Studio Matrx Monthly · Volume 1 · Issue 4 · September 2026
Amogh N P
 In loving memory of Amogh N P — Architect · Designer · Visionary 
Design & CoordinationLesson 8.2
Reality Capture & Scan-to-BIM/Module 8 · Applications Across the Lifecycle

Lesson 8.2 · Applications Across the Lifecycle

Design & Coordination

Once you hold an accurate record of what is really there, you can design against real geometry, coordinate new work with the existing inside one model, detect clashes against the as-built before anything is built, and even manufacture parts to fit a building you have never guessed at

12 min Interactive lessonFree · open lessonByAmogh N P· Architect & interior designer
The hook

Capturing the building is only half the value. The other half is designing with it -- against real geometry, coordinated in one model, clash-checked before a single part is cut.

An accurate scan answers the question 'what is here?'. Design asks the harder one that follows: 'given exactly what is here, will the new thing fit?' For a long time that question was answered on site, the most expensive place to discover the answer is no. New steelwork fabricated to the drawing would arrive and foul a wall that was 60 millimetres out. A beautifully detailed ceiling would clash with a beam nobody had recorded. The fit-out would fight the floor's slope. Every one of these is the same failure in a different costume: new work designed against an assumed building rather than the real one.

Reality capture breaks the pattern by turning the captured existing conditions into the basis you actively design against. The point cloud, or the scan-to-BIM model built from it, sits inside your design environment as measured ground truth. You set out new elements to real geometry, coordinate the new with the existing in one federated model, and run clash detection against the as-built so conflicts surface on screen, weeks early and free to fix. And because the whole team shares that one accurate basis -- and because you now know the real dimensions in advance -- you can even manufacture parts off-site to slot into an existing building, which is impossible when the building is a guess.

One measured source of truth beats five confident guesses. Coordinate against the building that exists -- then you can even build parts for it off-site.

Designing against reality

Setting out new work against real geometry, not an idealised box

The first shift reality capture brings to design is psychological as much as technical: you stop drawing against an idealised, orthogonal fiction and start drawing against the building as it actually is. Bring the point cloud into your modelling environment and the real geometry is simply *there* -- the wall that leans 50 millimetres over its height, the floor that falls 30 millimetres across the room, the column that is not quite where the grid says, the soffit that steps where an old alteration left it. You set out new elements to that reality. Joinery is scribed to the wall it will actually meet; a new floor build-up is designed around the real levels; a partition lands where it will genuinely clear the existing beam.

This sounds obvious, yet its absence is the source of a huge share of site problems. When you design against a drawn rectangle and the building is a slightly skewed parallelogram, the gap between the two becomes the contractor's problem, discovered late and resolved expensively by cutting, packing, re-ordering or improvising. When you design against the captured reality, that gap is closed on screen, deliberately, by someone who can see the consequences. The cost of accommodating a lean is trivial at the model stage and punishing at the installation stage, and the only thing that moves it from late to early is having the real geometry in front of you while you design.

There is a discipline to doing this honestly. The captured geometry is a measurement with a stated accuracy and with gaps, so you design knowing roughly how much the cloud can be trusted and where occlusion has left voids you must verify by other means before committing. You also design against a moment in time -- the clutter, the furniture, the temporary works captured alongside the permanent fabric -- so part of the skill is reading which points represent the building you are keeping and which are noise. Used with that judgement, designing against captured reality is the single most effective way to kill the 'it did not fit on site' failure before it is born. Used naively -- treating the cloud as perfect and complete -- it simply moves the surprise to a different place. The cloud is ground truth only to the accuracy it was captured at, and never beyond its occlusion shadows.

DESIGN AGAINST THE REAL WALL, NOT THE IDEAL ONEassumedplumbreal captured wall~50 mm leannew joineryscribed to fitOne shared basisarchitect +engineer + fabricatormodel the sameas-built geometry
Zoom
New work set out against the real wall: the captured cloud shows the wall leaning ~50 mm off the assumed plumb line, so the joinery is scribed to fit reality and every discipline models against the same as-built geometry -- one shared, measured basis.

Design to the wall that exists, not the wall you wish existed. A 50 mm lean is free to solve on screen and brutal to solve on site.

Coordinating new with existing in one federated model

Design on a real project is a team sport, and its hardest moments are the seams where disciplines meet: where the architect's form, the engineer's structure and the services engineer's ducts, pipes and trays all have to occupy the same real space inside a building that already exists. Reality capture transforms this coordination by giving every discipline the same accurate basis. The scan-to-BIM model, or the point cloud itself, becomes the shared representation of the existing building, and each discipline models its new work against it in a federated model -- the combined model in which everyone's contributions are brought together and checked for conflict.

