Raytracing / Simulating LED in enclosed compartments

[Korabel47]

Hi to all and big thanks to @HarrisonKramer and the other project contributors for this cool software!

However, honestly saying, I am yet to fully understand this tools potential and capabilities, as I do not possess a good education in optical engineering and am currently teaching myself Python.
I would like to simulate, in 3D, light propagation (shown by a finite number of beams) from a few LEDs in an enclosed compartment, like a box.

1. The objects I plan to have there, include:

• mirrors
• lambertian surfaces
• LEDs
• focal planes, or light beams projection planes.
  

2. The current problems for me now are:

• how to create an impenetrable (I remember about the Materials class in Optiland) box and other physical obstacles inside it
• how to code propagation of the rays so that their intensity/power change with the traversed distance (thinking about limiting the beams length by some f(distance, power, etc.))
• how to express rays' power onto the projection planes once those, filtered out by their power, reach it.

I would like to use primarily Optiland's defined classes and functions and write additional if needed.

---

I researched a bit about the similar open-source software used for raytracing, and found pyVista, but it seems to become very resource-hungry during multiple-LEDs simulations. I hope Optiland might be more optimal choice for me here, but at the moment I do not know how to proceed...

I would greatly appreciate any advice, hints, and pointing to good references!

[HarrisonKramer]

Hi @Korabel47,

Thanks for the question and the kind words.

First, I'll point out that Optiland uses sequential raytracing, which means you define the order in which rays propagate through the system. From your description (and the drawing), it sounds like you want to perform non-sequential ray tracing. In other words, it seems you want to simply define the sources and objects, then let rays propagate through the system in any order. It's not impossible to do this in Optiland, but it would take quite a few changes in the code to be able to handle this.

Can you elaborate a bit more on what specifically you want to compute here? What's the background for this problem? I think that will help me provide a better answer. If you really need to perform non-sequential ray tracing, then you might have an easier time with other programs. Depending on your goals, a ray tracer designed for 3D computer graphics might also work. Also, Optiland does not natively support sources like LEDs, although this could be added without too much difficulty. This is much more in the realm of non-sequential raytracing. While I realize you may not have access to commercial tools, this sort of problem can be solved very quickly in programs like Fred, OpticStudio, etc.

If you do want to use Optiland (in its current state), then you'd need to trace only one ray path at a time. You could define several (100s, 1000s, etc.) of possible ray paths, simulate each independently, then combine them all at the end. You'd also need to extend the ray generation functionality to be able to generate something LED-like.

Hope that helps. I can provide more context/info, if needed.

Regards,
Kramer

[HarrisonKramer]

Thanks for the clarification. While this is possible, I will admit that you may need to make significant extensions to the code to build this. I would encourage you to investigate other options too, especially if your goal is to get something working as fast as possible.

If you're keen on using Optiland, then I would propose that you look further at the example gallery and learning guide examples, and perhaps even the developer's guide. This will give you an idea of how to customize/extend the code for your use case. I do not know how complex the obstacles are in your scene. If they are not simple geometries, e.g., if they are CAD objects or complex multi-surface components, then raytracing in a "pseudo-sequential" way may be impractical.

As for sources, you could create your own class for the LEDs that generate rays. These rays could be passed to the ray tracing facility in the standard manner. More info here. And yes, you can certainly define multiple beams at the same time, but they must follow the same path through the system. You could use the strategy I outlined above.

Hope that helps. If you find a nice solution, with or without Optiland, I would be interested to hear more.

Kramer

[Korabel47]

Thanks for the directions! Will update my progress here some time later, hopefully.
But I would like to fork your repository to experiment with the code.

[Korabel47]

Hi @HarrisonKramer,

I'm still struggling to comprehend how an optical system can be built in Optiland.

While digging through the posted issues and the tutorials, I found something I'd like to use as a concept:

1. Decentered reflective systems #42 - as an example of a decentered system
2. https://optiland.readthedocs.io/en/latest/gallery/reflective/three_mirror_anastigmat.html
3. https://optiland.readthedocs.io/en/latest/gallery/reflective/laser_system.html

In my attempt to make a first draft of something like what I described earlier, I wanted to create an optical system consisting of:

• a few rays sources with added fields angle + wavelengths each directed at a specific angle,
• multiple image planes to take readings from (distribution analysis).

