Docs · Troubleshooting

Troubleshooting & FAQ

Some of my calibration frames show as “Unclassified” in Build Masters

They won’t be built into a master until they’re identifiable. Tag them via the FITS IMAGETYP header, a filename prefix (Bias_ / Dark_ / Flat_), or by placing them in a subfolder whose name contains “bias”, “dark”, or “flat”.

“Masters folder is empty” dialog

The folder you picked contains no FITS files, so the stack would run uncalibrated. Choose Build masters… to create them from your raw cal frames, or pick a different folder. The same dialog appears for an empty Flat masters folder in calibration-library mode — there it offers to build just this project’s flats.

“No master flat for <filter>” on the Run screen

Since 1.8.4 a run says when some lights have no matching master flat, rather than skipping the flat silently: “No master flat for Ha (…) — those lights are calibrated without a flat”. Either that filter has no master flat yet (build one — see Building masters), or the flat doesn’t match on camera, binning, or gain. In calibration-library mode remember that flats come only from the project’s flat-masters folder: flats sitting in the library are deliberately ignored, so each project’s own flats are used.

My stack ran but looks uncalibrated

Check the Run screen for “No master frames found — calibration will be skipped.” Your Masters folder was empty or nothing matched your lights (matching is by camera + binning + gain, plus filter for flats and ±10% exposure for darks).

“I/O error: Too many open files (os error 24)”

Versions before 1.6.0 could stop with this while building masters from — or stacking — a large set of frames, typically a few hundred bias or darks. It was most common on macOS, where an application is allowed only 256 open files by default no matter what your shell’s ulimit says. Rapid Stacker now raises that limit for itself at startup, and where a group is still too large to hold open all at once it re-opens each frame as it reads instead. The masters are identical either way; the fallback costs about 6% on the integration pass. Update to 1.6.0 or later if you hit this.

Does it use my GPU?

Backends are probed automatically: CUDA (NVIDIA, Turing+ on the Windows CUDA 13 build) → wgpu (any modern GPU via Metal/Vulkan/DX12) → CPU. Older NVIDIA Pascal cards (GTX 10-series) use wgpu on Windows. If no GPU is usable, the CPU path produces identical results, just slower.

Plate solving didn’t run / no WCS in my master

The solver needs RA/Dec and pixel-scale hints in the frame headers; without them it’s skipped silently. Also check that you either have an internet connection or a Local Gaia DB (XPSD) folder configured for offline solving. A failed solve never blocks the master from being written.

One filter produced several stacks instead of one

Each distinct exposure time forms its own group by default (WBPP parity). Turn on Settings → Grouping → Combine exposures to merge them, or raise the Exposure tolerance if capture jitter split a single set.

Where do my settings live?

In the OS config directory (e.g. %APPDATA%\com.ekeservices.rapid-stacker\ on Windows), saved about a second after you stop making a change — so settings survive even if the app is closed abruptly. Deleting it resets the app to defaults.

A red “error:” bar appeared at the bottom

That’s the error toast — the message describes what failed (e.g. an unreadable file). Click dismiss; the app keeps running.

A run failed and I want to report it

The failure message stays on the Run screen — it isn’t cleared when you navigate away — and the text is selectable, with a Copy error button next to it. Paste that into your support email or feature request. Dismiss clears it once you’re done.

The Frames or Project screen says “scanning…”

Folder discovery reads the header of every FITS file in the tree, which takes a few seconds on a multi-thousand-sub project. It runs on a background thread, so the window stays responsive and shows a spinner while it works. If the scan fails outright (an unreadable folder, a disconnected drive), you get an error card with a Retry scan button rather than a repeating toast.

Current limitations

• Auto-cropped multi-session masters carry standard FITS WCS only. The cross-group crop currently drops the PixInsight-native astrometric properties, so SPCC on those masters needs a re-solve inside PixInsight (the FITS WCS itself remains valid).

Still stuck?

Email support@ekeservices.com, or submit a feature request if it’s a missing capability rather than a bug.