Difference between revisions of "Calibration pane"

Line 161: Line 161:
  
  
[[File:Begin_wanding.jpg|left|100px|Wand Sample Collection Start]]
+
[[File:Begin_wanding.jpg|left|Wand Sample Collection Start]]
[[File:Finished_wanding.jpg|100px|Wand Sample Collection Finished]]
+
[[File:Finished_wanding.jpg|Wand Sample Collection Finished]]
  
  

Revision as of 06:02, 22 October 2014

Camera Calibration Pane

Back To Main Motive Reference Page

CALIBRATION

Block Visible

This blocks all pixels that are above the set threshold. By default the threshold is set to 200 but this can be changed by the user in the cameras pane. Pixels in a camera image will have a grayscale value between 0 and 255 inclusively. If the default threshold is used, a pixel that is above 200 will be blocked along with the surrounding pixels.

This feature is a quick way to block data that is not needed and can be used in tandem with manual masking.


Start Wanding

This will start recording wand samples. After masking the cameras, press the start wanding button to begin your wand wave.


Reset

This will stop wand acquisition and the calibration solver.


WAND SETTINGS

OptiWand

This options allows the user to select which calibration wand their using.

Wands generally come in 250, 400 and 500 mm sizes. However, custom wands can be made.

A 250 mm wand should be used for smaller volumes or for systems that have cameras with lenses that have larger focal lengths. The reason being that the cameras will not be able to see all 3 markers on a 500 mm wand if the wand is close to the camera or the camera has a very narrow view angle due to it's lens type. If your camera's are not collecting wand samples while wanding, it may be a good idea to bring up a new wand. A 250 mm wand is good to use in most small to medium volumes.

When making a calibration wand, understand that the system accuracy will be tied directly to the accuracy at which the wand is constructed. A poorly measure wand will result in poor calibration results.

To make a wand all that is needed is 3 markers at set distances in a line.


Wand Length (mm)

This can be set when creating a custom wand and is the measure of the distance between the two outer marker centers. The accuracy of this measurement will directly impact camera calibration results, so be careful when creating and setting a custom wand.


Center Distance (mm)

Defines the distance in millimeters between the outer post and the center post (use the shorter of the two center offset distances). For use with custom calibration wands.


ADVANCED SETTINGS

Camera Group

Default: Show

Change this to the camera group you want to calibrate. Only one camera group can be calibrated at a time. The available camera groups will appear in the drop down menu. Group one is the master tracking group, group two is the reference camera group and additional camera groups will appear if they were created in the camera pane.


Visual Solver

Default: Show

This toggles the display of wand samples and point cloud calibration visuals in the 2D and 3D view. If you're running on a lower end machine or graphics card with a large system, it is best to turn this feature off. The visual display will in fact eat up some computing power you may want reserved for getting quicker calibration results.


Lens Distortion

Default: Show

Use this to toggle the display of the lens distortion solution results in the 2D view. The lens distortion will be represented by a square grid that maps the distortion.

Lens Distortion Visuals


Cameras

Default: Show

This toggles the display of the cameras during the solve.


Wanding Projection

Default: Show

This toggles the display of wand samples projected in the 3D view. Turn this off if you're calibrating a very large system.


Projection Error

Default: Show

This toggles the display of error, reported as a color in the projected wand samples and markers. The wand samples will have a color between blue (good sample) and red (poor sample). Make sure the samples you collect are mostly good samples. As will all visual feedback, it may be a good idea to turn this off if you're calibrating a larger system.


Residual Error

Default: 6 mm

Set the tolerance for the reported error in the projected marker sample during calibration.


Wand Error

Default: 8 mm

This sets the tolerance for the reported error in the projected wand sample.


Sample Space

Default: 1

Use this to increase the spacing between displayed samples that are projected in the 3D view. Increasing this will skip more samples but will make the visual wand projections easier to see.


Min Object Size

Default: 2 pixels

This sets the minimum number of pixels above the threshold that comprise a wand marker for the marker to contribute to a wand sample. If the markers on the wand are small and the wand is far from the camera the camera will see the markers as only a few pixels. Use this feature to eliminate higher error wand samples. Keep the setting at 2 unless you're certain of how the cameras are seeing the wand markers.