I use the following code to start with:

Code snippet

    ```python

    lens = optic.Optic()
    # # define the physical aperture for the off-axis parabola
    # aperture = OffsetRadialAperture(r_max=17, offset_y=45)

    # Constructing optical system
    # Basis
    lens.add_surface(index=0, radius=np.inf, z=-np.inf)
    lens.add_surface(index=1, y=100)
    # LED-source
    lens.add_surface(index=2, z=10, radius=np.inf, is_stop=True, material='mirror', rx=np.pi/4)
    lens.add_surface(index=3, z=10, y=25, radius=np.inf, material='mirror', rx=np.pi/4)
    lens.add_surface(index=4, radius=-40, y=-15, z=150, material='mirror', conic=-1, aperture=aperture)
    lens.add_surface(index=5, y=-15, z=131, rx=np.pi/2)

    lens.set_aperture(aperture_type='EPD', value=15.0)

    # Defining rays' properties
    lens.set_field_type(field_type='angle')
    # Typical (red) wavelength - from datasheet of CH DELSS1.22 LED
    lens.add_field(y=0)
    lens.add_wavelength(value=660, unit='nm', is_primary=True)
    # Minimal (red) wavelength - from datasheet of CH DELSS1.22 LED
    # at 60-degrees half-angle of view
    lens.add_field(y=10)
    lens.add_wavelength(value=648, unit='nm', is_primary=False)
    lens.add_field(y=-10)
    lens.add_wavelength(value=648, unit='nm', is_primary=False)


    lens.draw(num_rays=5)
    
    ```

I get then an error saying:

Error msg

```python

        ---------------------------------------------------------------------------
        ValueError                                Traceback (most recent call last)
        Cell In[11], line 8
              2 # # define the physical aperture for the off-axis parabola
              3 # aperture = OffsetRadialAperture(r_max=17, offset_y=45)
              4 
              5 # Constructing optical system
              6 # Basis
              7 lens.add_surface(index=0, radius=np.inf, z=-np.inf)
        ----> 8 lens.add_surface(index=1, y=100)
              9 # LED-source
             10 lens.add_surface(index=2, z=10, radius=np.inf, is_stop=True, material='mirror', rx=np.pi/4)

        File optics_VENV/lib/python3.11/site-packages/optiland/optic/optic.py:149, in Optic.add_surface(self, new_surface, surface_type, comment, index, is_stop, material, **kwargs)
            119 def add_surface(
            120     self,
            121     new_surface=None,
           (...)    127     **kwargs,
            128 ):
            129     """Adds a new surface to the optic.
            130 
            131     Args:
           (...)    147 
            148     """
        --> 149     self.surface_group.add_surface(
            150         new_surface=new_surface,
            151         surface_type=surface_type,
            152         comment=comment,
            153         index=index,
            154         is_stop=is_stop,
            155         material=material,
            156         **kwargs,
            157     )

        File optics_VENV/lib/python3.11/site-packages/optiland/surfaces/surface_group.py:244, in SurfaceGroup.add_surface(self, new_surface, surface_type, comment, index, is_stop, material, **kwargs)
            241     if index is None:
            242         raise ValueError("Must define index when defining surface.")
        --> 244     new_surface = self.surface_factory.create_surface(
            245         surface_type,
            246         comment,
            247         index,
            248         is_stop,
            249         material,
            250         **kwargs,
            251     )
            253 if new_surface.is_stop:
            254     for surface in self.surfaces:

        File optics_VENV/lib/python3.11/site-packages/optiland/surfaces/factories/surface_factory.py:71, in SurfaceFactory.create_surface(self, surface_type, comment, index, is_stop, material, **kwargs)
             68     raise ValueError("Surface index cannot be greater than number of surfaces.")
             70 # Build coordinate system
        ---> 71 coordinate_system = self._coordinate_factory.create(
             72     index, self._surface_group, **kwargs
             73 )
             75 # Build pre and post surface materials
             76 material_pre, material_post = self._material_factory.create(
             77     index, material, self._surface_group
             78 )

        File optics_VENV/lib/python3.11/site-packages/optiland/surfaces/factories/coordinate_system_factory.py:59, in CoordinateSystemFactory.create(self, index, surface_group, **kwargs)
             57 else:
             58     if self.surface_factory.use_absolute_cs:
        ---> 59         raise ValueError(
             60             'Cannot pass "thickness" after defining '
             61             '"x", "y", "z" position for a previous '
             62             "surface.",
             63         )
             65     thickness = kwargs.get("thickness", 0)
             66     x = kwargs.get("dx", 0)

        ValueError: Cannot pass "thickness" after defining "x", "y", "z" position for a previous surface.

    ```

My current issue is that I am trying to build something like this:

Image 1


When I played with the Hubble telescope example, I noticed that during a surface declaration with any big (>=100 mm) offset by X-axis, that surface just disappears - when changing x= kwarg for the index=3-surface. Example in the collapsible section below:

Hubble-example code