The power here is a single source of truth. Without capture, each discipline works from its own interpretation of old drawings and site notes, and the discrepancies between those interpretations become clashes that only reveal themselves when the trades arrive. With a captured as-built as the common reference, the architect routes the design around the real structure, the structural engineer sizes new interventions against the frame as it genuinely is, and the MEP engineer threads services through the space that is actually available -- all coordinated against one measured reality rather than several guesses. The federated model becomes a faithful composite of real existing conditions plus proposed new work, which is exactly what good coordination needs.

This is why reality capture and BIM are such natural partners, and the pairing is especially potent in existing buildings, where the cost of getting coordination wrong is highest and the information is otherwise weakest. It is worth repeating the honest caveat from the scan-to-BIM modules: building the existing-conditions model is still largely skilled manual work, and it is modelled to a chosen level of accuracy and detail -- so the coordination is only as trustworthy as the model's fidelity and the cloud's accuracy behind it. A wall modelled as a clean plane when the real one bows will coordinate beautifully in the model and fight you on site. Good practice keeps the point cloud available alongside the model precisely so the team can check the model against the raw evidence where it matters, and records what accuracy and level of detail the existing model was built to, so nobody over-trusts it.

Clash detection

Clash detection against the as-built -- catching conflicts on screen, not on site

The sharpest coordination tool reality capture unlocks is clash detection against the as-built. In a federated model you can automatically, or by eye, find where new work conflicts with the existing building -- where a new duct runs into a down-stand beam, where a riser collides with a slab, where a new partition fouls an existing column. The decisive point is *what* the new work is being checked against: not an idealised or out-of-date model of the existing building, but a representation derived from the captured reality. You are testing the design against the building as it genuinely is.

The value is almost embarrassingly simple to state. A clash caught in the model costs a model edit -- move the duct, resize the opening, reroute the run -- and a little of a coordinator's time. The same clash discovered on site costs rework: demolition, re-fabrication, delay, disruption to other trades, and sometimes a compromise that damages the design. Clashes with *existing* structure and services are among the most common and most avoidable, because they arise precisely from not knowing what was there -- and capture is the cure for not knowing. Running the new design against a captured as-built systematically converts expensive site surprises into cheap screen decisions.

Two honest qualifications keep this from becoming over-claimed. First, clash detection is only as good as the as-built it checks against: if occlusion left a service unrecorded, or the model simplified a feature away, the clash will not be caught -- so the quality, accuracy and completeness of the capture directly bound the reliability of the clash check, and the team must know where the blind spots are. Second, a detected 'clash' is a prompt for judgement, not an automatic verdict; tolerances, what counts as a real conflict versus acceptable proximity, and how to resolve it are design decisions. Used well -- against a well-planned capture, with the point cloud on hand to verify, and with the limits understood -- clash detection against the as-built is one of the highest-return uses of reality capture in the whole design process. It is also the bridge to the next idea: if you trust the geometry enough to clash-check against it, you may trust it enough to manufacture against it.

CLASH DETECTION AGAINST THE CAPTURED AS-BUILTexisting slab soffit (from point cloud)down-stand beam nobody knew was therenew duct (design)CLASHCaught on screen, for the price of a model edit --- not on site, for the price of rework.The cloud is the honest referee: new work is tested against the building as it IS,so prefabricated parts can be made to fit the first time.
Zoom
Clash detection against the captured as-built: a new duct meets a down-stand beam that only the point cloud revealed, flagged on screen as a clash -- caught for the price of a model edit rather than site rework, and letting prefabricated parts be made to fit the first time.

Clash on screen = a model edit. Clash on site = rework. Capture moves the surprise to the cheap side of the fence.

One trustworthy basis -- and the link to prefabrication

Pull the threads together and the real prize of capture-driven design emerges: one trustworthy, measured basis shared by the whole team. The architect, the engineers, the contractor and the fabricators all work from the same accurate representation of the existing building, so decisions compound instead of conflicting, and the disputes that usually erupt from differing assumptions about 'what is there' simply do not arise, because there is a measured answer everyone can see. This single-source-of-truth quality is the quiet foundation under every other benefit in this lesson; coordination, clash detection and confident design all flow from the team no longer arguing about reality but reading it.

That trust is what makes the most demanding modern method possible in existing buildings: prefabrication and off-site construction. Manufacturing a component -- a facade unit, a bathroom pod, a riser module, a run of plant -- away from site and delivering it to slot precisely into place is transformative for speed, quality and waste. But it is also unforgiving: the factory cannot improvise, so the real dimensions it manufactures to must be known *in advance* and must be *right*. You cannot safely fabricate a module to fit an existing structure you have only estimated. Reality capture supplies exactly the ground truth prefab depends on -- the accurate as-built geometry that lets a part be made to fit the first time. Capture, BIM and prefabrication form a chain: capture the reality, coordinate the design against it, manufacture to the captured dimensions.

