AkastroidImage Doctor
Measure every sub before you stack it
Before anything is stacked, every frame is measured for star size, star shape, signal to noise, sky gradient, cloud, trails and clipping, and every rejection has its reason beside it. You can point Akastroid at a folder and read the table without stacking anything.
Part of the free download. Nothing is uploaded and your files are never written to.
To try the measurements first, run them on one imagein your browser. Nothing to download.
The session planner shows target altitude, moon interference and a suggested order for tonight, also free.
The problem with blinking through 400 subs
The standard advice for culling frames is to look at them and delete the bad ones. Try it once. By frame 60 you cannot remember what the good ones looked like; by 150 you are rejecting frames because you are tired. And the faults that matter most, such as a four percent drop in sharpness or a background creeping up as the target sinks toward the horizon, cannot be seen at any size a screen can show you.
Culling is a measurement problem. So Akastroid measures, and shows you the measurements rather than acting on them silently. Software that quietly threw away a third of your night and told you nothing would be worse than software that threw away none of it.
What gets measured, and what each number lets you conclude
Every one of these is measured from the frame itself. Nothing is inferred from your equipment list, and nothing is a preset.
Star size (FWHM)
How sharp the frame is, in pixels across a star's profile. Rising through a session means focus drifting as the tube cools; high on every frame means the seeing, the focus or the sampling. It is measured from your own stars, so it is in your pixels rather than in somebody else's arcseconds.
Star shape (eccentricity and roundness)
Whether stars are round or stretched. Stretched means something moved (a tracking error, a gust, a cable snag) or the optics are at fault. Which of those it is has its own answer below, and it is the difference between buying a guide camera and collimating what you already own.
Signal to noise
How far the subject stands above the noise in this frame. It is what decides whether more frames or longer frames is the better use of your next clear night.
Sky brightness and gradient
How bright the background is and how steeply it varies across the frame. A bright but flat sky is recoverable; a steep gradient is much less so, and both are measured relative to the rest of your session rather than to a constant.
Cloud and haze
Called only when two things agree: the sky is brighter than the session's median and the star count has dropped. Either signal alone gives false positives: a rising target brightens the sky without cloud, and a passing focus wobble drops the count.
Satellite and aircraft trails
Counted, with the longest measured. A trail is not usually a reason to discard a frame, because outlier rejection removes it during integration. It is still worth knowing how many there are, and a frame with several deserves a second look.
Overexposure
What fraction of the frame is clipped. Clipped stars cannot be color-calibrated and a clipped core cannot be recovered by anything downstream, so this is one of the few faults processing cannot fix.
Star count
How many point sources were detected. It is the sanity check on everything else: a frame with almost none is a frame where cloud, condensation or a lens cap is the real story, whatever the other numbers say.
Your optics, or your tracking?
Getting this wrong costs money, and the data in your own session usually answers it.
Elongated stars have two very different causes. If the mount drifted, some frames are stretched and others are not, and the direction changes through the night as the guiding corrects and overshoots. If the optics are at fault (coma, field curvature, camera tilt, a pinched mirror, a sagging focuser), every frame of the night is stretched the same way, because the optics did not move.
So Akastroid looks at the elongation across the whole session rather than frame by frame. When it is present in all of them, it says the optics are the cause rather than the tracking, and rejects no frames for it. A rule that discards frames for a fault every frame shares leaves you with an empty session and no explanation.
The practical difference is whether your next purchase is a guide scope or an hour with a collimation tool. Optics or tracking covers the one-look test that separates them.
Why the thresholds come from your session
Nothing here is judged against a fixed number. A 3-pixel FWHM is superb at one focal length and soft at another. A background at 0.05 is a dark site from a city and a bright one from a desert. A star count of forty is a rich field through a fast lens and a failure through a narrowband filter.
