Fixed manual exposure and autoGain at night - possible bug or feature

[MMysBelec]

After successfully solved bug in 16-bit mode autoexposure on ASI cameras (thanks, now works flawlessly) I´m playing with camera, trying to find optimal setting. Under my sky, camera is capable to do 30 sec. exposures at quite high gain 255, showing nice image of the sky and brighter meteors and bolides, even if close to the 100k city. See image in attachment. When the sky become brighter (moon, haze), the auto-exposure function now correctly shorten the exposures, but interval remains the same, as max. exposure is defined. This lead to blind intervals (if exposure is shortened to 10sec. 2/3 of time the camera is blind). I think it is normal feature, to have all images in night equally spaced (I know, I can disable that, but this is not good solution).

So I tried to use fixed exposure, and autoGain instead of AutoExposure. Hoping it will reduce gain instead of time on brighter sky. But not. The algorithm starts with specified low manual gain (for example 50), and continuously rise it to the max. gain, even if image becomes totally white. Is this normal, or is this similar bug as with auto-exposure ?

The problem is present in both 8 and 16bit mono modes. Camera ASI 178MM. Can you look on that, if you are able to reproduce that behavior.

[SolCygnus]

Great Shot! I like to work on the color side of the cameras, however in my search for optimal settings I have noticed the software does lean towards max gain prior to anything else then adjusts exposure once max gain is reached.

[MMysBelec]

In my case, the gain reached max. gain and stay here, never returned back to value given by the Mean target. I can check it again, may be I was so hurry.

[MMysBelec]

No, algorithm is going to the max. gain forever. Below is the detailed log, and image. I always end with white image with mean of 255 in ROI which is 900x900 px. (marked by red square). Is this probably bug ?

allsky.log

[SolCygnus]

I'd wait and see if one of the devs drops into the convo, if they do not in the next 24 may want to move the convo over to issues forum. I did not see any similar issues open or closed. Your mean value seems really high just taking a quick look.

[EricClaeys]

With ZWO cameras, autoexposure uses an Allsky algorithm (which uses the histogram box) but autogain uses the camera's algorithm which in our opinion isn't very good, as you have noticed. I tried to get ZWO support to explain their algorithms but never had any luck. However, I am pretty sure ZWO uses the mean brightness of the whole image, which is the FULLMEAN value in the log file which was 115 which is probably considered fine.
Since we don't control ZWO's algorithm there really isn't anything we can do about it.
You can try decreasing the nighttime Brightness.

A future release of Allsky will port the RPi algorithm to ZWO; it supports autoexposure and autogain.

[MMysBelec]

Yes, understand fully. If fullmean is 115, when image is totally white (and averaged with black parts of image), I must consider that number as maximum instead of 255 and lower the nighttime Brightness accordingly. I will try that later at clear night, and will wait for future release.

Thanks for explanation, I did not know so autogain is not under your control now yet.

[SolCygnus]

FWIW, lots of people get good results using the ASI178 mm and mc including myself. If you get tired of working the issue, I would try a reset to default values in the settings and do minor tweaks from there. Some settings I use on the 178mc, Auto-exp,gain,wb max set to 20sec expsoure, 250gain, 5sec delay, auto image type, filename .jpg, consistent delays. ~ Clear Skies

[MMysBelec]

My settings are similar, max. 30sec, gain 255.

[MMysBelec]

I don´t know how ZWO autogain algorithm works, but it ignores nighttime Brightness, even if set very small or zero. Simply, for any value behaves the same and the gain is screwed up to specified maxGain.

[EricClaeys]

It's possible the Brightness only impacts auto exposure.

[MMysBelec]

Who knows, I will wait for AllSky algorithm for ZWO too.