```python

    class HubbleTelescopeDecenteredXYZ(optic.Optic):
        def __init__(self):
            super().__init__()

            self.add_surface(index=0, radius=np.inf, thickness=np.inf)
            self.add_surface(index=1, x=0, y=0, z=0)

            # add physical aperture in primary mirror
            obscuration = physical_apertures.RadialAperture(r_max=np.inf, r_min=300)

            self.add_surface(index=2, radius=-11040.02286, x=30, y=-20, z=4910.01016,  # <-- add decentering in y here
                             material='mirror', is_stop=True, conic=-1.001152,
                             aperture=obscuration)
            self.add_surface(index=3, radius=-1349.31166, x=100, y=110, z=0,
                             material='mirror', conic=-1.483014)
            self.add_surface(index=4, radius=-635.38227, x=-20, y=0, z=6365.20955)

            self.set_aperture(aperture_type='EPD', value=2400)

            self.set_field_type(field_type='angle')
            self.add_field(y=0)
            self.add_field(y=0.15)

            self.add_wavelength(value=0.55, is_primary=True)


    hubble = HubbleTelescopeDecenteredXYZ()
    hubble.draw()
    hubble.draw3D()

```

Although I do sense that the root of a solution is in how surfaces and coordinate systems are used in Optiland, I cannot understand how to manually place objects in 3D-space. Do I have to declare separate local coordinate systems?
Is there a good, informative tutorial for this?

Thanks!

[Korabel47]

The image in the collapsible section Image 1 doesn't render when unfolding

[HarrisonKramer]

Hi @Korabel47,

The error you point out is due to you not passing a z value on surface index 1. You may have already seen this, but this tutorial gives a nice overview of how to define coordinate systems in Optiland.

I believe the clipping that you're seeing with the Hubble is due to rays being clipped by the physical aperture. If the rays are clipped, then they won't be rendered. When you move the mirrors significantly, then all rays will strike the physical aperture and won't propagate to the image plane. In some cases, you will also cause rays to deviate so significantly that they won't find an intersection with the surface. This also prevents rays from rendering.

You're defining the field and aperture for your system, which results in rays being generated in a particular way (the standard in Optiland). This works fine for typical systems, e.g., imaging systems. However, I don't know if this makes the most sense for your case. You may want to manually define rays instead and pass them yourself to the surface_group. You're doing things in a slightly different way than how I originally intended the code to be used, which means you might need workarounds in some cases. Unfortunately, that also limits how systems are rendered using the Optiland visualization framework, which assumes sequential systems with apertures and fields defined.

Ultimately, you need a SurfaceGroup object consisting of your surfaces and a RealRays object consisting of your rays. You can then simply call SurfaceGroup.trace(rays). This will trace rays through the surfaces one at a time. The SurfaceGroup can be built through an Optic instance (like what you're doing above), but it can also be done manually. You can look through the code directly to see how this is done, or look at the API/docs.

Regards,
Kramer