Sorry I wasnt clear...the preblur is only to aid alignment - it doesnt blur the final image. It can work actually, but its a pain.

Does the QHY163m have this checkerboard like the ASI1600mm?
#26
Posted 30 May 2017 - 09:28 AM
#27
Posted 30 May 2017 - 09:31 AM
Why can't you calibrate this out in pre-processing? I thought most the planetary stacking programs accepted a master flat.
If you could divide out a master flat, that should help. That said, flat division is not guaranteed to completely correct this kind of FPN, so I would still recommend dithering. If you have a really stable scope, that might mean taking shorter video sequences and shifting the position of the planet between each one. Something along those lines.
Hi,
Because of seeing and the generally very long FL there is enough motion in the sequence that dithering isn't needed. I have never heard anyone dithering during planetary work but perhaps it is done. In fact many use the auto guiding feature in FireCapture to keep the target on the small ROI for the duration of the exposure. This causes enough motion that dithering isn't needed. In any case, this motion does not appear to remove the pattern.
Sander
#28
Posted 30 May 2017 - 09:32 AM
Sorry I wasnt clear...the preblur is only to aid alignment - it doesnt blur the final image. It can work actually, but its a pain.
Ah. You were plenty clear it's just that I did not understand the feature properly :-) In that case I don't know why the preblur did not work for me.
#29
Posted 30 May 2017 - 09:44 AM
Sorry I wasnt clear...the preblur is only to aid alignment - it doesnt blur the final image. It can work actually, but its a pain.
Ah. You were plenty clear it's just that I did not understand the feature properly :-) In that case I don't know why the preblur did not work for me.
I found that a preblur of 3 or 4 is needed to reduce the grid to a level where real features mostly dominate - at least for the moon - and even then you see some grid in 'featureless' areas. The grid is such a strong pattern it's hard to make it 'disappear' completely - without binning!
Edited by happylimpet, 30 May 2017 - 09:44 AM.
#30
Posted 30 May 2017 - 09:51 AM
Thanks I will give that another try just for giggles (tears). I'll also preprocess the same stack with PIPP and created a binned master. I think it can do that. I'll then be able to compare the results.
#31
Posted 30 May 2017 - 09:56 AM
If you could divide out a master flat, that should help. That said, flat division is not guaranteed to completely correct this kind of FPN, so I would still recommend dithering. If you have a really stable scope, that might mean taking shorter video sequences and shifting the position of the planet between each one. Something along those lines.Why can't you calibrate this out in pre-processing? I thought most the planetary stacking programs accepted a master flat.
Hi,
Because of seeing and the generally very long FL there is enough motion in the sequence that dithering isn't needed. I have never heard anyone dithering during planetary work but perhaps it is done. In fact many use the auto guiding feature in FireCapture to keep the target on the small ROI for the duration of the exposure. This causes enough motion that dithering isn't needed. In any case, this motion does not appear to remove the pattern.
Sander
If this is the case, then it sounds like the problem is possibly not the camera, but something occurring during post processing. If it was from the camera, dithering would eliminate it.
#32
Posted 30 May 2017 - 10:02 AM
Oh - and there is another option....you can tell autostakkert its a bayer array, stack it and then convert back to mono. Theoretically this shouldnt cause any loss in resolution, but the conversion to mono will 'slightly' reduce SNR as one of the colours has twice as much data in it, but receives the same weighting as the other colours. But this is a detail - its a method worth using.
#33
Posted 30 May 2017 - 10:04 AM
If you could divide out a master flat, that should help. That said, flat division is not guaranteed to completely correct this kind of FPN, so I would still recommend dithering. If you have a really stable scope, that might mean taking shorter video sequences and shifting the position of the planet between each one. Something along those lines.Why can't you calibrate this out in pre-processing? I thought most the planetary stacking programs accepted a master flat.
Hi,
Because of seeing and the generally very long FL there is enough motion in the sequence that dithering isn't needed. I have never heard anyone dithering during planetary work but perhaps it is done. In fact many use the auto guiding feature in FireCapture to keep the target on the small ROI for the duration of the exposure. This causes enough motion that dithering isn't needed. In any case, this motion does not appear to remove the pattern.
Sander
If this is the case, then it sounds like the problem is possibly not the camera, but something occurring during post processing. If it was from the camera, dithering would eliminate it.
No, because the stacking software locks onto patterns, not stars. This isnt deepsky stacking. The FPN is a pattern, and the software locks onto it. You can dither all you like, and the software will still sense the grid and ensure it is coherently locked onto.
#34
Posted 30 May 2017 - 10:26 AM
No, because the stacking software locks onto patterns, not stars. This isnt deepsky stacking. The FPN is a pattern, and the software locks onto it. You can dither all you like, and the software will still sense the grid and ensure it is coherently locked onto.If this is the case, then it sounds like the problem is possibly not the camera, but something occurring during post processing. If it was from the camera, dithering would eliminate it.
I am not really sure of that. I have examined hundreds of my individual flat, dark, bias and light frames. The checkerboard pattern is not visible in any of them. The pattern only barely begins to show up once you have stacked a few dozen frames, and for it to really show up well, you gotta stack well over 100. If the software was somehow capable of latching onto a pattern in each frame, then it wouldn't be able to "see" a checkerboard pattern in the first place...not in each individual sub.
I think there must be something more complex going on. I don't know what it is, but it does not make sense to me that somehow the software would be latching onto an aspect of FPN that is 50x fainter or more than the random noise of each sub...
#35
Posted 30 May 2017 - 10:31 AM
No, because the stacking software locks onto patterns, not stars. This isnt deepsky stacking. The FPN is a pattern, and the software locks onto it. You can dither all you like, and the software will still sense the grid and ensure it is coherently locked onto.If this is the case, then it sounds like the problem is possibly not the camera, but something occurring during post processing. If it was from the camera, dithering would eliminate it.
