March 13, 2018
Optical engineers can save time using Contrast Optimization
Contrast Optimization drastically increases the speed of MTF optimization and results in better optical designs.
Most imaging systems, including digital cameras and microscopes, include Modulation Transfer Function (MTF) specifications. Until recently, optimization of the MTF performance was widely known as a time-consuming and difficult process. Contrast Optimization, however, drastically improves the speed of MTF optimization, and in addition, results in better design candidates.
Optimizing on MTF had been too slow
Optimization techniques are used to find the best optical system that satisfies the required specifications. Optimization involves changing system parameters by a small amount, checking if the change improved the performance, and repeating. Until recently, the only way to optimize for MTF was to first optimize on the wavefront error, and then to optimize directly on the MTF value. This is an inconvenient multi-step process that is computationally expensive and often unsuccessful. Even a single MTF calculation is slow, because it requires calculating an autocorrelation of the wavefront in the exit pupil of the imaging system. Optimization is even slower, however, because this MTF autocorrelation calculation is repeated thousands of times. Therefore, designers often leave MTF optimizations running for many hours at a time.
Contrast Optimization speeds up MTF optimization by 10x or more
Contrast Optimization, a capability in OpticStudio 17 and later, speeds up MTF optimization by at least a factor of 10. Contrast Optimization targets the best MTF response up to a specific spatial frequency, but doesn’t require the slow autocorrelation calculation that was previously used. The new contrast calculation is based on a simple calculation of the wavefront difference between two rays in the exit pupil of an optical system. The rays are separated by a distance determined by the specified spatial frequency, and minimizing the wavefront difference between these pairs of rays maximizes the MTF at that spatial frequency.
“Optimization of optical systems constitutes the daily routine for an optical engineer. Fast Fourier Transform and Huygens’s methods are the common merit function operands of choice for calculating MTF at specific frequencies. While these standard tools offer great results, the optimization routines for complex systems can take numerous hours. Contrast Optimization brings a great deal of efficiency to optimization! It avoids the computational drain from calling repeated auto-correlation functions that were required for MTF optimizations.”
- Dr. Ravi C. Bakaraju, Head of Research & Development, Brien Holden Vision Institute
Contrast Optimization brings additional benefits
More information = better solutions.
Contrast Optimization is faster than other techniques, but it also finds better solutions because it has access to more information. Optimizations work best when calculating values that are descriptive of system performance and that are responsive to changes in the variable system parameters. Previous MTF optimizations could only predict whether the MTF value at a spatial frequency would go up or down after changing a system parameter. Contrast Optimization, however, traces pairs of rays across the exit pupil, and thus calculates more descriptive information. This contrast calculation easily identifies areas the exit pupil that correspond to poor MTF performance, which means that the optimizer can adjust the system parameters to improve MTF across the pupil.
Smoother parameter space = faster convergence.
Contrast Optimization converges faster than other MTF optimization techniques. The contrast calculation is essentially calculating a derivative of the wavefront, which has a much simpler optimization parameter space than the direct MTF calculation. With a simplified parameter space, the optimizer is less likely to get stuck in a local minimum and is more likely to quickly converge to a good solution.
It works when MTF calculations fail.
Contrast Optimization is also more reliable than previous MTF optimization techniques. Early in the design process there is typically a significant amount of wavefront error, and this causes direct MTF optimization to fail. It fails because the MTF does not increase or decrease smoothly as the variable system parameters are changed during optimization. The system performance may need to get worse (in a local minimum) in order to find a design candidate with better overall performance. Contrast Optimization, however, utilizes wavefront data in the exit pupil, which is less likely to contain local minima. Therefore, Contrast Optimization can be used early in the design process.
See for yourself how Contrast Optimization enables:
Specific targeting: You can target a specific spatial frequency in the MTF
Faster calculations: Much faster than calculating MTF directly, and almost as fast as wavefront optimization
Reliability: Works early in the design process, and even when direct MTF optimization cannot be used
Faster convergence: Contrast Optimization simplifies the parameter space so that optimizations reach a good solution more quickly
< Return to blog