dwarfastro.com

At 4:22 in the morning my phone showed me something I couldn’t explain in my M13 Hercules Cluster image.

561 frames of M13, the Hercules cluster, nearly at zenith, 9.74 GB of data collected while I slept. The Great Hercules Cluster is half a million stars packed into a sphere 145 light-years across, sitting 25,000 light-years away. One of the showpieces of the summer sky, high enough that night to clear every tree and horizon issue I’d ever dealt with. Perfect polar alignment. Settings I’d used a dozen times. Everything about the setup said this should work.

Instead, across the middle of the stacked image, thick dark horizontal bands. Precise and almost geometric, running nearly the full width of the frame.

My first thought was the scope. My second thought was stranger.

Those lines were too clean. Too deliberate. I’ve been imaging long enough to know what sensor artifacts look like, and this wasn’t that. At first I honestly wondered whether I had captured a secret satellite, a strange aircraft, or the world’s least subtle UFO. Think about it: if something classified was transiting the sky over Westford at 3am and didn’t want to show up in civilian astrophotography data, wouldn’t it make sense to subtract itself out? To leave a dark trace instead of a bright one? Satellite trails are bright. They add light. These were the opposite, dark and removed, like something had passed through the frame and taken light with it on the way out.

I’m not saying it was a UFO. I’m also not saying it wasn’t.

I sent it to DwarfLab support that morning. They didn’t tell me to ship anything. They asked for the session data, which was a better sign.

So that night I pointed the Dwarf 3 at M13 again. Same target, same settings. And there it was again, same bands, slightly different position. Which meant it wasn’t random. It wasn’t a one-time event. It came back because whatever caused it was still in the pipeline.

Sue at DwarfLab found it. One dark calibration frame captured at 24°C had light in it. My best guess, and Sue’s too, is that a car drove past during the dark capture sequence and headlights got into the frame for a fraction of a second. That’s all it takes.

Here’s what that actually means. Dark frames are exposures captured with the lens covered, no target, no sky, just the sensor running in the dark. The point is to record the noise the sensor generates on its own: thermal heat, hot pixels, fixed electronic patterns. That dark signal gets subtracted from your light frames during calibration, cleaning up the image. When a dark frame is clean the process works. When a dark frame has stray light in it, you subtract that light from every single stack. Instead of removing noise you’re carving a dark shape into your data. One contaminated frame, applied to 561 stacks, printed the same banding across M13 every time.

Pull the file. Run Megastack again without it.

The result is the image above. 9 hours and 37 minutes of M13 across two nights, clean background, the cluster resolved from edge to core, background galaxies visible in the field. The data was there the whole time. The telescope was working fine.

What I should have done before any of this was open the dark library in FITS Studio and look at the individual frames. A bad dark is not invisible. Stretch it hard and it has structure where a clean dark would just look like sensor noise. Gradients, banding, a bright patch in one corner. That check takes under a minute and it’s now part of my routine before every session.

So if you see something like this in your stacked result, start with your darks. Open them in a FITS viewer, stretch hard, and look for anything that shouldn’t be there. And when you capture new darks, make sure the environment is clean. Lens covered, away from traffic, no phone screens nearby. A dark frame captured with a car driving past is not a dark frame. It’s a problem waiting to show up six hours later in your stack.

---

Frequently Asked Questions