I am not really sure of that. I have examined hundreds of my individual flat, dark, bias and light frames. The checkerboard pattern is not visible in any of them. The pattern only barely begins to show up once you have stacked a few dozen frames, and for it to really show up well, you gotta stack well over 100. If the software was somehow capable of latching onto a pattern in each frame, then it wouldn't be able to "see" a checkerboard pattern in the first place...not in each individual sub.
I think there must be something more complex going on. I don't know what it is, but it does not make sense to me that somehow the software would be latching onto an aspect of FPN that is 50x fainter or more than the random noise of each sub...
There is variation between cameras, and I suspecty you use yours at a lower gain setting than planetary imagers, which I believe makes the grid less evident.
However, with a typical alignment point in planetary imaging being perhaps 80 pixels square, thats 6400 pixels across which to cross correlate a fit to the pattern. It can be buried in the noise visually, but still it fits the alignment algorithm with fewer residuals when the girds align.
#36
Posted 30 May 2017 - 10:49 AM
Edited by Shiraz, 30 May 2017 - 11:07 AM.
#37
Posted 30 May 2017 - 03:43 PM
I have not yet tried that. Taking solar flats is tricky. At least I could not get them done with a PST. The standard technique is to defocus the sun but then your optical train doesn't match anymore so I don't see the use of that. I tried to create Ha solar flats when I had Newton rings. I was not able to blur the sunlight while still allowing Ha through.
#38
Posted 30 May 2017 - 06:32 PM
https://stargazerslo...lar-flat-field/ this guy suggested a diffuser for solar - haven't done NB solar myself, but it looks like a reasonable approach.
Although, come to think of it, maybe your binning solution is probably good enough for solar? - my understanding is that daytime seeing rarely supports really fine resolution. Ray
Edited by Shiraz, 30 May 2017 - 06:35 PM.
#39
Posted 01 June 2017 - 01:29 PM
Yes, for full frame moon shots flattening is certainly possible. I'll give it a try.
#40
Posted 11 November 2017 - 04:33 PM
I ran into this on my moon shots and managed to solve last night - now just discovered this thread, LOL. Wondered what the hell was going on TBH. Had all sorts of weird downstream processing effects in Photoshop and subsequent rendering too, which led me up some blind alleys in search of a solution!
Fix for me with the moon was simply to denoise Layer 1 Wavelets in RS6 to 0.2 with 0.15 sharpening and 100 preview.
Basically, I'd forgotten to denoise Layer 1 at all, originally. I could still leave denoise at 0 on Layer 2 with 0.10 sharpening and 70 preview. No other layers needed in my shot (all off).
I will also look for that driver setting mentioned in this thread - was just on defaults here.
Nice and crunchy result here:
http://www.astrobin.com/full/321125/0/ or
Before and after 1300% crops attached.
Edited by dswtan, 11 November 2017 - 11:32 PM.
#41
Posted 13 November 2017 - 12:24 AM
Hello,
I think this is due to the channel difference between R,G,G,B channel. Many cmos sensor has this problem because it designed as color version. For mono version the circuit inside the cmos sensor is the same, just the different that without bayer filter.
If the four channel has different parameter, it will cause this problem.
If the different parameter is the offset (like our QHY5II-M which has similar problem and we give an option that adjust the R,G,G,B offset individually and this can solve the problem in camera side). But not all sensor has the R,G,G,B offset register inside.
For offset , normally it can be calibrated by dark frame or bias frame. Because the offset is a constant for each panel. after subtract it the different will be gone.
If the different parameter is the gain. The dark frame or bias frame can not solve it . This is the multiple ratio . Only the flat frame can calibrate it.
Unlike CCD camera using only one channel to readout . The CMOS has some different so I think the calibrate method should be update. And we hope the image processing software have more tools for the CMOS camera.
The question is how about this performance in QHY163M. Actually I have not test it before. I will try that and see what's results. If anyone has the QHY163M flat image you can also show here. It is best to do a heavy stretch to show this flat frame.
Best regards,
Qiu Hongyun
Founder, CEO of QHYCCD
Edited by QHYCCD, 13 November 2017 - 11:07 AM.
#42
Posted 13 November 2017 - 10:38 AM
I have dealt with this checkerboard issue quite a bit with lunar imaging and the ASI1600MM. From a camera standpoint, there is definitely and issue the designer needs to contend with as Qiu Hongyun stated with more elegance than I could.
In my experience as an end user, there are two potential fixes.
- Use a master flat in the stacking process, this obviously works only if you are capturing the full frame and not ROI. I have found this to COMPLETELY correct the issue. My master flat has a very obvious checkerboard and it is due mainly to the photoresponse of the pixel. So, bias and darks seem to have no effect on the pattern for planetary. For deep sky, the obvious recommendation is to fully calibrate. For planetary, I get by with just flat calibration in autoskakkert. If Firecapture or Sharpcap allowed an on-the-fly flat correction with ROI cropping, that would be ideal and could allow ROI capturing.
- In the absence of flat correction, I have had very good success with Autostakkert's "experimental features" under the advanced tab. I apply a 1.5 pixel blur in the pre-processing module for vertical and horizontal. This completely wipes the pattern clean, but also comes at the cost of ultimate resolution. Fortunately, because the blur is created in software, it also seems to respond very well to deconvolution and, to be honest, I don't see that much of an issue with the finished product. It would be worth a try, if #1 is not an option. The key aspect that this addresses is that the stacking algorithm does not lock onto the pattern and only aligns on real features of the target. I am quite pleased with the result.
This is an issue with the sensor and it is not unique to your camera.
Edited by Thirteen, 13 November 2017 - 10:41 AM.