Plots with different axes

[nsotom]

Hi all:

First of all, thanks for making this amazing project. I have been looking for a long time for an open source optical design software such as this one.

I'm trying to understand how to use optiland to design non-rotationally symmetric systems. The tutorial is a very good start, but I'm struggling to plot the 2d view and in particular the spots diagrams for a pair of axes other than the default.

For example, if in the tutorial system I remove the last mirror and plot the spot pattern in a shifted surface in the x-z plane , I get a
"side" view of the spot diagram in the x-y plane:

import numpy as np

from optiland import analysis, optic

lens = optic.Optic()

# System settings
lens.set_aperture(aperture_type="EPD", value=5.0)
lens.set_field_type(field_type="angle")
lens.add_field(y=-0.5)
lens.add_field(y=0)
lens.add_field(y=0.5)
lens.add_wavelength(value=0.633, is_primary=True)

# Lens data
lens.add_surface(index=0, radius=np.inf, z=-np.inf)
lens.add_surface(index=1, z=0)
lens.add_surface(index=2, z=10, material="bk7", surface_type="standard")  # material in
lens.add_surface(index=3, z=15, material="air", surface_type="standard")  # material out
lens.add_surface(
    index=4,
    z=25,
    material="mirror",
    rx=-np.pi / 4,
    is_stop=True,
)  # first mirror
lens.add_surface(
    index=5,
    z=25,
    y=-15,
    material="mirror",
    rx=-np.pi / 4,
)  # second mirror
lens.add_surface(index=6, z=45, y=-15, material="mirror", rx=np.pi / 4)  # third mirror
lens.add_surface(
    index=7,
    z=45,
    y=-10,
    radius=-30,
    material="bk7",
    rx=np.pi / 2,
)  # material in
lens.add_surface(
    index=8,
    z=45,
    y=-5,
    radius=30,
    material="air",
    rx=np.pi / 2,
)  # material out
lens.add_surface(index=9, z=45, y=20, rx=np.pi/2)  # x-z image plane

lens.draw(title="System 2D Layout")

spot = analysis.SpotDiagram(lens, fields='all', num_rings=5, distribution='hexapolar')

spot.view()

I tried looking into spot.data, but only x,y coordinates are available. I also tried, unsuccesfully, to get the fields to propagate down from y=+np.inf to rotate the whole system.

Is there a way to tell Optiland to change the default axes for its diagrams? (I'm thinking something like plotting the diagrams in the local coordinate system of the surface). A similar issue would appear when trying to use the draw() method for a system with elements rotated in an axis other than x.

Thanks in advace!

[HarrisonKramer]

Hi @nsotom,

Thanks for the kind words.

There are a few things to point out here. Optiland assumes the optical axis is along z, with light propagating in the +z direction. This currently cannot be configured otherwise. Second, as you pointed out, the SpotDiagram data does not include z points. There is not currently a way to tell Optiland to change the axes for diagrams, but this is certainly something that could be added. How would you like to see that implemented? I can imagine something like the following:

spot = SpotDiagram(lens, fields=[(0, 1)], axes="xz")
spot.view()

This would not be too difficult to implement for the SpotDiagram. It may be more difficult for other analyses, but I would need to investigate this.

If you'd like a workaround, you can manually retrieve, then plot ray intersection points using matplotlib. Retrieving x, y, z data is straightforward and is covered in this tutorial.

As for plotting on other axes, this is also not supported, but in theory is also possible. I will consider adding this to the roadmap, but it will not have high priority. If needed, plotting in 3D may be a suitable alternative.

Regards,
Kramer

[nsotom]

Thanks @HarrisonKramer !

Yeah, initially it would be good to have the option to set another pair of axes specified as you suggest. Altough in that way the spot diagram would still be restricted to planes which are orthogonal to the x,y,z axes.

A more general approach would be plotting the diagrams in the local coordinate system of the surface. Something like

spot = SpotDiagram(lens, fields=[(0, 1)], axes="local")
spot.view()

I think having that as the default would probably be ok in almost all cases. I'm not familiar with the data structure used by optiland to know how difficult would be implementing something like that.

I totally overlooked the linked tutorial, I'll have a look at it and also to the source code to see if I find a quick way of doing the local axis plotting.

Cheers.

[HarrisonKramer]

Hi @nsotom,

Thanks a lot for the suggestion. I like the approach of plotting the spot diagram in the local coordinate system by default.

As for implemented this, I think it's as simple as

• Adding an argument to SpotDiagram to choose local or global coordinates for plotting.
• Using optiland.visualization.utils.transform to transform the global intersection points into local coordinates, if local coordinates are specified.

Perhaps a coordinates argument could be added ("axes" is also fine):

spot = SpotDiagram(lens, fields=[(0, 1)], coordinates="local")  # or "global"
spot.view()

I'd gladly merge a pull request if you (or any other contributor) want to work on this. In any case, this will be added to the roadmap.

-Kramer

[manuelFragata]

Hi all,

I believe I have submitted a PR for this fix a few months ago. See here:
#68

Isn't it what is wanted?
Which version of Optiland did you use?

Best,
Manuel

[manuelFragata]

Hi @nsotom :)

As I mentioned in my other message, I believe you had a different version of Optiland where this bug was still not fixed.
In the current version, you can rotate the image plane easily, for instance (following the example in the tutorial and removing the last mirror like you suggested) by simply passing rx=-np.pi / 2 (-90° rotation). If you then check the spot diagram, it will look something like this:

Like you and @HarrisonKramer suggested, I also like the idea of passing coordinates='local' to spot = SpotDiagram(lens, fields=[(0, 1)], coordinates="local") . It shouldn't take so long, at least for the spot diagram. I'll get started with it and keep you updated here :)

[manuelFragata]

Hi all, already pushed the changes to the current master branch. Let me know if it is what you expected :)

[HarrisonKramer]

Awesome, thanks for adding that, @manuelFragata !

[nsotom]

I just tried with version 0.4.1 and it works perfectly. Thanks @manuelFragata