IMX662 vs IMX462

[robhorchner]

I have the IMX462 and think about upgrading to the ZWO ASI662MC but i wonder i it will make much difference on a bortle 6 sky?
Will i be able to make longer exposures with the 662 or am i already at the limit with my IMX462 at 115 seconds (without tracking)?

[jimmauck]

I have the ASI676MC, which I ran in Bortle 5 during testing, and now Bortle 2 at the remote observatory. FWIW, I found under x Bortle skies (2 or 5) that aiming for shortest exposure time and highest gain with acceptable noise level gave the best results. In my limited experience, shorter exposures resulted in sharper images (less star trailing) and smoother timelapse sequences. Then match dark frames to improve from there.

I'm not sure if that helps answer your question. It would be great to learn what others have found.

[robhorchner]

Maybe hard to compare different sensors but interesting and useful observation.

When i set the exposure to auto it will go to max (115 sec) when the sky is clear.
I don't see much difference between short and long exposures tbh.
I wonder how the auto exposures are calculated.
It seems like when the JSQM goes down the exposure time goes up.
Maybe Aaron can answer that?

[aaronwmorris]

Auto-exposure time is simply based on image brightness. jSQM stands for "janky SQM". It is really just the average ADU value from the camera. As the sky gets darker (decrease in jSQM), the exposure times will ramp up.

[aaronwmorris]

The only issue I have seen with the CSI/MIPI Starvis sensors (IMX327, IMX290, IMX462) is I do not believe there is enough EMI shielding compared to their sensitivity. Higher gains will introduce interference, usually horizontal banding. Astronomy/planetary cameras usually fare much better in terms of interference.

The IMX662 is a Starvis 2 sensor, vs IMX462 is Starvis (version 1). I do not know how much of a difference that will translate for an allsky camera.

I would rather have just about any astronomy/planetary camera rather than a CSI module.

[jimmauck]

Per my limited experience, I would start with dialing in maximum acceptable Night Gain (moonless), and then set Max Exposure time to get desired sky brightness. Next, set Moon Mode Gain to be acceptable sky brightness within the already established Max Exposure parameter.

Aaron, would it be possible to have separate "Moon Mode Max Exposure" time to segregate moon and moonless settings?

[robhorchner]

Thanks guys, appreciate it!

[aaronwmorris]

@jimmauck Regarding the Moon Mode max exposure. Having a separate max exposure would not really be very useful. Maybe if it were paired with a different "moon mode period" setting.

However, I am currently working on an Auto-Gain setting that should dovetail nicely in this use case. My current theory on Auto-gain is to allow the gain to scale up and down to maximize the exposure time. Maximizing the exposure ensures less lost time between frames.

[jimmauck]

Auto-gain is intriguing. Is there a way to optimize, or scale, darks to scale with a range of gain values (Auto-Gain), considering temp and exposure?

[aaronwmorris]

No, but indi-allsky already supports using the dark frame from the next highest gain in the database, ie if you have darks for gain 1, 75, and 100, an image using gain 50 will use the dark @ 75. It may over-subtract the data, but if you enable the Fix Calibration Pin Holes option, the pixels will just be replaced instead of just subtracted. This should result in a very clean image.

[robhorchner]

I have set the default exposure to auto and gain to 10, although a little bright it gives me a decent no of stars and a unknown object (yellow circle)

[robhorchner]

Got it. I found this article comparing gain and exposure times for dark scenes: https://afterhourscoding.wordpress.com/2024/11/30/exploring-imx462-sensor-settings-in-dark-scenes/

Perhaps try Night Gain = 50 and Max Exposure = 15? Are you able to utilize darks to subtract the noise?

Per the article, I wonder if the OS is patched to provide HCG control yet.

Thanks, interesting article!
I have not activated the HCG control yet.
What exactly are the benefits of HCG?
Are the darks automatically applied to the captures?

[aaronwmorris]

HCG gives greater dynamic range and lower read noise at the same gain level. You can see the diagrams on the ZWO site for the IMX462. https://www.zwoastro.com/product/asi462mm/

Note: ZWO cameras have a fixed gain level where it switches between HCG and LCG, but the behind the scenes the IMX462 (and other sensors) the switch between HCG/LCG is arbitrary.

Yes, if a dark is found for a particular gain, exposure, etc level, it is automatically applied (unless disabled in the config).

[jimmauck]

Technically, the highest analog gain supported by the Arducam IMX462 is 29.5. Anything higher than this just adds digital gain.

Would it make sense to use Night Gain = 29 (rounded down from 29.5) with Max Exposure = 15 as trial point?

[robhorchner]

I have tried Night Gain at 29 but it's too bright, 10 is still a little bright but acceptable (see image below).
I've set the exposure to auto to let indi-allsky find the optimal exposure.

[aaronwmorris]

Absolutely. And FYI, the latest versions of indi-allsky now manage gain as a floating point value and you can go all the way to 29.5. :-)

[robhorchner]

[robhorchner]

Still torn if i should keep the Arducam IMX462 or get a ZWO ASI662MC.
I know the 662 is much better but will i be able to see the milky way and maybe some vague dso's (at bortle 6)?
In order to do that i have to do longer exposures, how long can you go without tracking?
I also wonder if trying to see more is maybe beyond the purpose of Indy-Allsky?

[jimmauck]

I am able to "see" the Milky Way with my ZWOs under Bortle 5, streetlights, and neighbors' light trespass. There was little MW contrast against the bright sky, and I did not detect any DSO color. However, I did not tune the image settings under B5 skies as I focused on a robust all sky 3D printed enclosure design intended for remote, harsh winters. (I shall see how it fairs this winter.)
It seems some all-sky setups are using more advanced techniques to post-process images that may help under light polluted skies, many of which are revealed in this forum, such as transferring image files to a powerful computer (e.g. running Docker), then processing on the fly, and presenting the processed results from there.
It depends on how much resources we want to invest in this hobby.

[aaronwmorris]

More exposure is not the answer. In fact, I would suggest using a shorter exposure with higher gain.

I think your real issue is light pollution and local light sources. I see a slight "doughnut" ring of light in the image, which is light reflecting inside your dome. This is killing contrast a bit, but the real issue is just the light dome in your area. I think your light pollution profile is worse than B6.

[robhorchner]

There are indeed a lot of light sources at my location.
Clearoutside tells me it is B6 at my city but my location is probably B7.

According to chatgpt the following settings should be applied; Auto exposure, Max Exp 6–8 s, Night Gain cap ~24–27, Bin 2

Can't wait to see how the upcoming Auto-Gain will handle this.

[aaronwmorris]

FYI, I do not think libcamera binning functions the same as regular astronomy cameras. I believe libcamera only supports averaging mode for binning, so there is no signal benefit. (There is technically a Signal-to-Noise [SNR] ratio benefit, but that is not incredibly useful for an allsky camera). Many astronomy cameras use a "sum" where 4 pixels are added together so there is 4 times the signal, but libcamera does not do this.

[robhorchner]

Turns out that Arducam's IMX462 PiVariety driver runs in HCG by default.

[aaronwmorris]

That is interesting...