Using the PointCloudCoercer to Convert Point Clouds

Liz Sanderson
Liz Sanderson
  • Updated

FME Version

  • FME 2018.x


The PointCloudCoercer is used when writing to formats which do not support point clouds. It will transform the point cloud into either a multipoint feature, which is an aggregate of points, or into individual points.

Transformer Highlights

There are 5 options for output geometry with the PointCloudCoercer:

  • Spatial Equal Points
  • Sequential Equal Points
  • Nested Equal Points
  • Single Multipoint
  • Individual Points

Spatial, Sequential and Nested Equal Points can produce several multipoint outputs. This is demonstrated using colours in the examples below. For each of these output geometries, the maximum points per multipoint were set to 1,000 (left), 10,000 (middle) and 100,000 (right) to show how results may vary. The total number of points in the point cloud is divided by the value set in the maximum points per multipoint parameter to determine how many multipoint features will be made.


Running the Workspace

The attached workspace demonstrates the differences in results between the methods described below. If you choose to run all methods at once, the workspace will take a very long time to finish running (approximately 8 minutes). The PointCloudCoercer used for generating Individual Points will take the most time because there are millions of points that are being written individually. To run the methods separately, objects within bookmarks can be enabled or disabled. Simply right-click on a bookmark and select Disable or Enable All Objects in Bookmark.



To view the results of the PointCloudCoercer you may connect inspectors to the FeatureColorSetter transformers as seen below in image a., or turn on Feature Caching before running the workflow as seen below in image b. This will allow you to view the results in the Data Inspector.

a. 6AR-_GfDgKa0KSB59OaLiCkmag9RrpMFR047LZ4Y_Z0QphJS4dh4SxVngah6W3-Z04lvWidkJ-_PIC4jljIzJ0VLeJ67AjhwmKZDkD8IG_6ZdYLX_8M16RVlHuG1HDbLXuKM0KBL b. e9gu9SWJyY69_uMtaE5R9ldRh4XTCFp1ucbOJD6NBj7Q5gwDNX9yW1b3ebbT4WBUN0yuPOpks-wGNyLmNvbmF-2PzQZ48yf6gWTnZCeJNkwFX8Uu0AlFMs2opJD8WKN3var7g-z0


1. Spatial Equal Points

Spatial mode attempts to split the point cloud into multipoint features where each have the approximate same number of points.

Z8EljlLPuSLbPPzStgakDxrizjjIQDuy0M03UqmtyhS16JxxwNsPLO8pUulUIu0f9TA3Qp9wV9V8MrbFLMDZ47AJgJgXBWw-64le5qptYXA9eCPzP3FlUkyfBoPRxnatKMv458ze TG1CnzDXjvERkp3S45MtA-EvNqYUw5BOUJZwafQmZsxepgrtytS8WRdVoM7e042ZTVS0C1bAtSkJAxL1KvH2PjcACYOOA9pGRiroDTs3ia4iXlUEiJdcAS5O-tHH4OMj58pYlyq- GcDXHGhh43LWJmcSIT7TR58GjpcjVV1Qp0hlaL0mdn0meH1hElnuxVp-YwRUESyD_fR5VhlchhpvIyoALmBYEMOklf4F1oH2lwFfimIvCKgsf_9sNc6tNVk3CtZ1tdczTjbdHggN


2. Sequential Equal Points

Sequential mode splits the incoming point clouds in the sequential order, that is, making a chunk as soon as the number of points reached some specified amount. In the point cloud, points are not necessarily ordered in a reasonable spatial manner, so the chunks in sequential mode can look quite random and vary significantly from point cloud to point cloud.


1fSVbY6eduU9nDb0X6iVSgQ8WKyF7YfVn30rKbyrzRZp1KtH62IZbUtIFZzNtSH660w08LnJl3anTxRoIL1quLakBZ8EYhCvg8zSZgJtI10yKL5MzmCjjdCH-3V4m1JGJHndxCxO J8sx2KpwsG2KsotloWxKFaE35lQUeXnyp521kZ7bLJrlPpGV4-gIsUYI8Ss2wxrJvRP_Xr0uq-gkHCvSU93Tcz2Rs5JCfVfK879apXNBCMjbh4qBTibEVQk-EArVzsr8c6y7si5e AL-92ePNXS4GRuDCUGw-701ragru3bjK84t-4sFqjSXuEndCRl_VpO_2wj8b3S2CSpZvSx9tqjp8MwvayvCKkz-wvZvzDv6qUv1-mtxtbScz4r4K8ymr67lSWcoQ4bCqt9gQUTz8


3. Nested Equal Points

Nested mode attempts to create multipoint aggregates which are approximately the same size by dimensions. This produces a tiled like output of aggregates.



4. Single Multipoint

The single multipoint option converts the entire point cloud into a single multipoint aggregate.



5. Individual Points

The individual points mode will convert each point from the LAS point cloud into a single point feature. Since point clouds are large and can have millions of points, this conversion can be quite slow.



Please visit the documentation page for the PointCloudCoercer to learn more.


GeometryCoercer vs. PointCloudCoercer

Both the GeometryCoercer and PointCloudCoercer are similar in that they can change the geometry type of the input feature. For this reason, it may seem unclear how the outputs may differ when changing the geometry type of a point cloud.

The major distinction between the two transformers is that the PointCloudCoercer can retain the components of the point cloud in the multipoint output and the GeometryCoercer will not. This is shown in the image below. The parameters for the PointCloudCoercer were adjusted to preserve the components intensity, classification and scan_direction. In the feature information this information is retained. In the GeometryCoercer this information cannot be stored.



Please visit the documentation page for the GeometryCoercer to learn more.


Data Attribution

  • The data used here originates from open data made available by the City of Vancouver, British Columbia. It contains information licensed under the Open Government License - Vancouver.

Was this article helpful?



Please sign in to leave a comment.