Fusion.retrieve_measure() changes depth units

When i get the depth measurement from a fusion object subscribed to a camera, it seems to change the depth units to mm:

            status_rgb = self.fusion.retrieve_image(self.rgb,camera.uuid)
            status_depth = self.fusion.retrieve_measure(self.depth,camera.uuid,sl.MEASURE.DEPTH)

Any ideas why? The sl.Fusion() and sl.Camera() objects are in different threads.

Hi @fdunbar,

This is probably due to how the InitFusionParameters are set, the coordinate_units parameter is set to millimeters by default.

Could you try changing this parameter and try again?

This is how I am constructing the sl.Fusion module:

    def _build_(self):
        self.current_gps = sl.GNSSData()
        self.fusion = sl.Fusion()
        self.current_pose = sl.Pose()
        self.rgb = sl.Mat()
        self.depth = sl.Mat()

        fusion_params = self.zed_config.parse_fusion_params()
        fusion_params.coordinate_units = sl.UNIT.METER

        fusion_params_tracking = self.zed_config.parse_fusion_params_tracking()
        init_status = self.fusion.init(fusion_params)
        if init_status.name != "SUCCESS":
            logging.error("Fusion: Cannot initialize: ", init_status)

        fusion_params_tracking = self.zed_config.parse_fusion_params_tracking()
        positional_tracking_status = self.fusion.enable_positionnal_tracking(fusion_params_tracking)
        if positional_tracking_status.name != "SUCCESS":
            logging.error("Fusion: Positional tracking not activated: ", positional_tracking_status)
            logging.info("Fusion: Positional tracking enabled")

Also experiencing this, meters is set in both Fusion and Camera but receive millimeters back

1 Like

Hi @fdunbar, @mars3,

Thank you for reporting this, we will investigate this on our end and will get back to you as soon as we have more information on a fix for this.

Hello again @fdunbar, @mars3,

Thank you for the report, this was indeed identified as a bug and will be fixed in an upcoming version of the ZED SDK.


1 Like

While you are at it, could we please get a way to generate a sl.Pose() object from a (4,4) numpy.ndarray?

I’ve logged this as a feature request, this can probably be a sample code or a helper function that we can share.