the strange case of the missing pixels - ASI1600

[Shiraz (Ray)]

Peter pointed out that one of my ASI images had slight hot pixels that had not been fully removed by the calibration process - have been trying to pin down the cause.

It seems that the chip retains some signal processing from it's DSLR heritage - and it removes the hottest pixels. I looked at subs varying from 30 seconds to 10 minutes and at different gains and sure enough, when a single bright pixel exceeds its local background by about 8500ADU?, it blinks out of existence (presumably replaced with data from surrounding pixels).

implications are that dark scaling will not work properly when calibrating images from this chip. Darks must be taken using the same gain, offset, temperature and time as the lights. In addition, for bad pixel mapping, it is probably best to use darks taken at the same gain, offset and temperature but at slightly shorter time than the lights. This will ensure that all of the nasty hot pixels of interest are present in the dark and that there will be no odd hot pixels that have been substituted out by virtue of the read noise tipping them over the removal threshold. Depends on the exact mechanism, but it is also possible that this process may be able to remove stars in severely undersampled images - not tested, but maybe something to be aware of if very fast short fl lenses are used (will not be an issue with normal scopes).

This chip is turning out to be a bit "different" in almost every possible respect - it is taking a lot of time to come to grips with what it does so well and where it requires a bit of extra care.

[glend (Glen)]

Thanks for that info Ray. I always use Darks shot at the same gain, offset, duration, and temperature. My bias frames are also shot with the same gain, offset, and temperature.
I have Dark and Bias Library sets for what are becoming the standard settings that i favour. If i stray outside those settings i add new Darks and Bias frames to equate. So i have these various Master Darks and Master Bias that i add to DSS and do not have to restack/process each time. Hot pixel detection in DSS is used, as well as cosmetic cold and hot cleanup in final stacked filter image.
I haven't notice what you identified in my images but welcome Peter or anyone else reviewing them.
However, i don't know if i over drive a pixel given that my data always seems relatively skinny compared to others.

[Shiraz (Ray)]

Originally Posted by glend

Thanks for that info Ray. I always use Darks shot at the same gain, offset, duration, and temperature. My bias frames are also shot with the same gain, offset, and temperature.
I have Dark and Bias Library sets for what are becoming the standard settings that i favour. If i stray outside those settings i add new Darks and Bias frames to equate. So i have these various Master Darks and Master Bias that i add to DSS and do not have to restack/process each time. Hot pixel detection in DSS is used, as well as cosmetic cold and hot cleanup in final stacked filter image.
I haven't notice what you identified in my images but welcome Peter or anyone else reviewing them.

using your disciplined approach, there will not be an issue . Although I regularly take new bias data, I have been trying out various gain settings, depending on conditions - and have often ended up with less than perfectly matched dark calibration data. The old approach of using whatever darks were fairly close and letting PI sort out the scaling/details is no longer an option . eg, For one image that showed slight residual hot pixels, I had combined data taken at gain50/300s, gain100/300s, gain100/600s and gain200/600s. I had darks taken at gain200/600, but had tried to use scaled darks taken at gain100/600 for the rest - it didn't work perfectly.

[Camelopardalis (Dunk)]

Sounds like a case for regular dithering to me

[Shiraz (Ray)]

Originally Posted by Camelopardalis

Sounds like a case for regular dithering to me

Always dither Dunk, but the problem remained. I assume that, if some slightly brighter pixels remain in all subs, it is possible for some to overlap on a few subs (out of a few hundred) and fail rejection - the warmer pixels in the final image are only very slightly above the surrounds, but they stand out with sharpening. Still not altogether sure I understand what is going on, but using matching darks fixes the problem. Am pretty sure that the on-board processor gets rid of hotter pixels though, so assume that it has something to do with it.

[Atmos (Colin)]

So as long as darks match lights, doesn't matter

[Camelopardalis (Dunk)]

Interesting observation Ray I've only got one set of data out of my 1600 so far (due to travel, weather) so it'll encourage me to take a closer look at it when processing.