Technical guide

Degradation mapping according to UNI 11182: how to prepare the vector base

Degradation mapping is rejected more often because of the geometric base than because of the diagnosis. Here is how to build a vector survey that holds up to review, from scale to layers.

Published · 2026-09-14

Where degradation mapping sits in the deliverable sequence

The most common setup error is to produce the degradation map as the first thematic drawing. It is not. The correct sequence is: geometric survey — the metrically correct elevation, with no themes; material mapping — identification of constituent materials by surface area; degradation mapping — the forms of alteration, referred to the materials of the previous drawing; intervention project — the works, referred to the forms of the previous drawing.

The order is not bureaucratic. UNI 11182 describes forms of alteration of natural and artificial stone materials, and several forms are defined in relation to the substrate: a crust on marble and a crust on lime plaster are the same lexical entry but two different works, with two different prices. Skipping material mapping means the degradation drawing cannot be turned into a bill of quantities without returning to site.

The three thematic drawings must share the exact same geometric base, i.e. the same starting file with the same architectural layers. If the base changes from one drawing to another, the surfaces will not overlap and comparison becomes impossible.

What UNI 11182 prescribes

UNI 11182:2006, Cultural heritage — Natural and artificial stone materials — Description of the form of alteration — Terms and definitions, establishes the lexicon for naming forms of alteration and degradation visible to the naked eye. It replaced the previous NORMAL 1/88, expanding and updating its repertoire, and applies to both natural stone and artificial stone materials: mortars, stuccoes, plasters, bricks, concrete.

It is not an academic reference. The facade handbook of the Padua and Venice Soprintendenza explicitly requires the use of UNI 11182 terminology for the recognition and classification of degradation in graphic deliverables, and similar provisions appear in the regulations of many other Soprintendenze. If the legend contains terms such as 'darkening', 'blistering' or 'powdering', you are using a vocabulary the authority does not recognise, and the drawing is challengeable on a purely formal level before any substantive review.

A distinction made by the standard that is systematically confused when drafting: alteration means a modification of the material that does not necessarily imply a worsening of conservation status, whereas degradation implies a worsening. A stable oxalate patina is an alteration and in many cases must be preserved; a black crust is a degradation and must be removed. If the legend treats them as equivalent, the resulting intervention project will prescribe cleaning on surfaces that should not be touched — an error that in restoration cannot be corrected.

The full definitions are in the standard, a paid document available from the UNI catalogue. What matters here is another level: how each form behaves geometrically on the drawing.

Forms of alteration classified by representation geometry

This classification is not in the standard: it is functional to the drawing, and decides whether quantities can be extracted from the sheet.

Areal forms — mapped as closed, hatched polylines. These yield square metres and make up most of the repertoire: chromatic alteration · alveolization · concretion · crust · differential degradation · surface deposit · disintegration · efflorescence · erosion · exfoliation · incrustation · stain · patina · biological patina · film · pitting · pulverization · vegetation · scaling · sub-efflorescence.

Linear forms — mapped as open polylines, with coded thickness or hatch. These yield linear metres, the unit used to estimate pointing and injections: cracking · fracturing. For these two the 2D drawing loses information the bill of quantities needs: width. A 2 mm crack and a 15 mm crack have the same linear development but two different works. It is best to set up separate layers for width classes — for example under 5 mm, 5–20 mm, over 20 mm — or an attribute on the polyline, stating the convention in the legend.

Forms with loss or change of material — require the surface but carry a third dimension the drawing does not represent: deformation · detachment · lacuna · loss · blistering. For these the surface alone is not enough for quantification: a loss 2 cm deep and one 15 cm deep have the same area and incomparable restoration volumes. They must be accompanied by spot dimensions of depth or a schedule keyed to a polygon identifier.

The distinction between lacuna and loss is the one most often mistaken. The first presupposes an element whose original presence is documentable and whose perimeter can be read; the second is the loss of parts of the material. They are two different bill items and must go on two different layers.

Required precision and tolerances

Before discussing layers we must settle what 'to scale' means in numerical terms, because this is the point on which a drawing is defended or falls.

