Software Bisque

Take a virtual tour of the Software Bisque campus Watch Steve Bisque discuss the Paramount mount with S&T's Dennis Di Cicco Read the Paramount MYT Review Read about the Paramount Software Suite

The Hot Pixel Elixer...
Richard S. Wright Jr.'s Blog

... to cure your ills.

Two of the X2 camera plug-ins I'm responsible for do not have shutters (Lodestar/Starlight Xpress, and the SSAG). For main imaging cameras without a shutter, this is not a big deal as nearly all imagers build a dark library, or we'll use a bad pixel tool as part of our post processing procedure (if you have one of those ultra quite Sony chips and don't bother with darks). Properly calibrated images normally don't have many hot pixels left, if any. Normally.

The way we approach guiding though is a bit different. Most people don't bother with darks when they guide (they should, see later), or if they do they have a guide camera with a shutter and they can always do the "autodark" routine, supplying them with a fresh dark whenever needed at the cost of only a few seconds of automated delay. Hands-on imagers will also twiddle the guider exposure to account for the seeing of the moment, or to grab a particularly dim star that won't require changing their chosen main imager frame up. Compared to "imaging", "guiding" can be somewhat chaotic.

The "default" approach of just plugging in an autoguider with no shutter and pressing "go" has some issues however, and a search of our forums will reveal a number of users tripping over the hot pixel issue when guiding. First it's with trying to get a good calibration:

  1. Mount did not move "enough".
  2. Select a different guide star.
  3. Try again, maybe longer time.
  4. Go to #1.

Programmers call this an infinite loop. Often someone will cry out in desperation, "But I CAN SEE the star moving!"

Alas, the star has a peak ADU of 5,000, and there is a nearby hot pixel with an ADU of 42,000. That hot pixel is not moving and it looks like a really bright star just sitting still. Image link would fix this 100%, but most users don't have a guiding system that will produce enough stars for a successful image link. The second problem is a cloud comes through and the guide star dims slightly. A warm pixel in the corner of the frame leads the guider off on a wild goose chase chasing a pixel it will never catch.

"Can't TheSkyX tell a pixel from a star?" 

Well, TheSky took an image and the camera told TheSky that there is something bright at that pixel. We both know sometimes the camera is lying though. Proper calibration, again, is the best way to distill the truth.

I'm going to talk about the best way to fix this in a moment. First though, in order to provide a better "out of box" experience for the beginning imager, the next update to the Starlight Xpress and SSAG X2 plug-ins will have a routine that detects and removes hot pixels. This feature will be turned on by default (for guide cameras only on Starlight Xpress), so new users just trying off the bat should immediately get better results when they try and calibrate or select a guide star near a hot or warm pixel. The algorithm is not perfect, and it might eat the edges of bright stars, but should still improve results for most people without them having to do anything in addition to what they are already doing. This feature can of course be turned off, and there is a threshold adjustment to increase or decrease it's aggressiveness. The idea is any pixel that is greater than a certain threshold above all of it's neighboring pixels is probably a hot pixel. If it's not, then this probably makes the star unsuitable for guiding anyway. Bad pixels are replaced by the average of surrounding pixels.

Again, this is not perfect, but it will catch A LOT of bad cases for a lot of people. If your reading this blog however, this feature is not intended for you. It's intended for people who do not read blogs or documentation, and expect things to "just work". We might even improve this algorithm in the future, and it could be the norm for guiding one day.

There is for now however a better way, and it's been part of the Camera Add On for some time now already, and that's using the calibration library. For even better results, turn off hot pixel detection, and take a dark frame with your guide camera (possibly having to cover your scope if the camera does not have a shutter). This dark should be the same length as your anticipated longest guider exposure. Now, on the autoguider tab, select "Full Calibration", and an "Image Calibration Library" button will appear.

Now, when the image calibration library dialog appears, select the "Dark Frames" portion under "Autoguider" and then click the "Add Frames" button.

You can add a single frame here, or a group of dark frames all of the same length and choose to have them averaged or median combined when applied. Further, you can also elect to have the dark frame scaled to match the exposure time. My preference is to not scale dark frames, but your results may vary based on your particular camera.

Now, whenever you take an image with the guider, it will automatically be dark subtracted. Even when you sub frame on a camera without a shutter there will be no dreaded "please cover the scope" that you get with the autodark feature (which wasn't intended for cameras with no shutter).

Not scaling the dark is actually an option, but as Steve Bisque said about this to me recently, this is a bit like pounding a nail with a wrench, but it does work. Yes, the dark time does not match the light frame, but it's actually "good enough" for short exposures that are being used for guiding purposes only. I call this an, Incredi-hack". Simple, easy, and it works. Or you can scale... but wait, there's more!

Now, for the discerning imager, the honors students among you who have found this blog and don't mind reading the instructions and taking some extra steps in order to get superior results, we take this one step further. For us, we'll take several darks, one or more for each exposure length we may use with our guider. We can probably get away with just one dark for each exposure length, but you may of course stack a few. Create a new imaging group for each exposure for your guider, say for example "Guider - One Second", "Guider - Two Second", etc.

Now, when you actually are guiding, select the matching group for your exposure time. This is about the best you can possibly do when it comes to eliminating hot pixels. The big advantage to these last two methods is that it corrects more than just the errant stray hot pixel. I have a guide camera for example that has "clumps" of hot pixels that the first algorithm will not work on. Also, what about when a guide star is on or too near a column defect? For the discerning imager... this is as good as it gets, and the path to guiding nirvana. 

Happy Imaging,



Posted 04-08-2014 12:22 PM by Richard Wright


© 2018 Software Bisque, Inc. 862 Brickyard Circle, Golden, CO 80403-8058 USA - phone: +1 303 278 4478 fax: +1 303 278 0045