Low rejection pattern

Colin Clackson

I'm getting a checkerboard pattern on the low rejection image of a master light. The master light is an integration of three hours of O3 data taken with a 1600MM pro camera. I don't get this pattern in any other lights, even those taken through an Ha filter. I wouldn't be troubled by it but the master light has a faint, fine, mottling that looks sort of wind-swept. 

The guiding program dithered between each frame.

Any suggestions on where to start to investigate the cause of the pattern?

BlockHead

Oh yes...I know that pattern. 

It means you are not using a pedestal for your NB data. 

You SHOULD actually be troubled by this... because you are losing data (in the form of zeros from over subtraction). Please see my famous video on this:

https://www.adamblockstudios.com/articles/step-up-your-narrowband-imagery-pedestal

This very pattern is demonstrated in the M51 processing demonstration in Horizons. Part 2 specifically..

https://www.adamblockstudios.com/categories/m51

-the Blockhead

Colin Clackson

Thanks. I knew I had seen it somewhere in your videos but couldn't remember which one.

Colin Clackson

So I ran the O3 images through WBPP again setting the pedestal to automatic. Resulted in DN of 1000, which the process console helpfully labeled as "Huge!" so I'm assuming that is undesirable.

Would an increase exposure time help resolve the issue?

BlockHead

No, increasing exposure time will not help most likely unless you can get your sky counts to 1,000!  But did the 1,000 help the issue?

My guess is that you need to electronically increase the offset of your sensor first. You may have zeros in your data right from the get-go. Did you look at your dark frames for zeros? (You should never have any.)

However, I disagree concerning the pedestal. Once you have the image in PixInsight you are working in a normalized value space (32-bit blah blah). The 1,000 that is added is just a constant that does not affect the data at all..and there is no loss of information. If you tried to keep the data in 14-bit range... then yes, 1,000 is marginally significant...but even then you do not need to worry! You will not be achieving bright values in NB. You will never get to the point that it matters in terms of the nebula. 

Finally... the automatic pedestal will still leave a small percentage of zeros... whereas you can find a value that results in no zeros at all.

-the Blockhead

Colin Clackson

Thanks always for your quick response. 

The 1000 DN took the checker board pattern out of the low rejection image, but didn't have an appreciable effect on the faint mottling in the master image

I  checked the darks as you suggested using your pixel math method - no zero value pixels.

I'm currently using an offset bias of 64.

Not sure if its of any interest but here are the stats on the low rejection with the pedestal in place:

masterLight_BIN_1_4656x3520_EXPOSURE_300_00s_FILTER_O3_mono_rejection_low

K

count (%) 100.00000

count (px) 16389120

mean 1742.093

median 1008.230

avgDev 1784.869

MAD 1494.802

minimum 0.000

maximum 61502.077

BlockHead

Once the checkerboard is removed... the other effects are not about the pedestal issue. 

The faint mottling can be caused by poor dithering, lack of enough exposures and old dark frames. 

The  mottling I see is diagonal... which strongly indicates dither is not good enough. But I always guess from a point of a severe lack of information.

-the Blockhead

Colin Clackson

I'm dithering up to 16 pixels after every frame, although the dithering sometimes doesn't execute correctly (or at all). Didn't think it would impact the frames but I guess I should pay a little closer attention.

Thanks for your thoughts.