Graphical error — established convention fixes the minimum appreciable graphic error on a drawing at 0.2 mm. Multiplied by the scale denominator, this gives the admissible metric tolerance: at 1:20 = 4 mm; at 1:50 = 10 mm; at 1:100 = 20 mm.

Degradation maps of façades are normally drawn at 1:50, with details at 1:20 or 1:10 for significant details. The operational tolerance is therefore of the order of a centimetre: a overall scale error exceeding this value makes the drawing formally non-compliant with the stated scale.

Resolution of the photographic base — for vectorisation to respect that tolerance, the photoplan must have a ground sample distance (GSD) smaller than the graphical error: GSD = real width framed / image width in pixels.

A 12 m façade shot with a 4000 px wide image gives a GSD of 3 mm, adequate for 1:50 and still acceptable for 1:20. The same façade shot at 1200 px gives 10 mm, i.e. exactly the tolerance at 1:50: already at the limit, and insufficient for any detail. Practical rule: aim for a GSD equal to or less than half the tolerance of the output scale.

The limit of planar rectification — photographic rectification projects the image onto a plane. Anything projecting from that plane is shifted radially from the principal point by an amount estimable as: shift ≈ projection depth × (lateral distance from camera axis / shooting distance).

An example: a cornice with 30 cm projection, shot from 15 m, at a point on the façade 6 m laterally from the camera axis, is shifted by about 12 cm. That is an order of magnitude ten times the 1:50 tolerance.

Operational consequences: planar rectification is valid for substantially coplanar façades; significantly projecting elements (cornices, string courses, balconies, portals) must be surveyed separately and not derived from the photoplan; shots must be set up as frontal and distant as possible, because both conditions reduce the ratio governing the shift.

Scale calibration and error propagation

The vector base must be anchored to a real measurement taken on site. The choice of that measurement determines the precision of the whole drawing, and almost nobody thinks about it.

The relative scale error is the ratio between the uncertainty of the reference measurement and its length. With the same uncertainty — let us say 5 mm: on a 1 m base → 0.5% scale error → on a 12 m façade, 6 cm error; on a 10 m base → 0.05% scale error → on the same façade, 6 mm. Same measurement uncertainty, one order of magnitude difference in the result.

Hence the rules: choose the longest readable base in the drawing — the spacing between the outermost openings beats the width of a portal; choose materially sharp ends — a live arris, not a degraded plaster edge or a mortar bead; measure with a laser distance meter, not a measuring wheel — over 10 m a wheel, with catenary and variable pull, introduces more uncertainty than the long base removes; take a second measurement on a different axis and use it for verification, not calibration — if after calibration the second measurement on the drawing diverges beyond the scale tolerance, the problem is upstream: the rectification is wrong.

It is worth being explicit about scale error in economic terms too. The areas of the degradation polygons become the quantities of the metric bill: cleaning, consolidation, pointing, biocide treatment. Scale error propagates as the square on surfaces: a 1% linear error becomes about 2% on areas. On a 400 m² façade with priced works, that is thousands of euros of deviation in a bill someone signs off on.

CAD file structure

Layers — one rule only, no exceptions: one layer per form of alteration. Never a single 'degradation' layer with different hatches inside it. If everything is grouped together, extracting quantities means selecting by hatch manually, and on a façade with two hundred polygons that is a job that will not get done.

Naming that holds up: DEG_CRUST · DEG_LOSS · DEG_LACUNA · DEG_BIOLOGICAL-PATINA · DEG_CRACKING_05-20 · DEG_EFFLORESCENCE · MAT_LIME-PLASTER · MAT_ISTRIA-STONE · ARC_PROFILES · ARC_OPENINGS · ARC_DIMENSIONS. Constant prefix by family (DEG, MAT, ARC) so an entire theme can be isolated with a filter; name of the form exactly as in the standard; hyphen instead of space; optional class suffix. Colour and line type assigned by layer, never by object: assigning them by object keeps the layer filter working but breaks plotting.

File units — set the drawing unit before starting and declare it. In DXF this information travels in the $INSUNITS variable: if it is undefined, the file may be rescaled by a factor of 10, 100 or 1000 when imported into another environment without any warning. It is the most trivial and most common way to lose a good calibration.

