Way to disable automatic dark subtraction/calibration on ASI178MC?

jpaana

Hi,

I'm using current ASI SDK 0930 with INDI driver with my new 178MC on Rpi3B+ and noticed that my darks/lights/flats have very low ADU values where as bias frames look more like I would expect. Bias frames have median value of around 150 and over depending on configured gain and offset values where as all other frames have mean of only around 4-5 regardless of gain/offset (RAW16 images). Is there a way to disable this automatic "calibration" and just get raw unaltered frames from the camera? Calling ASIDisableDarkSubtract() doesn't seem to have any effect at least

chad

Hi Sir，

I am not sure what caused the problem of low values on your light image.

But I think it should does not matter with ASIDisableDarkSubtract. We did not have an automatic "calibration" inside our camera. The data you get from our SDK is the raw unaltered data.

Thanks

Chad

jpaana

Hi, thanks for the reply, but there definitely is something going on that is exposure time dependent and seems to trigger somewhere between 0.02 and 0.025s exposure and even more so after 0.1s and after 0.2s the median value doesn't seem to grow at all regardless of 500x longer exposures. I took a series of light frames with cap on and only exposure time changed (it's unregulated camera, but sensor temperature was roughly the same).

According to statistics in PixInsight the median pixel values of each are (as 16-bit data):

32us: 124.0

0.01s: 158.0

0.02s: 158.0

0.025s: 123.0

0.03s: 123.0

0.04s: 124.0

0.05s: 124.0

0.1s: 52.0

0.2s: 7.0

0.25s: 5.0

0.3s: 5.0

0.5s: 5.0

1s: 5.0

10s: 5.0

100s: 5.0

Gain was set to 210, offset to 10, raw 16 format. I uploaded a few of the FITS files as http://s2.org/~jpaana/Light.zip if you want to take a look.

chad

Hi，

I think you should increase your offset. It is too small.

For now, with the time increase, more data is losing, you can find it in the histogram. I think it is the reason why the median value get smaller.

Thanks

Chad

jpaana

Thanks for the idea, offset makes some difference and eventually around offset 120 the longer exposures also start to rise and no longer clip, but shorter exposures still have way higher median ADU values than longer ones, which doesn't make sense if the data is raw. These two frames were taken right after each other, gain 210, offset 120, 0.01s exposure has median value of 644 while 1s has 95: http://s2.org/~jpaana/Light2.zip

Do you have a suggestion how much the offset should be raised and why the default is only 10 if it clips that badly?

chad

Hi,

The current one is still not enough.

You can use a software to check histograms of images.

You should add the offset to make sure the histogram like my example. The left side of the wave is complete.

Thanks

Chad

jpaana

Hi,

I tried raising offset all the way to 500 (gain still 210) which definitely lifted the range off the left edge for the shorter exposure, but some pixels of the longer one still clip. So the main issue that longer exposures have lower ADU values still remains, even the maximum values, it looks like around 400-500 ADUs have been subtracted from the longer exposure, which causes the low end to clip:

0.01s exposure has minimum value of 176, median 2517 and maximum 6475

1s exposure has minimum value of 4 still, median 1936 and maximum 6015

Both images have similar three peak histogram shapes, but 1s exposure has them at lower values. The fits files are again at http://s2.org/~jpaana/Light3.zip

jpaana

Just to test I also took a 100s exposure which already shows the amp glow quite nicely, but the values are very similar, minimum is still 4, median 1840 and maximum 65534. Maximum is due to a hot pixel, but median shouldn't go down unless the "correction" is applied more when exposure time increases. Average deviation is roughly similar in all exposures, 769.9 for 0.01s, 588.6 for 1s and 679 for 100s.

chad

Hi Sir,

I tested.

And i have the same result like you. And the following is our analysis:

1st, for cmos sensor, the Pixel‘s sensitivity is not very linear, expecially at exposure of short time. After 0.7 seconds, this value changes linearly.

2nd, for 178, there are some unknown operations inside the image sensor. This will cause this value to change irregularly. We had asked sony, but they did not answer us.

Thanks

Chad

jpaana

Thanks, this does make calibrating longer exposures a bit challenging, but I plan on using the camera mostly for planetary and lunar imaging where it probably doesn't matter as much. I guess the default offset value could probably be a bit higher in the SDK for this camera though, now the default 10 causes clipping to black very easily and people might not know to raise it. But then again even 500 doesn't completely eliminate that so it's a bit hard to say what the value should be. Best would obviously be if Sony could allow disabling that "correction" in the first place.

Anyway, thanks for your time Chad, I hope there still is some better solution, but I think I can live with this for now.