Simplification vs Smoothing in GIS

Simplification and smoothing both change the shape of vector geometry, but they solve different problems. Simplification primarily reduces geometric detail, usually by removing vertices. Smoothing primarily reduces sharp or irregular changes in direction and may add vertices to create a softer shape.

A jagged line containing hundreds of points might be simplified into a line containing only the vertices needed to preserve its broad form. Smoothing the same line might instead replace its sharp turns with more gradual curves—potentially producing more vertices than the original.

That distinction matters because a smoother feature is not necessarily a simpler one.

If your goal is smaller geometry, faster rendering, or less unnecessary detail, you are usually looking for simplification. If your goal is a less angular representation of a line or polygon, you are usually looking for smoothing.

Simplification and smoothing at a glance

• Column 1: Primary goal; Simplification: Reduce geometric detail; Smoothing: Reduce angularity or irregularity

• Column 1: Typical vertex effect; Simplification: Removes vertices; Smoothing: Often adds vertices

• Column 1: Main parameter; Simplification: Tolerance or importance threshold; Smoothing: Iterations, offset, angle or algorithm-specific controls

• Column 1: Broad shape retained?; Simplification: Ideally; Smoothing: Ideally, but geometry is reshaped

• Column 1: Useful for reducing data size?; Simplification: Often; Smoothing: Usually not

• Column 1: Useful for cartographic softness?; Simplification: Sometimes indirectly; Smoothing: Yes

• Column 1: Can coordinates change?; Simplification: Yes; Smoothing: Yes

• Column 1: Can measurements change?; Simplification: Yes; Smoothing: Yes

QGIS exposes the distinction directly in separate Simplify and Smooth operations. Its Simplify tool creates geometries with fewer vertices, while Smooth adds vertices and corners to produce progressively smoother geometry as additional iterations are applied.

Simplification asks which details can be removed

Imagine a boundary containing many small deviations:

/\/\/\/\_/\/\/\/\

At a small map scale, those deviations may be unnecessary.

A simplification algorithm asks, in effect:

Which vertices can be removed while keeping an acceptable approximation of this feature?

The result might look more like:

/\____/\____/\

The geometry still contains bends, but fewer of them.

As explained in What Is Geometry Simplification in GIS?, algorithms such as Douglas–Peucker and Visvalingam apply geometric criteria to decide which vertices are significant enough to retain.

The central trade-off is detail versus approximation.

Higher simplification usually means fewer coordinates and a less exact reproduction of the original shape.

Smoothing asks how sharp changes can be softened

Smoothing begins from a different question:

Can this geometry follow a more gradual path instead of turning sharply at each vertex?

Consider:

Original:

      /\
     /  \
____/    \____

A smoothing operation might produce something more like:

Smoothed:

     ____
   _/    \_
__/        \__

The sharp corner has been replaced by a gradual transition.

This does not require deleting coordinates. In fact, creating that gradual transition often requires additional points.

QGIS's current Smooth implementation explicitly describes this behaviour: smoothing line and polygon geometry adds vertices, and increasing the number of iterations produces smoother geometry at the cost of an increasing number of nodes.

PostGIS provides a particularly clear example through ST_ChaikinSmoothing. Each Chaikin iteration replaces interior vertices with new vertices along adjacent segments, and the number of vertices doubles with each iteration.

Smoother does not mean simpler

This is the key conceptual distinction.

Suppose a line begins with 20 vertices.

After simplification, it might contain:

8 vertices

After smoothing, it might instead contain:

40 vertices

Yet the second line may look visually simpler because it has fewer sharp turns.

This creates two different meanings of the word simple.

Cartographically, a smooth curve can look clean and uncomplicated.

Computationally, it may be more complex because it contains more coordinates.

So if the goal is:

Reduce geometry size.

Smoothing may move you in the wrong direction.

If the goal is:

Make this angular geometry look less harsh.

Simplification alone may not produce the desired form.

The operations can produce very different results from the same line

Consider a GPS track:

•--•-•---•--•-•--•---•--•

Some of its small changes in direction may reflect measurement noise.

Simplification

A simplifier might retain only the more significant changes:

•---------•-------•-------•

The result uses fewer points.

