Support for GeoJSON with 4-Component Coordinates and Dynamic Height Selection via modelMatrix

[FrancicoVerdu]

I'm working with GeoJSON data where each point has 4 components in its coordinates, for example: [x, y, z_relative, z_absolute]. My goal is to visualize these points in deck.gl so that, based on a configuration variable, either the relative or absolute height is used. I want to achieve this without preprocessing the data on the CPU—using GPU transformations instead by leveraging the modelMatrix property to "convert" the 4-component vector into a 3-component one by swapping or selecting between the third and fourth components.

I've encountered several challenges with this approach:

• Limitations of GeoJsonLayer: The GeoJsonLayer automatically extracts positions from the GeoJSON geometries, which means I cannot define a custom getPosition accessor. This prevents me from selecting between the third and fourth components at the layer level.

• Using modelMatrix for Transformation: My idea was to apply a transformation using a 4x4 matrix (for instance, a "swap matrix" that exchanges the third and fourth components) so that the desired value is placed in the height position used by the shader.

Question: Is it possible to use only the modelMatrix property in deck.gl to perform the necessary transformation so that, based on a configuration variable, either the relative or absolute height is selected for a GeoJSON with 4-component coordinates?

[grahambates-code]

Can you use 2 different geojsonlayers , 1 for each z and toggle between them?

[FrancicoVerdu]

Thank you for your response. Using two GeoJSONLayers is an alternative I’ve considered. The problem I see is that these layers, which I intend to be displayed and edited using deck.gl’s community EditableLayer, involve querying an API that returns the sea-level altitude of a point whenever I edit a geometry. Using two layers would mean managing two separate data states that need synchronization, though I don’t rule out eventually doing this.

For me, an interesting functionality—though I’m unsure how difficult it might be to implement—would be if the GeoJSONLayer had a property, for example, called coordinateKey, representing the key of the coordinates array. This way, you could toggle between coordinate pairs within the same layer.

[grahambates-code]

THat sounds quite niche for the GeoJSONLayer abstraction.

For something like your case you could make a composite layer contain the 2 geojsonlayers, the internal state, and the api calling. Then all that logic is wrapped inside a single layer component for reuse.