Judging each frame against the rest of its own session asks whether this frame is worse than the others taken that night. Two things follow. A uniformly good session is not culled, because nothing in it is unusually bad. And a poor session still produces an image, because at most forty percent of a session can be rejected, however bad the numbers look.
The faults it names
A score on its own is not a diagnosis. Where a measurement crosses into a fault, the fault gets a name and appears beside the frame: soft focus, star trailing, strong gradient, cloud or haze, satellite or aircraft trails, overexposed, or too few stars to register on. No frame is rejected without one; a rejected row with an empty reason would be a bug.
Cloud is the one most often called wrongly. Akastroid raises it only when two independent signals agree: the sky is brighter than the session's median and the star count has fallen. A rising target brightens the sky on its own; a focus wobble drops the star count on its own. Requiring both is what stops half a good session being labeled cloudy.
Using it as a diagnosis and nothing else
You do not have to stack anything. Point Akastroid at last night's folder, read the table, and close it. That tells you whether the focus held, whether the guiding held, and whether tonight is worth another try. Doing this is free.
Nothing is deleted, moved or altered. Frames are marked for the stack or against it, and any row can be clicked to overrule the decision. Your files are read and never written to, and none of this leaves your computer.
Read further
Reading a quality score goes through the numbers one at a time. A symptom-by-symptom guide works the other way round, from what you can see to what caused it. Optics or tracking takes the most commonly misdiagnosed fault in detail, and deconvolution explained covers what sharpening can and cannot recover once the frames are stacked.
If you are still deciding what to process with, what Akastroid replaces, and what it honestly does not is the short version, and the 2026 buying guide is the whole field.
What is Image Doctor?
It is the measurement pass Akastroid runs on every frame before it stacks anything, shown to you rather than kept internal. Twelve things are measured per frame, including star size, star shape, signal to noise, sky brightness, sky gradient, noise, clipping, satellite trails and star count. Each frame gets a score and, where something is wrong, a named fault. You get a table of your whole session with the reason beside every rejection.
Is it a separate product or an extra cost?
Neither. It is part of the free download and it runs on every session, because the same measurements are what the automatic processing decides from. There is nothing to enable and nothing to buy.
How can it tell an optics problem from a tracking problem?
By whether the fault is in every frame. A tracking error stretches stars in some frames and not others, and the direction changes as the mount fights back. An optical fault (coma, field curvature, camera tilt, a pinched mirror) stretches them in the same way in every frame of the night, because the optics did not move. So when the elongation is present across the whole session Akastroid says so, and says it is the optics rather than the tracking, and does not reject a single frame for it. Rejecting frames for a fault that is in all of them leaves you with nothing.
Why are the thresholds relative to my session rather than fixed?
Because a fixed threshold is wrong for almost everybody. A 3-pixel FWHM is superb at one focal length and soft at another; a sky at 0.05 is dark from a city and bright from a desert. Judging a frame against the rest of its own session asks the useful question: is this frame worse than the others I took that night? It also means a uniformly excellent session is not culled and a poor one still produces an image.
Does it delete anything?
No. Nothing is deleted, moved or modified. Frames are marked selected or rejected for the stack, and you can click any row to overrule it. Your files are read and never written to.
Can I use it just to check my frames, without stacking?
Yes. Point it at a folder and read the table at the end of a session, to see whether the focus held, whether the guiding held, and whether it is worth going back out tomorrow.
Does it work on single frames?
The measurements do, and they are shown. The scoring does not mean much, because a score is a comparison and one frame has nothing to be compared to. Faults that are relationships between frames, like too few stars to register on, are not raised against a lone image.
What does it not tell me?
It measures the frame, not the sky above it or the equipment under it. It cannot tell you your polar alignment error in arcminutes, it cannot see a dew heater that is off until the dew arrives, and it will not diagnose a fault that leaves no trace in the pixels. Everything it reports is measured; nothing is inferred from your equipment list.
Check last night's frames
Point it at the folder and read the table. No account is needed and nothing is uploaded.