Why does my stacked image look worse than a single frame?

Stacking means taking many raw frames ("subs") of the same target and combining them into one image, so the real signal builds up while random noise cancels out. For that to work, the software first has to line every frame up pixel-for-pixel — that step is called registration or alignment. A worse-than-one-frame result almost always traces back to that alignment step, or to something feeding bad information into it, not to stacking as a technique.

There are four real causes, and they leave different fingerprints. Look at your stack closely and match what you see below — that's the fastest way to know which one you're dealing with, and it's honest to say upfront that more than one can be true at once.

Doubled or ghosted stars — alignment matched the wrong position

If stars show as two faint copies slightly offset from each other, or a star has a soft twin next to it, the registration step matched some frames to the wrong reference point before combining. This is a software matching failure, not bad raw data. Re-running the stack with a different reference frame (the one all others are aligned to), or switching alignment method if your software offers more than one, usually clears it up completely.

Soft, smeared, or lower-contrast than one frame — too few subs for the rejection method

Rejection algorithms (the part of stacking that throws out satellite trails, hot pixels, and stray noise spikes by comparing each pixel across every frame) need a reasonable sample size to tell an outlier from real signal. Below roughly 10-15 frames, some of these methods misfire — either letting junk through or discarding data that was actually fine — and the result can look softer or patchier than a single sharp sub. If your stack used fewer than about 10 frames, this is the first thing to suspect; try a simpler averaging method instead of an aggressive rejection method and compare.

New blotches or a color cast that wasn't in any single frame — calibration mismatch

Calibration frames (darks, flats, and bias frames shot separately to cancel out sensor noise, dust shadows, and vignetting) only cancel out correctly if they match the light frames — same temperature, same gain, same optical setup, dust in the same place. Reused calibration frames from a different night, or flats taken after the camera or telescope was touched, can inject a blotch, a vignette that wasn't there before, or a color tint into the stack that no single raw frame shows. If the artifact looks like a smooth patch or a tint rather than a doubled star, this is the likely cause.

Wrong file format or bit depth on import

Less common, but worth ruling out: if the subs were saved or converted to a lower bit depth (8-bit instead of 16-bit) before stacking, or the stacking software misread the color pattern of the sensor, banding or posterization can appear only after the combine step, since the precision loss compounds across every frame. Check that your raw files went into the stacker in their original format, not a JPEG export.

• Doubled or twin stars → alignment matched the wrong frame. Re-stack with a different reference.
• Soft, smeared, lower contrast than one sub → too few frames for the rejection method. Try averaging instead.
• New blotches or a color tint → calibration frames don't match this session. Reshoot flats and darks together with the lights.
• Banding or blocky color → check the subs were fed in at full bit depth, not a compressed export.

If none of these match what you're seeing, that's a real possibility too — some combinations of causes don't leave a clean, single signature, and the honest answer is that it needs a look at the actual file rather than a guess from a description.