The Huygens PSF is computed on a flat patch of image surface, centered on the chief ray but normal to the surface, not the chief ray (as the FFT PSF is, for example). The size of the patch is determined by the image sampling and the image delta. If the image surface is curved, then the PSF patch is tangent to the image surface at the chief ray location.

Hi Mark,

Thanks for the information. 

Could you confirm that if I want to render from an ideal image, I should apply distortion first and then PSF, right?

Thanks,

Qiang

The PSF is a function of the wavefront error, and so is independent of distortion. I don't think it matters which you do first as a result. In the Image Simulation feature OpticStudio does the convolution of the image with the grid of PSFs, and then accounts for the image distortion. I'm not sure it really matters which order you do it in.

In the Image Simulation feature you can also replace the PSF grid with a grid of delta-functions to get diffraction-limited perfromance.

Hi Mark,

Thanks for the quick reply. 

The grid locations as well as the size of image area that the PSF patch covers are different before or after distortion. Would that change the resulting image, especially at the cornor of the image where there is a lot of distortion?

Huygens PSF is calculated by coherent ray tracing, right?

Would that account for the distortion already?

Thanks,

Qiang

Hi Qiang,

The grid is located where the real chief ray lands on the image surface, and so that will be the distorted (real) location and not a paraxial (undistorted) location. The Huygens PSF is computed by coherently adding the rays from the pupil as they jand in the image grid.

- Mark

Hi Mark,

Let's say you have a PSF grid point at [x, y] = [1, 1] mm in the image space and a PSF patch of size [width, height] = [0.05, 0.05] mm square also in image space (after distortion). The corresponding point before distortion might be at [x_ideal, y_ideal] = [1.2, 1.2] mm. If you were to apply the PSF at this point before distortion, the PSF patch size and shape would not be [width, height] = [0.05, 0.05] mm square in undistored space. Please see the attached image. How do you apply it before distortion?

Thanks,

Qiang

I don't really follow your question. The rays land where they land...these are real rays. The image grid is centered where the real, aberrated, chief ray lands. Use the single ray trace calculation to trace the chief ray, and note it's location on the image surface. That is the location of (0,0) in the PSF, and the normal of this location points along the normal to the image surface.

Yes. The (0,0) in PSF is at the correct location. Can a PSF matrix, calculated in the distorted image space, be convolved directly to an undistorted image?

Hi Qiang,

As Mark has noted, the PSF matrix is centered on the real aberrated chief ray, (or distored chief ray), location on the image surface. If you want to create a PSF matrix with PSF from each field points centered on the 'undistorted' chief ray, you'll have to do some post data processing yourself to figure out where the paraxial chief ray landing positions are and then re-create this PSF matrix. You can find the paraxial chief ray location either using Single Ray Trace analysis or use operand PARX PARY from each field points.

Best regards,

Hui

Hi Hui,

Thanks for helping out here.

My question is not how to create PSF. It is which order do you apply PSF vs distortion.

Mark said ' In the Image Simulation feature OpticStudio does the convolution of the image with the grid of PSFs, and then accounts for the image distortion. I'm not sure it really matters which order you do it in.' I don't understand how you can apply PSF on undistorted image first before distortion. Can you explain what steps are involved in such an operation? For example, how was PSF generated and how they are applied to the undistorted image.

Thanks,

Qiang

The steps followed are described in the manual and at https://my.zemax.com/en-US/Knowledge-Base/kb-article/?ka=KA-01359 Also, if you search for Image Simulation in the MyZemax search you'll find lots of links to OpticsTalks and Envision discussions of Image Simulation, several of them really in-depth.

The key point is that wavefront error and distortion are two separate things. Distortion affects the landing coordinates of the chief ray, and wavefront error describes the phase relationships of the other rays to that chief ray.

- Mark