The discipline, as ever, is honesty about accuracy. Prefabrication against a captured building raises the stakes on the capture: the accuracy must genuinely be good enough for the tolerances the manufactured part demands, the critical interfaces may warrant targeted verification rather than trust in the general scan, and where the fit is critical and the cost of error high, the capture accuracy at those interfaces is a matter for verified methods and, if it must be binding, a qualified surveyor. But the principle is powerful and clear: the better and more trustworthy your knowledge of what exists, the more of modern, model-based, manufactured construction you can safely bring to bear on an existing building. Capture turns the existing building from the riskiest unknown in the project into the shared, measured foundation the whole team builds on.

Verify-this: coordinate against the capture, respect its limits

Federated model / single source of truth

One combined model all disciplines design against

Captured existing conditions plus proposed new work, coordinated together. Only as trustworthy as the model's level of detail and the cloud's accuracy. Modules 6, 7.

Clash detection against the as-built

Finding new-vs-existing conflicts before construction

Catches conflicts on screen, not on site -- but bounded by capture accuracy, completeness and occlusion. A detected clash is a prompt for judgement, not an automatic verdict.

Level of accuracy / level of detail (LOA/LOD)

How faithfully the existing model matches reality

Record what the existing model was built to so nobody over-trusts it; keep the point cloud available to verify. Module 6.3; binding accuracy follows verified specs and a surveyor.

Prefabrication / off-site construction

Manufacturing parts to fit a captured existing building

Needs real dimensions known in advance and right; verify critical interface accuracy specifically. Binding measurement for critical fit defers to verified methods and a licensed surveyor.

Hands-on workshop

Workshop -- plan the coordination of a new intervention against a captured existing building

Reality capture earns its keep in design when the new work is coordinated against the real building. In this workshop you will take a real or imagined intervention into an existing building and plan how you would design and clash-check it against captured reality.

A building and an intervention you can reason about, plus a notebook. No software required -- this is about coordination thinking; the tools come in Module 7.

Given & goal
Goal: a one-page coordination plan for new work against a captured as-built
Inputs: an existing building + a plausible intervention (a new mezzanine, a services run, a fit-out, a facade) + this lesson
Time: ~45 minutes
  1. 1Describe the intervention and its critical interfaces: what new work is going in, and where must it meet the existing building precisely (the soffit it hangs from, the wall it scribes to, the structure it penetrates)?
  2. 2State what you would design against: the point cloud directly, a scan-to-BIM model, or both -- and at what level of accuracy and detail, matched to how tight the fit needs to be.
  3. 3Plan the clash checks: list the specific new-vs-existing clashes you would test for (duct vs beam, riser vs slab, partition vs column) and note which existing elements might be hidden by occlusion and therefore missed.
  4. 4Identify where you would verify against the raw cloud: pick the one or two interfaces tight enough that you would check the model against the original points rather than trusting the modelled geometry.
  5. 5Decide the prefab question and the boundary: could any part be prefabricated against the captured geometry? If so, where would you verify the interface accuracy, and where would a binding measurement need a licensed surveyor?

You’ll walk away with
A one-page coordination plan: the intervention and its critical interfaces, what you design against and to what level of detail, the specific clash checks and the likely occlusion blind spots, the interfaces you would verify against the raw cloud, and the point at which manufactured fit needs verified or binding measurement. Keep it alongside your 8.1 survey plan.

The worked example

Three altitudes on the same idea

Read the band that fits you — or all three.

For the architectCapturing sites and buildings as the reliable basis for design

Captured reality is the basis you design and coordinate against, and it is what makes confident work in existing buildings possible. Bring the point cloud or scan-to-BIM model into your environment and set out new work to real geometry -- leans, sags and all -- then coordinate the new with the existing in one federated model and clash-check against the as-built so conflicts surface on screen, not on site. Keep the cloud alongside the model to verify where it matters, and record the accuracy and level of detail the existing model was built to so nobody over-trusts it. This single trustworthy basis is also what lets you push prefabrication into retrofit; where manufactured fit is critical, verify the interface accuracy and defer any binding measurement to a surveyor.

For the interior designerAccurate existing interiors, as-builts and fit-out verification

Coordinating a fit-out with an existing interior is where captured reality pays you back every time. Designing against a scan means joinery is scribed to the wall that really leans, the floor build-up suits the real levels, and a new partition lands where it genuinely clears the existing beam and services -- so the install fits the first time. Share that one accurate basis with the contractor and trades so everyone works from the same measured interior instead of conflicting notes, and clash-check bulky elements (islands, ceilings, ductwork) against the as-built before fabrication. For manufactured or precisely fitted elements, verify the critical dimensions rather than trusting the general scan, and call in a surveyor when a dimension must be binding.

