This is a fork to keep track of my changes and bugfixes to the google code project pyoptic (code.google.com/p/pyoptic) which now seems to be unmaintained. Pyoptic is a python-based raytracing/optical CAD package which works from a programmatic definition of the optical system. It was originally envisaged as a 'toy' tracer for educational purposes, but is arguably capable of being used in some (limited) production uses. I use it for testing the arrangement of optical components in large systems such as microscopes.
By now, things have divereged a lot from the original pyoptic, motivating the change to pyoptic2.
pyoptic2 is released under the GNU GPL v3 license
- Spherical surfaces
- Mirrors
- 3D Visualisation
- Thin lenses (new in fork)
- Dispersion (new in fork)
- Import of Zemax lens files (e.g. for Thorlabs lenses) - reads geometry and glass information (new in fork)
There are also a number of bug fixes and visualisation tweaks that are new in the fork.
Pyoptic requires a number of dependencies:
- Python >= 3.6
- Numpy, Scipy, Matplotlib
- Mayavi2 (optional, for 3D visualisation)
A python distribution such as Enthought Canopy, WinPython, Python(x,y), or Anaconda ought to provide all of these.
After downloading and installing python & dependencies, pyoptic can be installed by executing:
python setup.py install in the pyoptic base directory.
The examples directory has a couple of ipython notebooks with comments:
pyoptic_testsv2.ipynb : Sets up a simple, folded 4-f system using catalog lenses. This is a good place to start. olympus_25x_objective.ipynb: Simulates a microscope objective (based on patent literature). This is a bit more involved, and less well documented, but is an example of how the package can be used for something practical. It shows how to construct systems from individual surfaces.
These packages are needed to automatically grab Zemax files from Thorlabs:
- Requests, Beautifulsoup4, Lxml
There are a few things that you might find in professional optical CAD that pyoptic does not yet do
- Aspherical surfaces (this would be resonably simple to add)
- Polarization
- Fresnel coefficients (or any other form of intensity tracking)
- Non-sequential ray tracing
- Both reflection and transmission at the same surface (this can be spoofed in order to model, e.g. dichroic mirrors separating excitation and detection paths in a microscope by creating a copy of the system up until the split)
On the other side of the coin, there are a few things which we are particularly well equipped to do. Because system definition is programatic, it is very easy to generate parametric systems, and, by leveraging the numpy/scipy/matplotlib stack, to generate plots of characteristics which might be specific to your system.