Description
Amp glow (short for amplifier glow) is a luminous halo that appears in a specific area of the sensor, usually a corner or edge, caused by parasitic infrared emission from the readout electronics (amplifiers, output registers) located immediately adjacent to the photosensitive array.
This emission is invisible to the naked eye but is detected by the sensor itself, which is sensitive into the near-IR, creating a zone of additional signal that adds on top of the useful image.
Amp glow was massive and highly characteristic on early astrophotography CMOS sensors (ASI1600, ASI071, first-generation ASI294) and remains present to varying degrees on most modern sensors.
Its signature is perfectly reproducible: same position, same intensity for a given gain, offset, exposure duration, and temperature, which makes it entirely removable by rigorous dark calibration.
When amp glow appears on the final master, it is invariably a sign of a failed calibration, not a sensor defect.
Visual signature
An extended bright zone in a corner or along one edge of the sensor, forming a progressive gradient from the corner toward the center. Intensity falls off exponentially as distance from the thermal source increases.
On older sensors (original ASI1600, ASI071), amp glow can reach 50-200 ADU at the hot corner in a 180 s exposure, several times the read noise.
On recent sensors (IMX571, IMX455, IMX585, IMX533), it is residual: just a few ADU, barely visible, sometimes in several distinct zones (multiple corners on recent Sony sensors).
The signature is fixed in the sensor corner (always in the same location, perfectly reproducible frame after frame under identical conditions) and scales with exposure duration (the longer the sub, the more the glow spreads and intensifies).
On an uncalibrated master, the result is a very bright zone that stands out immediately after stretching, one that no atmospheric gradient could explain.
Differential diagnosis
To be distinguished from a light-pollution gradient (progressive variation across the entire image caused by external lighting, not confined to one corner; changes depending on telescope orientation and sky position).
Not to be confused with incorrect flat vignetting (peripheral darkening, not illumination; symmetric around the optical center, not in a specific corner).
Different from an internal reflection in the optical train (halo usually linked to a bright star outside the field, position varies with target).
Not to be mixed up with a twilight atmospheric gradient (extends continuously from horizon to zenith, not confined).
The primary diagnostic test: if the bright zone is always in the same spot on the sensor regardless of target or orientation, it is amp glow. If the position varies with pointing, it is an optical or atmospheric artifact.
Also check on a single long-exposure dark: if the bright zone appears there identically, the diagnosis is confirmed.
Probable causes
- Parasitic IR emission from the sensor output amplifier, a physical phenomenon inherent to CMOS
- Failed dark calibration: inadequate master dark or no dark applied
- Darks taken at a different temperature from the lights (amp glow is highly temperature-dependent)
- Darks with a different exposure duration than the lights (amp glow does not scale linearly)
- Darks taken at a different gain than the lights (amp glow varies with gain)
- Outdated dark library relative to sensor aging
- On OSC, debayering applied before calibration (amp glow is then distorted and harder to subtract)
- Sensor known for significant glow (older ASI1600, some ASI071 units)
- Faulty thermal regulation of the sensor (temperature fluctuating during the exposure)
- On recent CMOS sensors with HCG (High Conversion Gain) mode, failing to match HCG/LCG mode between darks and lights
Course of action
- Build a rigorously matched dark library: same temperature, same gain, same offset, same duration as the lights
- Prefer active sensor cooling (-10 C or -20 C depending on the spec), with 5 minutes of stabilization before the first sub
- Redo darks every 6-12 months to track sensor aging
- Verify that the master dark actually contains the visible amp glow before integrating it into the pipeline
- In PixInsight: use ImageCalibration with the master dark, "Calibrate" checked, "Optimize" unchecked (for CMOS sensors without a separate bias)
- In Siril: preprocess with the master dark, check the "applied" status in the FITS headers
- For OSC, always calibrate before debayering (order: calibration, cosmetic correction, debayering, alignment)
- On a recent sensor with marginal residual glow, DBE/GraXpert can clean up remnants without risking nebulosity targeting
- Verify readout mode consistency (HCG/LCG, gain, offset) between darks and lights in the FITS headers
- As a last resort on an already-affected master, model a synthetic background with DBE using targeted samples (at the cost of local signal loss)
The Doc's advice
Amp glow is the defect most misunderstood by beginners: people think it is a sensor flaw, when it is just normal physics that a good calibration eliminates 99% of the time. If you have a halo coming up in the corner of your master, your ASI1600 is not dead, your dark library is just wrong. Exact temperature, exact duration, exact gain: all three must match, without approximation. And on modern sensors like the IMX571 or IMX455, amp glow is so weak that if you still see it after calibration, you simply have not applied your darks.
Think you can see this defect in your image?
Run a diagnosisFrequently asked questions
Why don't my darks completely remove the amp glow?
Because the master dark does not exactly match the subs. Amp glow depends strongly on temperature, duration, and gain: if any of these three parameters differ between darks and lights, the subtraction misses and leaves a trace in the corner. Check that your darks are taken at the exact setpoint (for example -10 C), at the same duration and the same gain/offset, then confirm in the FITS headers that the master dark was actually applied.
Are modern sensors completely free of amp glow?
Not entirely, but the difference is enormous. Older sensors (ASI1600, ASI071, ASI294 v1) exhibited massive glow. Recent sensors (IMX571, IMX455, IMX585, IMX533) leave only a residual of a few ADU, often invisible after calibration. On these sensors, glow still visible in the master almost always indicates a mismatched or unapplied dark, not a sensor defect.
Do I need separate bias frames to calibrate amp glow?
No, amp glow is removed with darks, not with bias frames. Because it depends on duration and temperature, only a master dark taken under the exact conditions of the lights contains and correctly subtracts it. On most modern CMOS sensors, calibration is done with darks and flats (plus dark-flats) without a separate master bias, and the "Optimize" option is left unchecked in ImageCalibration.
My final master shows a slight residual glow after calibration. Can it be removed in post-processing?
Yes, a marginal residual can be cleaned up with gradient removal. If the glow is faint and confined to a corner, DBE or GraXpert will model it as a low-frequency background and remove it, provided no sample points are placed on nebulosity. But this is a workaround: obvious glow signals a calibration that needs redoing, and forcing it through DBE digs into local signal.