SaratogaSkies Jim Solomon's Astropics


Latest news

December: Test shots with new scopes/mounts

Dec 21: TMB 80/480 Arrives!

Dec 3: AP1200 Arrives!

Nov 30: TMB 152/1200 Arrives!


Printer-friendly Version

Vignetting of Hutech Hα Front Filter?


I recently picked up on Astromart a Hutech, Hα, Front Filter (from herein abbreviated Hα-FF). Before going out under the stars for first light, I decided to play with it “on the bench” to figure out, among other things, what the right Tv (exposure time) and ISO setting should be to properly expose Flat Lights. Properly expose, that is, the red sensors (only) of my Modified 350D, since the green and blue sensors will “see” almost nothing from the Hα-filtered Light Box.

After a couple of experiments, I settled on 10sec @ ISO 800 which gave me a median value of 1700 or so, a bit shy of the target of 2048, the latter of which is the center of the linear range of the 12-bit A/D converter in the 350D. Those 12-bit unsigned values, by definition, fall in the range [0..4095]. I should probably crank the exposure time up a bit, but this is close enough for the purposes of this article.

Test Equipment

All of the analysis was performed and all of the images were collected with:

In all of the example images that follow, I'm showing only the red pixels from the 350D's Color Filter Array, and, for maximum clarity, I'm showing those pixels as a grayscale image. IRIS's convenient split_cfa command was used to extract just the red pixels from a raw, linear, uninterpolated, CFA image.

A Problem Arises

Upon inspecting the exposures, it appeared to me as if there was a radially symmetric “step function” present in the Flat Lights, which looked sort of like a bright circle centered in the image. Figure 1 below shows a single Flat Light exposure that has been calibrated against a Flat Master Dark (median combine of five Flat Darks at the same Tv and ISO as the Flat Lights). For historical reasons, the Flat Light has been normalized to a median value of +20,000 in IRIS' signed 16-bit integer space [-32768..32767]. Remember, this is a ”linear“ image, with no gamma, DDP, nor other stretching applied. It is shown here visualized in IRIS from [15000,25000] and shrunk by 25% for display in this article:

Figure 1: single Flat Light with Hα-FF

Figure 1 shows the Flat Light being generally brighter in the center and darker at the edges/corners, just as one would expect. But the “sharp” transition between bright and dark is not expected, and that's what I'm referring to as the radially symmetric step function.

Further Analysis

Not yet discouraged, I continued to process the Flat Lights by sigma-combining a handful of them (five), all of which had been normalized after calibrating against the Flat Master Dark. As expected, the general form of the image remained the same, though the sigma-combining of five exposures helped to reduce the random noise in the image. The lower random noise makes the ”pattern noise“ (checkerboard pattern, more evident vertically than horizontally) all the more evident, however. The result is in Figure 2. For comparison purposes, Figure 3 shows a sigma-combined Master Flat from the same optical system (200" f/5 Newtonian with MPCC) and same camera, but without the Hα-FF. Again, only the red pixels are show for comparison sake:

Figure 2: Sigma-combined (Master) Flat with Hα-FFFigure 3: Sigma-combined (Master) Flat, no filter

Figure 3 is exactly how I'd expect a Master Flat to look with an optical system that has a noticeable amount of vignetting. Note how smoothly the brightness diminishes as you look outward from the center, with no abrupt “step” change in brightness at any given radius. (The tiny amount of clipping in the center is a function of the upper visualization threshold chosen. The image itself is not clipped.)

One final way of looking at the problem is to take the Master Flat with the Hα-FF, treat it as if it was a Light, and divide it by the Master Flat without the Hα-FF. That tends to factor out the vignetting inherent in the optical system, and focus the attention on the difference between the expsoures with and without the Hα-FF. That result is shown in Figure 4. But since that's a little hard to see, Figure 5 is the same image as Figure 4, but visualized more aggressively at [18000,22000] to increase the contrast:

Figure 4: “Flattened” Master Flat with Hα-FF (original)
(Equals Figure 2 ÷ Figure 3)
Figure 5: “Flattened” Master Flat with Hα-FF (contrast enhanced)
(Equals Figure 2 ÷ Figure 3)


It's pretty clear from these images that there's a problem somewhere in the optical path. I'm not sure whether the Hα-FF is simply causing some nasty vignetting, or whether there are reflections being caused by the UV/IR replacement filter in the 350D, the coma corrector, the Hα-FF, or all of the above. Of course, it's also possible that at the Hα wavelength, the diffusion layers of the Light Box are such that it really looks like the image shown in Figures 4 and 5.