Min Object Circularity

Default: 0.8

This sets the tolerance for the circularity of the wand markers. Values range from 0 (not circular) to 1 (perfectly circular). The default will work in most cases.


Calibration Type

Default: Full

Sets the type of calibration solver to use.

Full - start the calibration from scratch, discarding any prior known position of the camera group or lens distortion information. A full calibration will also take the longest time to run.

Refine - increment the placement of the cameras based on prior known positions. The solve time will be faster then a Full calibration. Only use this if your previous calibration closely reflects the placement of cameras. In other words, Refine calibration only works if you do not move the cameras significantly from when you last calibrated them. Only slight modifications can be allowed in camera position and orientation, which often occurs naturally from the environment such as mount expansion.

Visual - only render the calibration solution visual and will not calibrate your cameras.


CALIBRATION ENGINE

Calculate

Initiates the calibration solver. Press this button after collecting enough wand samples.


Apply Result

Applies the calibration results to the cameras.

Once pressed, this button will bring up a dialog box prompt you to save your wanding as a .tak file. The wanding take can be used for a variety of reasons, including verification of wanding results and for record keeping. Often times it is not needed.

After you save the wanding the camera calibration pane will switch over to the Ground Plane tab so you can set the global origin.


Wanding Table

While wanding the bottom part of the Camera Calibration Pane will show a table of the number of samples collected for each camera in the system. The samples will increase as the wand is waved in the capture volume.


Wand Sample Collection Start

Wand Sample Collection Finished


Calibration Results Table
Calibration Results Table

The calibration results will show in the Calibration Engine portion of the Calibration pane. If no calibration is being processed this area will remain blank. However, when a wanding or a calibration solver is underway, this field will be populated with a table showing the live results of the solution. The components of that table are described below.

Column Header Description
Cam This column shows the camera number associated with the row of data, the wanding result or the average result of the camera group. The wanding has error and is reported as the deviation in the wand markers across all samples.
Samp The number of samples utilized at the current stage of the solution. This number can climb as the solution converges.
Quality The quality given to the current pixel error. You will see the quality increase as the pixel error drops. Quality ranges in the progress bar. Red is poor, yellow is good, and green is excellent.
Focal This is the calculated or given focal length of the camera. Doesn't apply to the average or the wanding.
PixErr The average pixel error of the camera. Represent the 2 dimensional error of the camera's ability to locate a marker.

The elapsed time of the solver is shown at the bottom of the list.

As the calibration proceeds through the various phases of the solution you may notice the results slowing when a phases is finishing. Let the calibration finish all phases of the calibration. Once the solver converges on an appropriate solution, press the Apply Result button to apply the solution to the cameras. If you are unsatisfied with the results, hit reset near the top of the pane to cancel the results.


GROUND PLANE

Set Ground Plane

Set the location of the global origin. Use an 'L' Frame or 3 markers in the shape of an 'L'. If only 3 markers are seen by the cameras, you can simply press 'Set Ground Plane'. If more markers are in view then you can select the 3 markers you want to use in the 3D viewport and then press 'Set Ground Plane'.

The long leg of the 'L' frame is used for the negative Z axis. The shorter leg will be set as the positive X axis.


Ground Plane Offset

Default: 55 mm

The Ground Plane Offset is the location of the L frame vertex in the Y direction relative to the global origin. Use positive values to set the global origin below the 3 marker vertex and negative values to set the global origin above the 3 marker vertex.

The global origin is arbitrary and can be placed anywhere the user desires.


Ground Plane Refinement

The Ground Plane Refinement allows virtual floor to be offset in the Y direction only.

To use this place 4 or more markers on the ground throughout the volume. Change the offset to half the diameter of the largest marker in the volume. Then press the Ground Refinement button. This will change the vertical location of the floor, ensuring all of the markers are above the floor.

A scenario where this feature comes in handy is when placing the L frame on a force plate that is not entirely level with the floor.


Volume Translation

The Volume Translation is used modifies the global origin after it has been set.

Simply enter the amounts you want to translate the origin in the X, Y and/or Z direction and press the Apply Translation button. There is no limit to the number of translations that can be applied and there is no memory once a translation is applied. To revert a translation, simply translate the origin be an equivalent amount in the opposite direction.


Volume Rotation

The Volume Rotation is use to apply a rotational offset to the current global origin.