As manufacturing capabilities have developed, the potential to create complex geometries for optical elements of a variety of materials has become a reality. Under these conditions, freeform optical components have become more prevalent in the industry, and have allowed a new class of optical systems to be engineered. In the past, freeforms have commonly been described parametrically for optimization, which in some cases limits their flexibility and utility. In this paper we demonstrate a novel freeform design methodology which uses a grid-based sag definition to serve as an alternative to polynomial coefficient optimization. The OpticStudio TrueFreeForm model allows direct optimization of sag values at grid points across the surface to enable spatially-selective surface sag optimization. Here, a case study is presented on the effectiveness of TrueFreeForm optimization in the design of freeform optical surfaces for use in modern augmented reality/virtual reality (AR/VR) optical systems. We demonstrate the flexibility and advantages of grid sag optimization and discuss key implementation considerations in the context of a freeform wedge prism system as might be found in an AR headset. In particular we demonstrate, via sub-aperture grid sag optimization, the integration of an eye-tracking subsystem into the AR wedge prism. An average MTF of 0.446 at 40 lp/mm is achieved in the user pupil imaging portion of the eye-tracking system, while introducing no degradation in the nominal performance of the AR visual display system.
In this work, we discuss grid-based surface optimization in the OpticStudio lens design software. This new approach allows flexibility in designing freeform optical components that is not possible with traditional surface types. The system we focus on is a corrector plate with the rear surface specified by a variable grid sag surface. The goal of the corrector plate is to correct for wavefront error introduced by a Zernike phase surface. We optimize the sag values of the grid points to correct for this wavefront error and study the resulting behavior as the number of grid points are varied. We test grids from 5x5 to 25x25 and report the scaling of total computation time and improvement in wavefront error. We describe several findings, such as the importance of constraints on the grid variables and techniques that improve the calculation efficiency.
Traditionally optical design via computer optimization uses a numerical merit function to represent the optical performance of the simulated system. The conventional design approach is to maximize the nominal performance of the design, and then as a separate step, add fabrication tolerances to the nominal parameters so that upon manufacturing the resulting system still performs to specification. This paper will demonstrate an alternate approach. Because the angle rays make with respect to the normal on each surface are the primary drivers of optical aberrations and tolerance sensitivity, the method uses these ray angle as a fast, numerical approximation for the sensitivity to tolerance defects. This hybrid merit function thus includes the fabrication errors as part of the design process. The resulting design is effectively optimized for as-built, rather than nominal performance. Design examples will be provided which show that optimization using the hybrid merit function yields designs of different forms, which may have inferior nominal performance but superior as-built performance. The resulting alternate designs will be compared to conventional post-design tolerance analysis to demonstrate the reduction in tolerance sensitivity and superior resulting performance.
Our new Contrast Optimization technique allows for robust and efficient optimization on the system MTF at a given spatial frequency. The method minimizes the wavefront differences between pairs of rays separated by a pupil shift corresponding to the targeted spatial frequency, which maximizes the MTF. Further computational efficiency is achieved by using Gaussian Quadrature to determine the pattern of rays sampled. Examples are given to demonstrate the advantages of the technique.
Reverse ray tracing from a region of interest backward to the source has long been proposed as an efficient method of determining luminous flux. The idea is to trace rays only from where the final flux needs to be known back to the source, rather than tracing in the forward direction from the source outward to see where the light goes. Once the reverse ray reaches the source, the radiance the equivalent forward ray would have represented is determined and the resulting flux computed. Although reverse ray tracing is conceptually simple, the method critically depends upon an accurate source model in both the near and far field. An overly simplified source model, such as an ideal Lambertian surface substantially detracts from the accuracy and thus benefit of the method. This paper will introduce an improved method of reverse ray tracing that we call Reverse Radiance that avoids assumptions about the source properties. The new method uses measured data from a Source Imaging Goniometer (SIG) that simultaneously measures near and far field luminous data. Incorporating this data into a fast reverse ray tracing integration method yields fast, accurate data for a wide variety of illumination problems.
This paper describes a computational approach to handling CAD objects within the framework of optical design software. Very fast ray-tracing speeds can be achieved through realistic objects used in the manufacture of optical systems.
Ray-tracing codes used for stray light analysis or illumination system design often require access to complex object shapes. This is sometimes achieved by designing objects in CAD packages, and then importing them into the ray-tracing package. The disadvantage to using imported CAD objects is flexibility, ray-tracing speed, and in some cases, ray-tracing accuracy. This paper describes a new approach in which a user-defined object is used. The user-defined object is an external, user-supplied program called a Dynamic Link Library or DLL. The advantages to defining an object using a DLL, rather than the other methods listed above, are described and examples given which compare the two approaches.
Global optimization of optical system designs is difficult because of the complex solution space and extreme non- linearity of system performance with respect to design parameters. Genetic algorithms (GA's) have been used to effectively solve complex non-linear optimizations in other fields of science. GA's model each design parameter as part of a genetic code for an imaginary creature, and the system performance is interpreted to be the fitness for survival of this creature. The GA's goal is to improve the system performance by selective breeding of simulated creatures to yield simulated offspring with desirable traits. An algorithm for doing this along with results will be presented.
LightPath has been actively commercializing a family of axial-gradient glasses called the GSF glasses. These glasses have the same dispersion and refractive indices found in the conventional SF glasses. Therefore, in polychromatic design, it has only been possible to apply the extra degrees of freedom offered by GRADIUM® to negative elements by replacing a flint glass. This has shown that GRADIUM can significantly and positively affect polychromatic designs. For example, a seven element double Gauss lens was redesigned with GRADIUM, resulting in a six element design and a 27% improvement in rms spot size.1 Similarly, the Kodak Ektapro varifocal projection lens was redesigned reducing the element count from seven to five. Despite these impressive results, since most of the work done by any polychromatic system is done by positive elements— typically crowns, it is clear that the real potential of GRADIUM to polychromatic design could not be unleashed until GRADIUM crown materials were available. The GRADIUM GK glass family is one of several GRADIUM crown families under development. A number of examples will be presented to illustrate the advantages realized in optical design when GK glasses are used. Simple doublets as well as more complex multi-element systems will be discussed, some of which use both GK and GSF glasses and others which use only GK or GSF glasses.
Recent advances in the manufacturing of large (up to 62 mm diameter) glass blanks with axial index gradients of up to 0.20 enable significant performance gains in optical designs. The new axial gradient material has an index and dispersion similar to SF (short flint) glasses, except the index varies continuously through the axis of the blank. The axial gradient material and numerous design examples using the new material are presented.
This paper reports results of a program to integrate geometrical and physical optics. Limitations to the theory, numerical considerations, and methods of optimization are discussed and the procedures illustrated by several examples.
Waveguide grating couplers have important applications to optical data storage. Focusing grating couplers eliminate the need for focusing optics and can be used in both read and write modes. It is convenient to fabricate the gratings at short visible wavelengths although they are used at laser diode radiation wavelengths in the near infrared to visible range. This paper discusses a design technique for introducing compensating aberrations during fabrication which account for the wavelength change. A sample design is presented and the tolerance limits on the alignment parameters are derived. 2. 0