Hatching — many specifications require the graphic symbol associated with the form of alteration. In the absence of a requirement, use distinct hatches with hatch scale set by layer, not by individual object: in print, hatches of the same form at different scales look like different forms. Hatches must be associative and polygons closed, otherwise the area cannot be calculated.

Quantity extraction — with layers set up like this, extraction is automatic. In AutoCAD, DATAEXTRACTION exports to CSV the area and perimeter of all objects grouped by layer; alternatively the Area field in an associative table updates quantities whenever the polygon changes. The CSV goes straight into the bill spreadsheet. This is why layer structure is not pedantry: it turns three days of manual recalculation into an export.

Delivery format — DXF for technical delivery, in a version the authority can open: R2013 is today the reasonable compromise, readable even by older installations. Before exporting: purge empty layers, check for open polygons, explode unnecessary blocks.

SVG is the right format if the mapping is done in vector graphics software rather than CAD — a widespread practice among conservators, who work faster there. The price is that area calculation is not native and must be reconstructed, so it only makes sense if the quantitative restitution is carried out on another file anyway.

Where the vector base comes from

The classic path is to redraw the elevation in CAD over the photoplan. It is the correct method and no one disputes it, but on an articulated façade it is hours of manual tracing, and it is the phase where the survey eats the margin.

The alternative is to obtain vectorisation automatically from the photoplan and dedicate the time to mapping, which is the part that requires the technician's eye. That is what CVCAD does: upload the rectified image, the application returns vector geometry, and before download you pick two snap points on the preview and assign the real distance — in millimetres, centimetres or metres. The scale is applied to the drawing, and the downloaded DXF or SVG is already to real scale, ready to be imported and mapped.

Two declared limits, because they help you use it well. Rectification stays upstream and CVCAD does not perform it: starting from an unrectified shot, the geometry inherits the perspective deformation, and no two-point calibration can correct it. And the output geometry is a drawing base, not a finished deliverable: it must be read and cleaned, because the degradation itself — the sharp edge of a crust, the shadow of a deep lacuna, the boundary of a rising damp stain — generates high-contrast contours that the algorithm reads as lines and that must not appear in the architectural elevation.

In development: assisted pre-segmentation of alteration forms

The CVCAD team is working on an algorithm for the automatic recognition of some alteration forms from the façade image.

The goal is not to replace diagnosis. It is to return, together with the architectural geometry, a pre-segmentation of areas with homogeneous contrast attributable to a limited number of forms — those with a sufficiently stable visual signature, typically areal forms with clear colour discontinuity — delivered on separate layers and already outlined, to be validated, corrected and reclassified.

The distinction is substantial and must be stated unambiguously: an algorithm can propose a perimeter, but it cannot assign a form of alteration under UNI 11182. That assignment requires reading the substrate, the cohesion state, often a tactile or instrumental check in situ, and remains the responsibility of the technician who signs the deliverable — with the liabilities that D.Lgs. 42/2004 assigns to qualified figures for interventions on cultural heritage. What automation can remove is the perimeter tracing time, which on a complex façade is the longest and least qualified part of the job.

The errors that get a drawing sent back

Legend with non-standard terms — the most frequent reason for rejection and the easiest to avoid.

Alteration and degradation treated as synonyms — leads to prescribing cleaning on patinas that should be preserved.

Open polygons — look filled on screen, but the area cannot be calculated and the bill breaks.

Undeclared overlaps — a crust over a chromatic alteration is legitimate, but adding the areas counts the same surface twice. They must be declared in the legend or resolved hierarchically.

Degradation on the same layer as the architectural base — makes automatic quantity extraction impossible.

Different geometric base between material and degradation mapping — surfaces do not overlap and the two drawings cannot be compared.

$INSUNITS not set — the file rescales by a factor of 10 or 100 on import, with no warning.

No indication of the calibration measurement — it must be shown on the drawing, with the point where it was taken and the instrument used. This is what makes the survey verifiable by third parties, and it is the first thing a checker looks at.

CVCAD converts images into downloadable DXF and SVG vector CAD drawings, with scale calibration on two snap points before download.

Try CVCAD on your next survey

Sign up and get 1 free conversion: upload a photo and download a DXF already in real millimetre scale.

Open the app