LOD 200 vs LOD 300 for Scan-to-BIM
For scan-to-BIM, choose LOD 200 when the team needs approximate existing-conditions geometry for early planning, and LOD 300 when modeled elements must be dimensionally specific enough to design, document, or coordinate against. LOD 200 elements are generalized: size, shape, and location are roughly right but not measured to a tight tolerance. LOD 300 elements carry specific geometry placed at the measured location, so a designer can dimension against the model inside the tolerance the scope states. Most existing-conditions projects do not need one level everywhere. The right answer is usually a mix, set category by category.
The trap is treating LOD like a quality grade where higher is always better. It is not. A higher LOD asks the modeling team to make and verify more decisions, which costs more time and produces a heavier Revit file. If the receiving team will never use that detail, you paid for confidence you will not consume.
What LOD 200 and LOD 300 actually mean
The LOD 100 through 500 definitions originate with the AIA. BIMForum publishes the Level of Development Specification that turns them into element-by-element language, and it added LOD 350 for coordination work. They describe how much you can rely on a modeled element, not how pretty it looks.
| Level | What you can rely on | Typical scan-to-BIM use |
|---|---|---|
| LOD 200 | Generic element. Approximate quantity, size, shape, location, and orientation. Treat dimensions as indicative. | Early renovation planning, space planning, concept design, owner record model, feasibility. |
| LOD 300 | Specific element whose quantity, size, shape, location, and orientation can be measured off the model, positioned against a defined project origin. | Renovation design development, coordination, drawings generated from the model, anywhere fit and alignment matter. |
| LOD 350 | LOD 300 plus the interfaces with other building systems: connections, supports, and the clearances that affect neighboring trades. | Trade coordination and clash detection where elements must align across disciplines. |
For existing conditions, LOD 200 is not an excuse for a sloppy model. It means walls, floors, ceilings, roofs, and major openings are present and organized, just not chased to a tight tolerance on every face. LOD 300 means those same elements are placed where the scan says they are, with geometry a downstream team can dimension against.
LOD describes the model. LOA describes the measurement
Here is the distinction worth getting right before the scope is written. LOD is an AIA/BIMForum concept about element development. Level of Accuracy, or LOA, is a separate USIBD standard, currently version 3.1, about how accurately a point was measured and how accurately the model or drawing derived from it represents that point. USIBD names those two things explicitly: measured accuracy and represented accuracy. It grades them LOA10 through LOA50, each level a defined accuracy band stated at 95 percent confidence. LOD and LOA are different axes, and existing-conditions work needs both.
The USIBD bands run from several centimeters at the loose end down to millimeter level at the top. Where a given project lands depends on the scanner, the registration, the site, and the control available - it is a per-project result stated in the scope, not a property of the service. A tight measured band says the point cloud is trustworthy - and that is a statement about the data. That does not automatically make the Revit model LOD 300. Modeling does not improve accuracy. It translates measured data into a usable representation, and every step of that translation can only add deviation, never remove it. The represented accuracy depends on how far each element is allowed to drift from the scan. A good scope names both: the LOD per category and a represented accuracy tolerance, for example model walls to within a stated tolerance of the point cloud. Without that, LOD 300 becomes a feeling rather than a measurable acceptance criterion. For a closer look at what the LOD 300 label does and does not promise, see what LOD 300 means in BIM.
Do not buy one LOD for the whole model
A building rarely needs the same effort everywhere. An architect doing a renovation needs reliable walls, openings, floor-to-floor heights, and stairs, while loose furniture can be excluded entirely. A facilities team may want LOD 300 on a mechanical room and LOD 200 on typical office areas. A facade project needs exterior detail and almost nothing inside. What a renovation team actually consumes is worked through in existing conditions documentation for renovation.
Scoping by category is what keeps LOD honest. It also keeps the quote honest, because what you pay rides on element count and target LOD, not on square footage. A clearer instruction than “LOD 300 everything” looks like this:
| Category | Example scope |
|---|---|
| Exterior walls and partitions | LOD 300, modeled to measured location |
| Doors and windows | Openings placed; LOD 300 size, generic families unless schedules are needed |
| Floors and levels | Primary planes at LOD 300; set levels from the scan and any control or survey data provided |
| Ceilings | LOD 200, or LOD 300 only where reflected ceiling plans are required |
| Structure | Visible columns and major beams at LOD 300; concealed framing noted as assumed rather than modeled as measured |
| MEP | Excluded, or visible equipment and primary exposed routes only |
| Furniture and small objects | Excluded unless specifically required |
That table is a better deliverable definition than a single label. It tells whoever builds the model exactly where to spend hours and where to stop.
What changes between LOD 200 and LOD 300
Moving a category from LOD 200 to LOD 300 changes how tightly each element is aligned to the cloud, how irregular real conditions get represented, whether generic families are acceptable or must be replaced with specific geometry, and how much QA the model goes through. Existing buildings are not orthogonal. A wall is out of plane, a slab slopes, a beam is half hidden behind ductwork. At LOD 200 you can model the nominal condition. At LOD 300 the scope has to say how those irregularities are handled, because that interpretation work is where the hours go.