For the studentHow the real world becomes measured 3D data and models

Understand that capture is not only a record -- it is the geometry you design and coordinate against. Learn the chain: design new work to real captured geometry; combine disciplines in a federated model; run clash detection against the as-built so conflicts are caught on screen; and share one trustworthy basis that makes prefabrication against an existing building possible. Grasp the honest limits too -- clash detection is only as good as the capture's accuracy and completeness, a simplified model can hide a real bow, and occlusion can leave a service unrecorded. Knowing why 'one measured source of truth' beats several guesses, and why prefab needs real dimensions in advance, is exactly the reasoning modern BIM-led practice hires for.

Misconception check

Once we have a scan-to-BIM model of the existing building, coordination and clash detection are basically automatic and reliable -- the model has the real geometry, so if the software finds no clashes, the new work will fit.

A clean clash report is only as trustworthy as the capture and the model behind it, and treating it as a guarantee is how teams get surprised anyway. First, the existing model is built from the captured reality to a chosen level of accuracy and detail, and it is largely skilled manual work: a bowing wall modelled as a flat plane, or a feature simplified away, will coordinate perfectly in the model and clash on site. Second, capture has occlusion -- a service hidden above a ceiling or behind a chase that the scanner never saw cannot appear in the model, so a clash with it cannot be detected. Third, the cloud has a stated accuracy, and if the new work is set out to tolerances tighter than the capture supports, 'no clash in the model' means little. Fourth, a detected clash is a prompt for human judgement about tolerance and resolution, not an automatic verdict. Good practice therefore keeps the raw point cloud available alongside the model to verify critical areas, records the accuracy and level of detail the model was built to, knows where the occlusion blind spots are, and -- where a manufactured part must fit precisely -- verifies the interface accuracy specifically rather than trusting the general scan. Coordination against captured reality is enormously powerful; it is powerful precisely because the team understands its limits, not because it is magic.
Try it

Do it yourself

No software needed -- reason it through.

  1. 1Explain what it means to design against real captured geometry, and why it prevents the common 'it did not fit on site' failure.
  2. 2What is a federated model, and why is a single measured source of truth better than each discipline working from its own interpretation?
  3. 3How does clash detection against the as-built save money, and what two limits bound its reliability?
  4. 4Why does prefabrication for an existing building depend on reality capture? What must be true about the captured dimensions?
  5. 5Give one case where you would verify against the raw point cloud rather than trust the scan-to-BIM model, and say why.
Take this with you

The one line to carry out

Captured reality is the measured basis you design against: set out new work to real geometry, coordinate the new with the existing in one federated model, clash-check against the as-built to turn expensive site surprises into cheap screen edits, and give the whole team one trustworthy source of truth that even lets you prefabricate against the real building -- always bounded by the capture's accuracy, completeness and occlusion, and with binding or critical-fit measurement deferred to verified methods and a licensed surveyor.
Take it further
References & further reading

Peer-reviewed journals & authoritative standards

  1. 01Building information modelingWikipedia -- Building information modeling, 2026.
  2. 02Point cloudWikipedia -- Point cloud, 2026.
  3. 03As-built drawingWikipedia -- As-built drawing, 2026.
  4. 04ConstructionWikipedia -- Construction, 2026.
Related lessons
Recap
Reality capture's value in design is that the captured existing conditions become the basis you actively design against. You set out new work to real geometry -- the leaning wall, the sagging floor, the column that is not on the grid -- so the gap between design and reality is closed deliberately on screen rather than expensively on site. You coordinate new with existing in one federated model, giving every discipline the same measured source of truth instead of several conflicting interpretations of old drawings. You run clash detection against the as-built, turning the most common and avoidable conflicts -- new services and structure fouling existing beams, slabs and columns -- into cheap model edits caught weeks early. And because the team now shares trustworthy real dimensions, you can prefabricate parts to slot into an existing building, which is impossible when the building is a guess. Throughout, the power is bounded by honesty: clash detection is only as good as the capture's accuracy and completeness, occlusion can hide a service, a simplified model can mask a real bow, so the team keeps the cloud on hand to verify, records the level of accuracy and detail, and defers binding or critical-fit measurement to verified methods and a licensed surveyor.
Carry forward →

Designing against captured reality assumes the capture is trustworthy -- but construction is where design meets the messiness of being built. Next we turn capture onto the building site itself: scanning during construction to verify the as-built against the design, catch deviations early, and monitor progress with measured evidence.

A

The author

Amogh N P

Architect, interior designer, and creative polymath. Studio Matrx began in his notebooks — his vision of design made honest, useful, and open to everyone. Its Academy is written and taught in his memory, and free, forever.

More about Amogh →