ZWO autoexposure at night time

[Jonk2]

It's been mentioned a couple of times that there's no way of controlling the ZWO autoexposure when the software starts 'nighttime' imaging.

I know that 60 seconds is the limit (maybe it's worth putting this in the wiki / instructions if not already there?) but in all other imaging software that you can use for ZWO cameras (OACapture, Sharpcap, Firecapture etc for example) you can set a max exposure limit and target exposure value (based on the histogram peak, normally 0-255).

On my 178MC and 178MC-Cool (essentially the same camera), I have it set to 60s (max) exposure at night, autoexposure, gain at 160, offset at 300 and during the day I get great exposure (as expected) and at night I get great exposure for clear moonless nights.

During a bright moon, such as tonight, my exposures are up at around 45-50 + seconds, with the moon high in the sky - it might as well be daylight looking at the images. (Yes I know my sensor needs moving down!)

On the text overlay, it tells me the mean brightness - this is up at 200 odd tonight. I assume this is in the range 0-255?

My auto-exposure brightness for nighttime is set to 50 (the accompanying text says

50 - 160. Changes the amount of light in the image.
Low value = dark; high value = bright. Only works with auto-exposure.

What does this mean? 'Amount of light in the image' doesn't mean a lot to me and isn't likely to be technically accurate. Is this just moving the black point?

What does the range 50-160 mean?

Mine is set to 50 - my image 'mean brightness' is up at 200, which I assume is the peak, or average of the peaks on the histogram of 0-255.

Is there not a way to have the 'target exposure value' choice, with a number between 0-255 much in the same way as the packages mentioned above, (which I would personally set to slightly less than half the histogram, say ~120) in the gui? This would mean that the moonlit and very cloudy nights it would not wash out but during darker clear moonless nights, the max exposure of 60s would still work as the target exposure value probably won't be reached.

Or am I misunderstanding something?

[EricClaeys]

@Jonk2 About a year ago we developed an improved auto exposure for ZWO cameras during the day, because they were producing flickering when using ZWO's auto exposure method. At the time, people weren't having problems at night, or at least they weren't reporting problems. The new algorithm improved daytime pictures, but it's far from perfect. At night, we use the ZWO method which we're finding produces flickering at night. You are correct that there's no way to control the ZWO auto exposure method.

All the ZWO cameras I've seen have a 60 second exposure maximum when in ZWO auto exposure mode, for example, at night. We chose to keep that limit in our algorithm, although I don't remember why. We determine the mean brightness (0 - 255) of an area of the sensor we call the "histogram box" - the default is a 500px x 500px box centered on the sensor (see the advanced options in the WebUI). You are getting 200 on a moon-y night most likely because the moon is in the histogram box (you can display the box). I had to move my box because the Sun went through it so I was getting very dark images during parts of the day. At night we report the mean brightness but just FYI - we don't do anything with it.

I honestly don't know what the "Brightness" setting does - ZWO's documentation doesn't help and neither did some emails with their tech support. ZWO actually has two "Brightness" settings. One they also called "offset" (and the WebUI calls "Brightness (offset)" in version 0.8 and was simply "Brightness" in version 0.7) and add a some brightness to every pixel. From what I was able to determine, it adds about 1/10 the number you specify - if you say 20, it'll add 2 points (out of 255). The other "Brightness" is ZWO's "target auto exposure brightness". I don't know if it adjust the exposure, gain, or both, but it makes the picture lighter or darker, but only when using ZWO's auto exposure. Since we use our own during the day, I have the Brightness setting adjust the exposure by 12% for each 50 units. If you specify 150 it'll increase the brightness by 24 percent. These numbers were arbitrary and one could easily argue for other numbers. The 50-160 or 50-600 or whatever the range is what each camera supports. Using Debug Level of 4, look at the top of the log file - it will list all the camera's capabilities including minimum, maximum, and default values. One of our "to do" items is to have the WebUI know what camera model you have and adjust what it displays. For example, the 178MC doesn't support Gamma or cooling, so why show those options? It's upper Brightness (I believe) is 600 so the comment should reflect that. etc.

The "Mean" value displayed I believe (don't quote me) is the value that appears most often in the histogram box.

Sorry, but I don't understand what you're asking for. Can you give an example of exactly what would change?

FYI, for the Raspberry Pi HQ camera, @AndreasLMeg implemented a much better auto exposure algorithm than we have for the ZWO - his bases exposure changes partly on past exposures which helps minimize the flickering. And his works at night. Our plan is to port it to the ZWO code.

What I would like to do with most of the numerical settings is to have them range from 0 to 100, regardless of the camera. For example, the native Brightness of one ZWO camera is 50 - 160 and another 50 - 600. That would be hidden from the user and they would simply enter 0 - 100 and the software would map their number to the associated native number. That cleans up the interface a lot and greatly simplifies things for users with more than one camera, or who replace cameras. I'm not sure yet if this will apply to Andreas' auto exposure code - I don't think he's implemented the concept of Brightness yet.

[Jonk2]

Thanks for the lengthy reply.

I will wait to see what improvements are made in the coming weeks. There must be a way to instruct the zwo autoexposure to target a user setting.

On a PI, try Oacapture by James Fidell. You can see the autoexposure working in realtime as you cover / uncover the lens.

Uncovered:

Covered:

Setting the brightness to 0 or 600, the exposure time is the same (in my example around 70ms when it settles) so the brightness setting cannot be the target exposure.

If you do the same with the same ZWO camera on Windows using Sharpcap you can see the autoexposure target brightness range is 50-160, as per the Allsky GUI, not 0-600.

Covering the camera on this program, the exposure value ramps up and up on each exposure (into the s range not ms).

I'm wondering if the 'auto-exposure brightness' setting you're using is not the correct 'target exposure' that I refer to? It's a range dynamic range setting, and the target exposure might be limited / preset to the centre perhaps.

In fact, the brightness setting in sharpcap when at 0 shows a very wide histogram graph when autoexposure to a target value is used:

With the brightness at max (600) the histogram is very narrow, presumably widening the range:

This must be the offset (dynamic range).

I might be talking nosense, but I would ask why sharpcap can set a target exposure value whereas oacapture / allsky doesn't appear to be able to?

The only difference is the OS so perhaps ZWO doesn't allow it or cannot do it.

It may be that there's no way around it with Linux as their sdk is the limiting factor (which you probably already knew, but I'm trying to understand hence the waffling!)

Hopefully you guys with the brains can figure something out!

For now, I can wait and am happy to test with my 2 setups (RPI4 32 Bit + 178MC-Cool, outside) and RPI64 Bit + 178MC, inside test setup.