To research the topic of imaging at the wavelength of Helium, I first watched the YouTube video by Christian Buil: https://youtu.be/TuF6R9K5cqg. Some other good tips are given by Peter Zetner in several old SolarChat posts: viewtopic.php?t=14044, viewtopic.php?t=19411, viewtopic.php?t=25437. On my second attempt, I was able to see something.
The first task: finding the Helium line. It is relatively close to the Sodium doublet, a little bid towards the blue. Near the limb, Helium can be seen as an emission line (rather than a typical dark absorption line). Here is a screenshot taken from a scan with our SHG, as the slit approaches an E/W limb (edge):
We needed to learn how our software would treat the various lines and how best to position them. Basically, we need one strong Sodium line to be entirely within the ROI of the scan (and of course, the Helium line). This leads to a rather larger width of ROI than usual (around 300 pixels vs 150 pixels), so the file size is about twice (or more) than normal.
This graphic is the normal output from our software and shows the D2 sodium line being successfully fitted by the software and the D1 sodium line is partially visible at the right edge. This graphic actually corresponds to the scan image above it, so we know the Helium line is around 1/6 from the left edge, or around the 50 pixel mark. The detected Sodium line is around the 230 pixel mark, so we can guess the Helium line should be around 180 pixels to the blue from the fitted line. (Note that in the graphic the scales are not the same so the line appears extremely curved.)
The fitted curve (a cubic polynomial) is what generates the centre line images in normal imaging (like H-alpha, H-beta, Ca-K/H, etc). From the centre of the line (assumed to be the darkest points), we normally do small pixel shifts (like -10 to +10), to see how the image changes with small wavelength displacements. In this case, we want a very large offset to the blue, so we input a pixel shift range of -190 to -170.
The software then generates a series of pixel shifted images. It turns out that looking at the "protus" images (the black disk image that highlights prominences), the exact location of the Helium line can be fairly easily recognised. In this case, it turned out to be at -179 pixels.
The second task: how to display the Helium absorption? We took the image at -179 pixels and another one close by (we chose -170). What is normally done is to subtract (or divide) the "helium image" from the "continuum image" to visualize the effect of absorption of Helium at the Sun's surface. When I played around with this way of visualising, I was not very satisfied. The reason is that the effect is so small that the differential image must be amplified considerably afterwards and any small amount of noise creates a lot of artefacts. I found it was more revealing (and potentially less misleading) to create a gif that would switch between the two images:
This way of displaying the data highlights how weak an effect Helium absorption is -- it is barely visible on top of the features of a normal continuum image, but the location of the helium absorption clearly corresponds to active solar regions. See, for example, the H-alpha and Ca-K images from Arne Danielsen taken on the same day viewtopic.php?t=37135. Generally speaking, the darkening caused by Helium occurs near where both H-alpha and Ca-H/K show relative brightness.
What about Helium prominences? They are quite weak compared to H-alpha but about the same strength as Ca-H/K prominences. To see them better, we took another SHG scan that was very overexposed -- the exposure time was increased until the D2 sodium line was just barely visible (1.14ms vs 0.38ms for the continuum, both with gain of zero). The sodium line needs to be visible or the software will not work.
Fortunately, on July 16 there was a huge filamentary prominence lifting off, visible in many H-alpha images (including our own viewtopic.php?t=37131). The image above is this giant feature in the famous yellow light of Helium emission, and the image below shows the prominences in H-alpha (on the same day).