is the part of the damage the event can reasonably be linked to.
The measurement is only as good as the reference used to scale the scene and the quality of the capture. No published tolerance applies, so record which reference you scaled from in the file.
Pin notes with hotspots
Hotspots placed in the scene carry text, an image, a video or a link. Put one on each finding: the stain, the tide mark, the previous repair, the point where the scope was measured. Attach the contractor's photo, the moisture reading or the policy clause it relates to. Anyone opening the scene later sees the note where it applies instead of hunting for "photo 14" in an appendix. Organised this way, the scene becomes evidence for disputed insurance claims that a reviewer who never visited can follow on their own.
Render the timelapse to MP4 for the file note
Multi Splat's 4D Timelapse mode plays the aligned captures in date order as one continuous pass, with a cross-fade or a wavefront transition between them and an optional orbit around the scene. Render it to an MP4 (or WebM) with each capture's name burned into the frame, so naming the files by date puts the date on screen. That clip goes into the file note, an email to the policyholder or a review pack, and plays anywhere a video plays.
Is a 3D record good evidence for disputed insurance claims?
It is a fuller record than a photo set, because it holds the whole space on a known date and lets a reviewer check any surface, not only the ones someone photographed. Whether it is accepted in a formal dispute, at an ombudsman or in court is a separate matter. That depends on the jurisdiction and on how the evidence was captured, stored and handled.
So treat it like any other evidence. Keep the source video or scan untouched, record who captured it and when, and note what was done to it afterwards (alignment, scaling, hotspots). The 3D scene is built from that source material; it does not replace it.
| Photo set today | Dated before-and-after 3D record | |
|---|---|---|
| What the "before" shows | Whatever was photographed, if anything was | The whole space on a known date |
| Comparing before and after | Two photos from different positions | One aligned scene with a comparison seam |
| Sizing the damage | Estimate from a close-up, or a return visit with a tape | Area and perimeter traced on a scaled scene |
| Where the notes live | A separate report referencing photo numbers | Hotspots pinned on the finding itself |
| A new question at review | Another site visit | Move the camera in the existing scene |
Objections
We rarely have a "before" capture. True for most claims today, and the honest answer is that this approach does nothing for a loss where no baseline exists. It pays off where the team controls a moment before the loss: surveys, pre-works records, tenancy starts. It also pays off on the claim itself, since the first post-loss capture is the baseline for every reinspection and supplementary claim that follows.
Our adjusters are not 3D specialists. They do not need to be. The capture is a slow walk with a phone. Alignment is clicking the same corner on two scenes three times. Reviewers open a link.
Is the measurement accurate enough to settle quantum? Treat it as a strong check rather than a survey instrument. Its precision depends on the scale reference and the capture, and no tolerance has been published. Where the figure will decide a large payment, confirm it on site; where the dispute is about whether the stain is 2 or 4 square metres, an outline on a scaled scene usually ends the argument.
What happens to the data? Files opened in Multi Splat stay in the browser and are not uploaded. Captures processed for a hosted scene are shared by link, embed or QR code, so access follows who you send it to. Multi Splat is free for personal and research use; commercial claims work falls under a paid plan, listed on the pricing page.
FAQ
Can a before-and-after 3D comparison prove pre-existing damage?
It can show what a surface looked like on two dates, which is often what a pre-existing damage argument turns on. It cannot show how a mark formed between captures, and it says nothing about anything the baseline walk missed. Use it as the visual record alongside moisture readings, expert opinion and the policy wording.
How far apart can the two captures be?
Any interval works, provided enough fixed features survive in both captures to place at least three matched control points. Window corners, door frames and sockets rarely move. A room that was stripped out between captures is harder to align, so pick features that the damage and any works left untouched.
Does the timelapse video replace the 3D scene?
No. The MP4 is a summary for a file note or an email, playable without a link or a browser tab. The scene is where a reviewer checks a particular surface, reads a measurement or opens a hotspot. Send the video for context and the scene link for the detail.
Is this useful for claim reinspection?
Yes, and it may be the most common use. The first post-loss capture becomes the reference, and each later visit is aligned to it. Drying progress, strip-out extent and reinstatement can then be compared in place rather than reconstructed from separate photo sets, which also helps on long-running works.
A disputed stain is a question about two dates. A photo set can only answer it for the angles someone happened to shoot. Two aligned captures can answer it for the whole room, and once the scene is scaled the follow-up question about how much is a number in the file, not a guess.