Smoothing

A smoothing operation might instead construct a flowing approximation through the original path:

~~~~~~~\________/~~~~~~~~~

The geometry may contain many newly generated vertices even though it appears less noisy.

The appropriate output depends on what the points represent.

If the original deviations are unnecessary detail, simplification may be sufficient.

If the underlying feature is believed to follow a gradual path but was represented with sparse or angular vertices, smoothing may be more appropriate.

Neither transformation proves what the real path was.

Simplification is often used for performance and scale

Simplification has a strong computational motivation.

Detailed geometry can increase:

• file size;

• network transfer;

• vector-tile size;

• rendering cost;

• spatial-processing cost;

• memory use.

When the detail is invisible or irrelevant at the intended map scale, simplifying it can make a dataset much more efficient.

For example, a coastline displayed on a world map does not need the same coordinate density as a coastline used for detailed local analysis.

The simplified representation can retain major bends while discarding finer structure.

This is one reason simplification is commonly part of multi-scale mapping.

Smoothing generally does not solve that problem because its purpose is not vertex reduction.

Smoothing is often cartographic

Smoothing can be useful when geometry is structurally correct enough but looks unnaturally angular for its intended representation.

Examples can include:

• contour lines generated from raster surfaces;

• coarse vectorised boundaries;

• hand-digitised curves with sparse vertices;

• some GPS-derived paths;

• schematic or cartographic linework.

QGIS's own training documentation, for example, contrasts simplification used to remove unnecessary detail with smoothing used to reduce sharp corners, including examples involving contour lines and sparse GPS tracks.

The word cartographic matters here.

A smoother line may communicate a phenomenon more naturally on a map without being a more accurate measurement of its real location.

Smoothing does not recover missing information

Suppose a curved river was digitised using only four vertices.

A smoothing algorithm can produce a visually flowing curve between those points.

But it does not know where the actual river flows between them.

It is constructing new geometry according to an interpolation or corner-cutting rule.

Likewise, smoothing a coarse coastline does not restore high-resolution coastline data.

The new vertices are derived coordinates, not newly observed positions.

A smoother representation can therefore look more realistic while containing no additional empirical information.

This is one of the main reasons smoothing should not be presented as “improving accuracy” unless external evidence supports that claim.

Simplification also does not reveal which detail is true

The reverse problem applies to simplification.

If a boundary is noisy because of poor source data, simplification may remove some of that noise.

But it does not know which deviations are measurement errors and which represent genuine geographic detail.

A narrow coastal inlet and a digitisation artefact may both be geometrically small.

A simplification algorithm evaluates geometry according to its mathematical rule, not according to real-world significance.

That makes both operations forms of generalisation or transformation, rather than automatic truth-improvement procedures.

Smoothing parameters control more than one thing

Simplification is commonly dominated by a tolerance parameter.

Smoothing often has several controls.

In QGIS's Smooth operation, these include:

• iterations, controlling how many smoothing passes are performed;

• offset, controlling how closely the smoothed geometry follows the original;

• maximum angle, allowing sufficiently sharp corners to be preserved instead of smoothed.

Increasing iterations can progressively soften the geometry while adding more vertices. Increasing offset can move the new boundary farther from the original. An angle constraint can protect deliberate corners such as right angles.

This last point is particularly useful because not every sharp corner is undesirable.

A parcel, building footprint, road junction, or engineered structure may intentionally contain angular geometry.

Smoothing those corners merely because they appear abrupt could make the representation worse.

Smoothing buildings is usually a warning sign

Consider a rectangular building footprint:

+---------+
|         |
|         |
+---------+

Apply indiscriminate smoothing and it might begin to resemble:

  _______
 /       \
|         |
 \_______/

The geometry is visually softer but less faithful to the intended building shape.

The same problem occurs with cadastral parcels, grid boundaries, engineering features, and other objects whose corners are meaningful.

Smoothing tends to be more natural for features that are expected to vary continuously, such as some contours or generalised natural boundaries.

Even there, the analytical purpose needs to justify changing the geometry.

Simplification can preserve angular character

Simplification does not inherently make straight-edged features rounded.

A complex polygon composed of many nearly collinear vertices may simplify to a smaller set of straight segments while retaining its basic angular character.

This can be useful for artificial features or generalised administrative boundaries.

For example:

Original edge:

•--•--•--•--•--•

might simplify to:

•--------------•

The unnecessary intermediate vertices disappear, but the result remains straight.

This illustrates why simplification and smoothing should not be treated as different strengths of the same operation.

They change geometry according to different rules.

Both operations can move boundaries

Simplification removes vertices and replaces several original segments with fewer new ones.

Smoothing constructs new segments around the original corners.

Both can therefore place the resulting boundary somewhere that the original boundary did not pass.

