Skip to content

Target management

The targets are a main element of Kratos. They are the representation of the tracked people in the 3D space. Target's position is what is send to third party systems. A list of the Targets of the show can be find in the Target tab of the Show explorer view. You can interact with the selected Target's parameter in this view. The Target patcher view allow to the link Targets with desired Trackers.

Appearance

img

The first parameter relate to the appearance of the Target.

  • You can change the Target's name. This name is the name send via PSN.
  • You can change the Target's color. The color is a hex code.
  • You can change the Target's avatar by clicking the "pick" button

You can choose other images than the ones proposed by default

To do so, you have to specify the image's path in the textfield, starting by file: Example : file://C:/Users/nao/Desktop/singer.png Supported image formats are .png, .svg, .jpeg, .gif

Standard postprocessing parameters

In function of the application you are in, it's may be important to work on the position of a Target before sending it to third party systems. Target parameters are here to be tweaked around until you find a configuration that suits your application. The following parameters are the most standard ones :

img

  • Prediction : You can use this parameter to predict the position of a Target in function of it's current speed and acceleration. The value chosen corresponds to the time you want to predict ahead in microseconds. This parameter is for example very useful to counter the inertia of lighting fixture during followspot applications.
  • Smoothing : You can use this parameter to smooth the position of a Target by averaging it's position over a period of time. The value chosen corresponds to the period of time Kratos will take to average the position of a Target in microseconds. This parameter is useful to smooth out any noise in the target's measured position.

Prediction and smoothing can be used independently and together to achieve the desired result.

  • Mode 2D : You can use this parameter to force the Target to be in 2D mode when tracked automatically. This useful to eliminate any measurement uncertainty in height that can occurs with the automatic tracking.

We recommend using 2D mode whenever there is no elevation change in your tracking area.

  • Manual Z : When you are in manual tracking mode, the vertical component of the Target (Z coordinate) is fixed at a certain height. This parameter allows you to change this height.
  • Manual Z From and To : These parameters allows you to define the minium and maximum values of the Target's manual Z fader in the Target patcher view.
  • Validity : The validity of a Target correspond to the validity of the Tracker associated (see Tracker validity)

Raw prediction

The speed used in the standard prediction is the speed predicted by the IA which apply a Kalman filter to the 3D tracks. This speed is very noisy and the AI has a tendency to over filter it.

When enabling the raw prediction, Kratos will calculate the speed of the Target on its own and not use the one predicted by the IA.

img

You can then have access to additional parameters about the prediction :

  • Prediction samples : This parameter is the number of Target's position samples (in elapsed frames) Kratos will use to compute the current speed of the Target.
  • Max Prediction Distance : This parameter is the maximum distance (in meters) Kratos is able to predict the Target's position from its actual position. This allows you to prevent Kratos to over predict the Target position when the Target is moving fast.
  • Reset Prediction Distance : This parameter is the maximum distance (in meters) a Target is allowed to move between two frames before the prediction is reinitialized. This is useful to avoid having a prediction stuck at the max prediction distance for a long period of time if you average over a lot of samples and that there is a sudden big gap in the Target's position (when you change the Target's patched Tracker for example).

When using raw prediction, you still need to use the standard prediction parameter to set the time you want to predict ahead.

Manual smoothing

The manual smoothing option is a Kalman filter applied to the Target's position when in manual tracking mode.

img

When enabled, you can manage the following parameters to tune the filter, which can be set independently for each axis :

  • Acceleration Standard Deviation : Expected noise in the Target's acceleration (in m/s2).
  • Measure Standard Deviation : Expected noise in the Target's measured position (in m).

The manual smoothing adds to the standard smoothing parameter.

Prediction smoothing

As for the manual smoothing, the prediction smoothing option is a Kalman filter applied to the predicted position of the Target.

img

When enabled, you can manage the following parameters to tune the filter, which can be set independently for each axis :

  • Min and Max Predicted speed : Clamp the speed values of the filter (in m/s).
  • Acceleration Standard Deviation : Expected noise in the Target's acceleration (in m/s2).
  • Measure Standard Deviation : Expected noise in the Target's measured position (in m).

As for the manual smoothing, the prediction smoothing adds to the standard smoothing parameter (and to the manual smoothing parameter if in manual mode).

Transform

You can monitor the position and speed of the Target here. The "Predicted Position" is the post-processed position (with all the parameters applied : prediction, smoothing, etc...).

img

Beakon

If needed, and if you have a BeaKon system set up, you can do without vision and base the Target's position only on the position of the associated tag.

img

Controls :

  • Enable tag : This allows you to switch to the previously described operating mode (tag's position as Target's position).
  • Enable kalman : If you activate the tag position mode, you can apply a Kalman filter to the tag's position to smooth it out.
  • Acceleration Standard Deviation : Expected noise in the tag's acceleration (in m/s2).
  • Standard Deviation Offset : Tag position offset compared to target's center of mass position (in m).
  • Standard Deviation Scale : Reported tag's standard deviation scale factor.