Getting Up & Running¶
What you will need¶
A computer with Python installed¶
Calcam will work with Python on Windows, Linux or OSX. If starting from scratch and you have a choice of Python versions, it is recommended to use Python 3.7. It will work fine with earlier versions of 3.x, however 3.8 is more difficult because the the required version of VTK is not easily available for Python 3.8 (as of July 2020). You can download installers for Python 3.7 for Windows or OSX from Python.org , or get it from your favourite software repository on Linux.
You will also need the
pip package installed for the Calcam installer script to work. If you are using Python 3.4 or newer from Python.org,
pip will be installed by default, and it’s also probably included in most other modern Python distributions you might be using. If you don’t have
pip installed, the Calcam setup script will give an error message telling you so. Documentation for how to get pip can be found here .
A copy of the Calcam source code¶
The latest release version of the Calcam source can be downloaded from the Releases page on GitHub.
Alternatively, the “cutting edge” development version can be downloaded using the green Code button on the GitHub repository page. If you want to use Git, which is recommended if you want to do any development on Calcam, the source can be cloned from the GitHub reporepository with the command:
git clone https://github.com/euratom-software/calcam.git
Once you have a copy of the source files on your computer, navigate to the directory where Calcam has been unzipped or cloned and open a terminal / command prompt. To install Calcam, use the command:
python setup.py install
This will check for and try to install Calcam’s dependencies (see section below for gory details); copy the Calcam source to the appropriate Python library path; and create a convenient launcher executable for the Calcam GUI. If all goes well, this script should end with a message which looks something like this:
*************************************************************** Calcam installation complete. It can be imported as a Python module with "import calcam" The GUI can be launched using the executable: C:\Users\username\AppData\Roaming\Python\Python37\Scripts\calcam.exe ***************************************************************
For convenience it is recommended to make a shortcut to the calcam GUI executable given in the setup complete message. After the setup script is finished you can delete the downloaded calcam files, if you want.
Installing without admin/root permissions¶
If installing on a system where you do not have the relevant admin/root permissions to install python packages globally, the setup script may crash with an error related to permissions, permission denied or a similar error message. In this case, adding the
--user option to the installation command will install the package for your user account and will not require admin/root permissions.
Installing in Development mode¶
If you plan to do any development on Calcam, and want to be able to edit the Calcam source code without having to run the setup script again to have your changes go live, you can use the below alternative command to install calcam in development mode:
python setup.py develop
In this case the copy of the code you are installing from remains the “live” version and can be used for development. Again, the
--user switch can be added to install in the current user’s library path rather than the system one.
Setting up CAD Model Definitions¶
Camera calibration in Calcam is based on matching features between camera images and a CAD model of the scene viewed by the camera. As such, it is necessary to define one or more CAD models for use in calcam. The current version supports importing
.obj format 3D mesh files. It’s usually convenient to split the model in to several individual mesh files containing different parts of the scene, and these can then be turned on or off individually when working with the model. Calcam packages these mesh files in to a custom zipped file format (.ccm) along with various metadata to create a Calcam CAD model file. You can have several such files and easily switch between them at any time. It is recommended to read the CAD Models section in concepts and conventions, then consult the user guide for the Calcam Settings interface for details of how to set up CAD model definitions.
Setting up custom image sources (optional)¶
As standard, Calcam can load camera images from most common image file formats. If desired, you can set up additional custom “image sources”, which are user-defined Python modules for loading camera images in to Calcam. For example you may want to load camera data directly from a central data server, or read images from an unusual file format. This can be done by writing a small python module which plugs in to calcam and handles the image loading. A full guide to writing such modules can be found in the Image Sources developer documentation page. Once written, they can be added to Calcam with the Calcam Settings interface.
Upgrading from Calcam 1.x¶
The update from Calcam 1.x to Calcam 2 includes large overhauls to the file formats, file storage conventions and Python API. This section covers the main things users need to know when upgrading from Calcam 1.x to Calcam 2.
In Calcam 1, CAD model definitions, other user-defined code, calibration input and results files were stored in a pre-prescribed directory structure. In Calcam 2 this is no longer the case; these files can be stored wherever you want and are opened either by graphical file browsing in the Calcam GUI or by file path in the Calcam API. The main change required to code calling Calcam to accommodate this will be that calibration results will now need to be loaded by supplying the relative or full path to the results file, rather than just the identifying name as before.
Whereas in Calcam 1, imported images, point pairs, calibration and fit results were all stored in separate files, in Calcam 2 all of these elements are stored together as a calibration. This is to maintain better traceability of calcam calibrations and make it easier for users to share data. Except for
.csv point pair files, Calcam 2 is not backwards compatible with Calcam 1 files, therefore to use existing data from Calcam 1 you must convert your Calcam 1 data to the new Calcam 2 formats. This can be done in bulk using the file converter utility provided in the
calcam1_file_converter directory of the calcam 2 repo. Running
convert_files.py from this directory as a script will open the tool, which is shown below:
At the top of this window, the “Source Directory”, where the tool will look for Calcam 1.x files to convert, is displayed. This is typically detected automatically, but you can also manually set the source directory manually using the Browse… button (this should be the complete Calcam 1.x data directory, i.e. the location of the
PointPairs etc directories).
Below this are 2 main sections: the top section for converting existing calibrations, and the bottom section for converting existing CAD model definitions. When the Convert! button is clicked in the relevant section, the large status bar at the bottom of the window will show the current progress during the conversion. The three text boxes containing file paths are used to specify where the output Calcam 2 calibration files should be saved to, since in Calcam 2 this can be wherever you want.
When converting calibrations, if the Try to match with image files based on name checkbox is ticked, the tool will try to match up calibration results with images by looking for Calcam image save files whose name also appears in the name of the calibration result being converted. If such an image is found, the image will be added to the resulting Calcam 2 save file. To disable this auto-matching, un-tick this checkbox, and Calcam 2 calibration results converted from Calcam 1 files will simply not contain any images.
Note: the conversion process does not alter or remove any of the original Calcam 1 data, so if anything goes wrong and you have to, or want to, go back to using Calcam 1.x, the data will still be intact, and it is left to the user to remove the old Calcam 1 data when you feel sufficiently comfortable to do so.
API Changes Summary¶
The change from Calcam 1 to Calcam 2 includes several compatibility breaking API changes. The main changes to the API are:
calcam.CalibResultsclass has been superceded by the new
calcam.Calibrationclass. This maintains the methods for working with calibration results which existed in
calcam.CalibResults, with the addition that
calcam.Calibrationnow contains data on the entire calibration process: image, point pairs, fit results and metadata.
calcam.VirtualCalibclass has been removed: virtual calibration results are now represented by the new
calcam.Calibrationclass, meaning all types of calibration use the same class in Calcam 2.
RayCasterclass has been removed. This is because although more functionality was originally envisaged for this class, that additional functionality is no longer planned for Calcam and therefore only a single method of this class was ever useful. In addition, the important element of this class’ state was already being held by other objects. The functionality of the
RayCasterclass has been moved to the function
machine_geometrymodule has been removed. Now instead of every CAD modeling having its own class inside calcam.machine_geometry, the
calcam.CADModelclass is used for all CAD models and is instanciated with string arguments to specify the name of the model you want. Also there have been various changes to method names and call signatures in the CAD model class.
Naming conventions: to be more Pythonic, throughout the API argument or function names which previously used capital letters and
camelCasehave been changed to
lowercase_with_underscores, while class names keep