Selecting and Stacking

RegiStax 4, available as a free download from the Internet, is one of the most powerful and widely used programs for sorting, registering, stacking, and sharpening webcam images. Sean Walker described its key functions in his review of version 3 in the December 2005 issue, page 94. Like him, I find I can produce excellent images by using the program’s default settings but even better results if I exert some manual control.

After opening an AVI file from my webcam, I first select a reference frame of average quality, as this tends to produce the best alignment compared to using an exceptionally sharp one. If I recorded my target on a night of average seeing, I’ll use an Alignment box that encompasses the entire planet’s globe, selecting a frame near the middle of the movie to minimize rotation artifacts and choosing the Quality Estimate Method Local Contrast. If the seeing was good during the movie clip, I’ll instead choose a smaller Alignment box centered on an interesting planetary feature, such as Jupiter’s Great Red Spot or an albedo feature on Mars, for my registration point. I then select Gradient as the Quality Estimate Method. Once I’ve chosen my alignment point and Quality Estimate Method, a window opens that displays the FFT (Fast Fourier Transform) Spectrum of the registration point. I usually end up with a better stack if I decrease the radius of the FFT filter, so that it displays a larger target area than with the default setting (Figure 1).

RegiStax allows the option of dark-frame and flat-field calibration. If I used a color webcam to record my target, it’s important to convert the flat frame to monochrome before applying it, or else the results will be unusable.

At this point, I initiate the Align command. After a few minutes, RegiStax will display the results of the Initial optimizing run, showing two lines in a window: a red one representing image quality, and a blue one displaying the registration difference between each frame. I try to end up with two roughly horizontal lines, though this isn’t always possible (2). If the lines intersect closer to the left of the graph, I may adjust my earlier settings and repeat the alignment routine.

Once I’m satisfied with the initial alignment results, I move the slider at the bottom of the screen toward the left to exclude the poorest frames and select the Limit option, which brings me to the Optimize menu. Here I choose the Reference Frame menu, and change the Frames option from the default of 50 to between 200 and 300. Once I press the Create button, RegiStax combines the best frames within this limited selection to create a smoother reference image for use with the remaining frames. At this point I sharpen this image with the Wavelet filter before reintroducing it to the remainder of the process. I then save the image both with and without the Wavelet sharpening, because occasionally this small stack ends up superior to the final result made with additional frames, especially if the seeing was poor during the original video recording (3).

I sometimes skip the Optimize command and proceed directly to the Stack tab. Occasionally I’ve found that the post-optimization images have serious artifacts, especially if the seeing was less than favorable. Usually a few frames were grossly misaligned; these appear as a ghost image when the image’s contrast is stretched.

In the Stack menu, I open the Stackgraph tab at the bottom right and exclude any remaining poorly aligned frames by moving the Difference Cutoff slider down and the Quality Cutoff slider toward the left — RegiStax isn’t perfect, so some lower-quality frames usually sneak in. As I adjust these settings, the percentage of these frames decreases. I find that stacking 800 to 900 frames is optimal; a larger stack tends to obscure finer details, while noise begins to dominate if too few frames are combined. Finally, I initiate the Stack command.

After stacking is complete, I click the Wavelets tab. Before adjusting the Wavelet sliders, I first reset them so that no sharpening is applied, then save the image as a 16-bit TIFF file. If I decide to reprocess the image at a later date, I can bring this “raw” stacked file back into RegiStax without having to repeat the alignment and stacking routines. If I used a color camera to record the image, it most likely displays color fringing caused by atmospheric dispersion. Here I utilize the RGBshift function noted in Walker’s review. When a satellite of the subject planet is in the field, I process the entire AVI movie a second time, registering only on the moon, and save this image as a separate file to add later in Adobe Photoshop.

Now that I’m ready to sharpen the image, there are a few self-imposed limits I apply to my processing to ensure I don’t add artificial detail to the picture through oversharpening. I limit the aggressiveness of the Wavelet filter based on various factors such as image quality, number of frames in the stack, and the type of detail I hope to resolve.