Why it moves the price
LOD affects cost because it affects time, and time is what scan-to-BIM is priced on. Modeling-only work commonly runs in the range of $0.10 to $0.30 per square foot for a basic shell, climbing toward $1 to $10 or more per square foot for dense, high-detail MEP, or roughly $50 to $150 per hour for scoped categories. These are vendor-derived US ranges, not a neutral benchmark. No standard $/sqft exists for scan-to-BIM, because the target LOD and how many elements sit inside the scope set the hours, not the size of the floor plate. A 5,000 square foot mechanical room modeled at LOD 350 can cost more than a 50,000 square foot empty warehouse shell.
That is also why the cheap quote is not always the bargain. Ordering too little detail produces missing categories, unreliable coordination, and a second modeling round when the design team discovers the door openings or ceiling heights they needed were never modeled. Over-modeling produces the opposite waste: heavy files, longer schedules, false confidence in conditions the scanner never actually saw, and detail that gets demolished anyway. The right level sits between those, and it is project-specific. For how the rest of the modeling variables move a number, see how much scan-to-BIM costs; the field capture is usually quoted separately, and those ranges are in 3D laser scanning cost.
The point cloud sets the ceiling
LOD does not override what the scan can see. A laser scanner records visible, accessible surfaces. It does not read through a wall, into a sealed plenum, or below grade. If something is hidden, blocked by stored material, glass or highly reflective, or simply outside the scanned area, it cannot be modeled at LOD 300 as though it were measured. A good scope states which areas are visible, which are inaccessible, what geometry is assumed rather than observed, and what is excluded.
The cloud also has to be registered before it can serve as a modeling basis at all, a difference covered in registered vs unregistered point clouds. The registered cloud, delivered as E57 for vendor-neutral interchange or as RCP and RCS for Autodesk tools, stays available so the design team can measure anything that was captured but not modeled. The model is built from what the source supports, and the scope should say so in writing. How capture and modeling hand off to each other, step by step, is in the scan-to-BIM workflow, and what a site needs to look like on scan day is in the laser scanning site prep checklist.
A short checklist before you ask for LOD 300
Before requesting LOD 300, send the intended use of the model, the target Revit version, the categories that must be modeled and the ones that can stay lower, the areas in scope, any views or sheets that will be generated, point cloud availability, known hidden or inaccessible areas, your model standard if you have one, and your review expectations. If you are unsure, ask for a recommended LOD by category instead of choosing one label for the whole building. That single change is what lets a quote come back scoped rather than hedged.
If you want a model scoped this way rather than priced as a guess, request a quote and we will recommend an LOD by category against your actual building and intended use.
FAQ
What is scan to BIM?
Scan-to-BIM is the process of capturing a site with a 3D laser scanner, registering the resulting point cloud, and using it as the measured basis for a building information model, typically in Revit. The scan records the visible and accessible existing conditions; the model interprets that data into usable elements at a defined level of development.
How do I import a point cloud into Revit?
Index the registered scans in Autodesk ReCap, which writes an RCS file per scan and an RCP project file that references them. Revit links RCP or RCS directly through Insert then Point Cloud; an E57 has to pass through ReCap first, since Revit does not link E57. Choose the positioning deliberately, usually origin to origin or by shared coordinates, then pin the linked cloud so it cannot be nudged out of position. The modeler builds Revit elements by tracing and aligning to the cloud. We cover the full sequence in the import point cloud into Revit article.
How much does scan to BIM cost?
No neutral per-square-foot figure applies, because the LOD you request and the number of elements in scope drive cost, not the area underfoot. For modeling only, when you supply the registered cloud, vendor-derived US ranges run roughly $0.10 to $0.30 per square foot for a basic shell up to $1 to $10 or more per square foot for dense MEP, or about $50 to $150 per hour. Turnkey scan-plus-model adds field capture, usually billed by the day; how much scan-to-BIM costs breaks down those ranges. LOD, categories in scope, site access, and schedule move the number.
Is LOD 300 more accurate than LOD 200?
Not exactly. LOD describes how developed and reliable an element is, while USIBD LOA describes how accurately the condition was measured and how accurately the delivered model represents that measurement. A model can be measured with high accuracy and still be delivered at LOD 200. Scope both the LOD per category and an accuracy tolerance so accuracy is a measurable acceptance criterion.
Can one model mix LOD 200 and LOD 300?
Yes, and it is usually the practical choice. A common pattern is LOD 300 walls, openings, and primary structure with LOD 200 ceilings and no modeled small-bore MEP. Mixed scope keeps effort and cost on the categories that matter to the downstream work.
Related reading: what LOD 300 means in BIM, the scan-to-BIM workflow, and as-built drawings for tenant improvement.
