Visualization of car windshields

Visualization of car windshields



Eclat-digital provides software, service and expert guidance on computer visualization of advanced materials, such as car paints or car glazings, which have a complex interaction with light and are especially difficult to render. In this example, we will focus on car windshields.

Two kind of windshields


The most common type of windshield is made of two glass sheets, laminated together with a polymer sheet in between. The glass sheets are typically 2.1mm thick, while the polymer sheet is most often 0.76mm thick. The polymer, generally PVB, is chosen to have a refractive index close to the glass. We have several choices for modelling this optically:

  • Model every sheet and interface. The software will compute the reflection at each interface and provide an exact result
  • Model as a plain material, with only two glass-air interfaces. The reflections at the glass-polymer interfaces will be neglected. This is reasonable, as the glass-polymer reflection coefficient is very low, below 0.01% with a refractive index difference of 0.02. The optical absorption of the materials can be accounted for exactly, by calculating an equivalent absorption spectrum for the full structure.
  • Pre-compute the windshield reflection and transmission spectra at any angle using Fresnel coefficients, and use them as tabulated values for the full windshield structure

The three approaches are possible within Ocean. The first one consists in modelling every interface in a modelling software, and assign each space a dielectric material with the n and k properties of glass and polymer. The second one would use a single dielectric material, with average refractive index n, and integrated equivalent k coefficient resulting in the same extinction coefficients at any angle. The third one would use the specular tabulated data model, where reflection and transmission spectra are given for several angles between 0° and 90° to normal incidence.

The first method would result in significantly slower calculation, for no real benefit (the third is exact as well, with enough data points). For this study, we have chosen the second method as it is very quickly implemented. The error on reflection coefficients is much lower than 0.1%, which is visually negligible compared to the 6-8% reflection coefficient of the windshield.

The optical properties of PVB were found in the literature. The properties of glass used in windshields vary between manufacturers and countries, it is generally tinted between gray and green to provide shading, while staying in the transmittance regulations (>70% in US/Japan, >75% in Europe). Therefore, the R and T spectra at normal incidence were measured, which allows calculating n and k at any wavelength.


More and more cars are equipped with athermic windscreens. They differ in a very thin silver coating added on one of the two glass sheets, bound to the polymer. This coating acts as a low-pass filter and reflects most of the solar infrared radiation, while transmitting most of the visible light. As such, they should not be visible to the human eye. But by design, they are not perfect filters, and modify significantly the optical properties in the visible wavelength range. The chart below shows the reflection spectrum of an athermic windshield, compared to a standard glass windshield:

Reflection spectrum of an athermic car windscreen compared to a standard one

The reflection is clearly low in the visible (380-780 nm) and high in the infrared (>780nm), but the reflection color of this windscreen is clearly not neutral, with higher values around 400nm and also 700nm, it will clearly look between purple and blue.

This complex color effect of athermic windshields is highly varying across manufacturers. Therefore, there is not “one” good optical model for athermic windshields, but many. In order to correctly visualize a given windshield, measurements must be performed on a matching sample. The images following in this article are not representative of all athermic windshields, only of the sample we performed the measurements on.

For rendering the car images, the windshield spectra were taken from 380nm to 780nm by steps of 10nm, for 17 angles between 0° and 90°, and for both s and p polarizations. All this data is given to Ocean as a material model.

Other modelling elements

The car model

The model is an Audi R8 model in IGES format, freely available on GrabCAD. It was triangulated with Rhino 5, the resulting triangle count being approximately 8.6 million.


The car paint material is an advanced, physically based metallic flakes paint model that will be the topic of a future article. Side and rear glazings use a 4mm green glass model based on measured spectra. Other materials are mostly non physically based, and were adjusted to provide nice results, nothing more.


All images were rendered using spherical HDR environment images for both light source and background.


Software performance

Despite the high polygon count and material complexity, Ocean renders the overcast sky images within 5 minutes in Full-HD on a single 12 core workstation. Sunny environments take two or three times longer. The memory use for these tests is 5GB for geometry, plus 1-2GB for the environment map.

Visual impact of athermic windshield

Move the mouse over the picture to compare

Is the tint more visible with different car paints?

The pictures below compare the same exact scene, the only difference being the car paint.

A dark car with athermic windshield

The bluish tint of the windshield is very visible on the dark car, as it is brighter than the car itself.

A white car with athermic windshield

The windshield tint is less visible on the white car, as the its relative brightness appears much lower.

Possible applications

We have shown here two simple examples of realistic rendering of car windshields achievable with the Ocean light simulation software. This technique could be used for better design of athermic windshields, in order to give them a better visual acceptance. As tints vary between manufacturers, the visual impact on car appearance could be compared between them. And finally, the actual windshield appearance could be more integrated into car design works.

Ocean is also capable of computing the amount of light energy hitting a surface, it could be used for evaluating the actual windshield performances for heat control. Windshields are specified with an energy transmittance (or shading coefficient) at normal incidence. This simplistic figure may differ significantly from the actual performance, with various sun orientations, windshield angles and shapes. On the same 3D model than the one used for visual renderings, Ocean is able to compute the amount of energy entering the vehicle.

Please contact us for any inquiry regarding light simulation for automotive applications!