The goal of autocalibration consists in obtaining one seamless display composed of multiple outputs like presented on the image below:

Setup

To perform autocalibration you will need this equipment:

  • PoE switch
  • Camera:
    - for “FLIR”:{https://www.flir.com/products/blackfly-s-gige/?model=BFS-PGE-200S6M-C} camera setup see this topic.
    - for “IDS”:{https://fr.ids-imaging.com/store/products/cameras/gige-cameras.html} camera setup see this topic.

Autocalibration steps.

Step 1/10. Instructions

In order to achieve a better autocalibration, prepare your environment :
1. For each projector, be sure the following build-in settings are :
On the projector
- automatic keystone set to OFF
- overscan set to OFF
- noise reduction set to OFF

2. for better image contrast consider minimizing the influence from the external light sources (turn the lights off)

Step 2/10. Defining groups

This page allows defining an autocalibration group. Each group contains a set of outputs that need to be calibrated and a camera.
In addition, in this window you can choose the autocalibration mode:
- Default. It should be used for the first autocalibration. You proceed through all the steps, i.e. full control on calibration parameters.
- One-click calibration. Brings you directly to step 10, image acquisition. In this mode, the new image capture is made using the parameters defined during the last Default calibration. This mode is useful in case when one of the projectors has moved and you need to update autocalibration and keep the same masks and warp. If it was the camera that has moved, the display mask and the warp may not be valid anymore (consider to run the autocalibration in default mode).

By clicking next you continue the autocalibration in the default mode.

Step 3/10. Selecting outputs

To add an output to autocalibration you have to select the corresponding VNode and then drag&drop the output you want to add.
Blend group should be adapted only for multiple layer configurations (such as dual projector configuration). In fact, the blending will be performed only between the outputs of the same group.

Some examples:

  • Flat screen. 2 projectors side by side

    Blend group Output
    0 outpuLeft
    0 outputRight

  • Curved screen. 4 projectors, 2 duals side by side

    Blend group Output
    0 outputLeft1
    1 outputLeft2
    0 outputRight1
    1 outputRight2

Step 4/10. Selecting camera

On this page, you have to choose the camera that will be used for autocalibration. The right panel displays the list of all available cameras.

Step 5/10. Setting camera parameters

The main parameter in this section is camera exposure. You need to ensure good contrast between projector black and white (see example below). As for gain, normally it should take the minimal value. The only reason to increase gain is when the exposure is at its maximum but the contrast is too low.

Exposure too low Exposure OK Exposure too high

The second group of parameters is responsible for image crop. While it doesn’t have a direct impact on the results of autocalibration, applying the crop when you can has the following advantages. It increases the speed of image acquisition and the acquired images take less space on the hard drive.

no crop with crop

Step 6/10. Pattern settings

Pattern size slider allows defining the size of the projected pattern. This parameter has a huge impact on the results of calibration. Unfortunately, the good value may be very different from one setup to another, and can only be determined experimentally. The good rule of thumb is to choose such pattern size so you have approximately 20 circles along the output width.

Avoid pattern sizes that are too small or too big.

Too small OK Too big

Step 7/10. Configuration

At this step, the user is asked to choose the type of display that is actually calibrated. Clicking on the corresponding button load the preset for the parameters in the advanced tab.

Deformation tolerance parameters takes four possible values: Low, Normal, High, Very high. It defines the rigidity of the transformation that the algorithm is allowed to apply to the output image. You should take a higher tolerance to compensate the imperfection of the screen surface.

Advanced parameters:
- Filter features by size: this option allows to discard features (blobs) that are too different from an average size of all detected blobs. It should be active only for flat or almost flat screens.
- Auto threshold: allows choosing whether pattern binarization is performed automatically or using the threshold defined in Step 9.
- Undistort image: undistort image during calibration
- Display outside camera view: tells the algorithm whether the camera sees the whole display or not. By setting this parameter to on, the algorithm will consider the zones invisible by the camera as zones without blending (full projection). Otherwise, it will set blending to zero (no image emitted).
- Update regions from displacement map: should be used only for flat screens in case where the camera doesn’t see the whole screen.

Step 8/10. Global mask definition

In this window, you can control the mask that is applied to the whole screen. In the following example, there won’t be any projection outside the blue region. In order to start drawing the mask, click on “+” button (add path). See section MaskEditor for more details (under development). The interface of this editor is similar to any vector graphics editor such as Adobe Illustrator or Inkscape. Specifying no mask is equivalent to drawing the mask covering the whole screen.

Step 9/10. Mask per output

Each autocalibration output has a mask. The mask corresponds to the region of the display that is covered by the given output (projector). The parameter binarization threshold allows to adjust it. The blue region on the image corresponds to the region that will then be used for the blending of the output.

The value of the threshold is right when the blue region on the image corresponds at the best to the region covered by the output.

too low ok too high

Step 10/10. Acquiring images

Displays the progress of image acquisition

Need more help with this?
Don’t hesitate to contact us here.

Thanks for your feedback.