That can change:

• area;

• perimeter;

• length;

• intersections;

• containment;

• distance to neighbouring features;

• adjacency.

The fact that smoothing may preserve the general visual character of a feature does not guarantee that its location remains unchanged.

Likewise, a simplified polygon that looks almost identical at normal zoom can contain or exclude small areas that were different in the source.

The upcoming How Simplification Can Distort Boundaries article will examine those effects more closely for simplification.

Shared boundaries require particular care

Suppose two administrative polygons share the same boundary.

If you smooth each polygon independently, the generated curves are not automatically guaranteed to remain identical.

Likewise, naive independent simplification can produce differences along what was originally one shared edge.

Potential consequences include:

• gaps;

• overlaps;

• broken adjacency.

Specialised topology-aware or coverage-aware workflows may be necessary when neighbouring features must continue to fit together exactly.

This is especially important for polygon coverages such as administrative areas, parcels, land-use zones, or other datasets where shared boundaries are structurally meaningful.

Neither “Simplify” nor “Smooth” should be interpreted as a promise that every relationship between independent features will survive.

Simplification and smoothing can be combined

The operations are not mutually exclusive.

A workflow might deliberately:

1. simplify a geometry to remove excessive vertices;

2. smooth the simplified result to soften the remaining angularity.

Or it might smooth first and then simplify the newly generated geometry to reduce the number of points.

PostGIS's ST_ChaikinSmoothing documentation even notes that smoothing can generate many more vertices and points users towards simplification functions when a reduced point count is required afterwards.

But combining the operations means applying two geometric approximations.

The result can move farther from the source than either transformation alone.

The workflow should therefore be justified by the final representation rather than applied automatically as a recipe for “cleaner” data.

Neither is the same as densification

Densification adds vertices along existing segments while generally intending to preserve the represented path.

For a straight segment:

Before:

•----------------•


After densification:

•----•----•----•--•

The line remains on the same straight path.

Smoothing can also add vertices, but those new vertices are introduced specifically to change the shape.

This gives another useful distinction:

• Densification: add coordinate samples to the existing geometry.

• Smoothing: add or reposition coordinates to soften geometry.

• Simplification: remove geometric detail.

The planned What Is Densification? article can own that third comparison more fully.

Which should you use?

Choose simplification when the problem sounds like:

This geometry contains more detail than my current map or analysis needs.

Examples include:

• reducing web-map geometry;

• creating small-scale boundary datasets;

• lowering vector-tile complexity;

• removing unnecessary vertices;

• producing lighter approximate geometry for exploratory processing.

Choose smoothing when the problem sounds like:

This geometry is too angular for the representation I need.

Examples may include:

• softening contour lines;

• reducing abrupt corners in certain generalised natural features;

• creating a more continuous-looking cartographic representation from sparse vertices.

And choose neither when:

The geometry is wrong and I need to make it accurate.

That is a data-quality or source-improvement problem unless you have an explicit geometric generalisation rule that justifies the transformation.

A practical decision table

• Question: Do I need fewer vertices?; Simplify: ✓; Smooth: Usually no

• Question: Is reducing file/rendering complexity a primary goal?; Simplify: ✓; Smooth:

• Question: Do I want softer corners?; Simplify: ; Smooth: ✓

• Question: Is the geometry too angular because of sparse representation?; Simplify: Possibly; Smooth: ✓

• Question: Do exact coordinates need to remain unchanged?; Simplify: Neither; Smooth: Neither

• Question: Can adding vertices be acceptable?; Simplify: Not normally required; Smooth: ✓

• Question: Are deliberate sharp corners important?; Simplify: Requires care; Smooth: Requires particular care

• Question: Do I need to preserve exact shared boundaries?; Simplify: Use specialised method; Smooth: Use specialised workflow

• Question: Am I trying to repair inaccurate source data?; Simplify: Neither automatically; Smooth: Neither automatically

The difference comes down to the goal

Simplification and smoothing can both make a map look cleaner, which is why they are easy to confuse.

But they optimise different properties.

Simplification asks how much detail can be removed.

Smoothing asks how the geometry can transition more gradually between directions.

A simplified coastline may still contain sharp bends, just fewer of them.

A smoothed coastline may contain more coordinates than before while appearing less angular.

Neither is inherently more accurate, and neither should automatically replace the original geometry.

If performance, scale, and vertex count are the problem, simplify.

If angular form is the problem and changing that form is justified, smooth.

And if the objective cannot be stated more precisely than “make the geometry look better”, define what better means before choosing either operation.

References

1. QGIS Documentation — Simplify and Smooth. Documents the separate QGIS Simplify and Smooth operations: Simplify reduces vertex count using methods including Douglas–Peucker and Visvalingam, while Smooth adds vertices and uses iterations, offset, and angle controls to soften geometry.

2. PostGIS — ST_ChaikinSmoothing. Documents Chaikin smoothing for linear and polygonal geometry, including the replacement of interior vertices and the growth in vertex count with repeated smoothing iterations.