vlaiv

Single operation in Gimp:

Problem is that you can't remove bloat from Luminance filter in this way as it captures all the wavelengths of light at the same time and some of it is bound to be out of focus as not all can be brought to a sharp focus at the same time. Luminance sharpness is very important for overall sharpness of the image - much more than color sharpness, as most of detail information in the image is carried by luminance layer. Here is little demonstration of the effect: I've found this nice bird image online that we will use to demonstrate effect: I'm going to blur first luminance component, and then both chrominance components in Gimp separately by same level of blur. We can then compare the two to see differences. It is not hard to tell the difference - chrominance blur is hardly noticeable, while luminance blur can't be missed. To further emphasize this point: Can you actually see any difference in star bloat between your two sample images once color is removed from them? Blue fringing is just aesthetic issue in images - real issue with unfocused light is blur that results from that. Blue fringing can be altered in post processing - there are even filters that remove chromatic aberration - or rather desaturate blue bloat from image automatically. If you want to have real impact on this and improve your images - you can do following: 1. Use RGB only and skip luminance. This way refocusing will have noticeable effect on sharpness of the image 2. Use special Lum filters that remove far ends of spectrum that are most offending - like Astronomik L-3 luminance filter 3. Use aperture mask to stop down your aperture a bit. This cleans up chromatic aberration.

No, no, no! Jpeg data is completely useless. Format uses 8bit data which clips away most of faint signal that is very important and on top of that - it adds compression algorithm artifacts. Jpeg was developed to alter image in visually acceptable way so that it compressed better. This alteration of data makes stacking useless as stacking expects "natural" noise and signal distribution in order to work properly.

@BrendanC If you want to test above - if sky flats are different to flat panel flats and if there is issue with filters, procedure is rather simple: - during daylight (or even better dawn / dusk) - point telescope to the sky (be careful not to point near the sun) and shoot some subs - just dozen or so. Be very careful not to saturate / over expose them as there is plenty of light (and why this is better done at dusk / dawn). - shoot flats and other calibration frames right there, using flat panel. Again - you don't need much of each, this is just for test purposes, dozen or so of each is good. - calibrate, stack, inspect. Resulting frame should not have any sort of gradient on itself - it should be perfectly "flat" (meaning same average ADU all over the place and uniform gray image when stretched). This test will rule out: - issues with filters - issues with telescope flocking and flats (if flat panel should be moved further out, or is it ok where it is sitting now)

Sensitivity can be used to explain at least pat of why it is visible now. Another thing to consider is filter wheel. It was not there previously. How light tight is it? It is mechanical component after all that needs to move freely. Don't think filters are to blame - I'm having hard time imagining what would cause them to have different response based on source of the light (they can't really tell where photon is coming from - they can only distinguish direction/location and wavelength - and that it shared between sky and flat panel).

Indeed - flip would rule it out as pre and post flip subs would have pattern rotated by 180 degrees. As all of the subs were taken same side of meridian - I'm guessing that we can't rule out dew/frost completely.

Few years ago - common wisdom was to use native drivers for high speed capture / short exposures like planetary and stick with ASCOM driver when doing long exposure. I did find that ASCOM drivers produced better result from long exposure, but I can't remember the details. From then on, I always advise people to use ASCOM drivers for long exposures. It probably has something to do with fast readout modes / electronics. Maybe something is in different mode with native drivers. Maybe you won't see any difference, and like I said - it is probably long shot, but I think in the end - its worth a try - nothing to loose really, and it might actually help

I see that you use APT. Out of interest, what driver for ASI1600 are you using? Ascom or native? If you are using Native / built in driver - I'd try ASCOM instead. Maybe even give NINA a go. I know it is a long shot, but sometimes drivers can cause issues.

Frost / dew on sensor can be recognized with meridian flip - it will rotate 180 degrees with respect to target. I just checked Blue, Red and Green stack as well. Pattern stays relatively the same with changing intensity - some patches are brighter and some fainter. Either all subs were captured on same side of meridian - or it's not dew/frost/sensor window related.

To be honest, I don't really see how it can be camera's fault? Camera is perhaps very sensitive and reveals flaws in the rest of the system, but I highly doubt it is down to it. Perhaps only thing to look for would be dew/fog/ice on camera window, but everything else should be fine.

For Olly's benefit (and for others who don't want to download) - here is two side by side stretched the same: And two side by side - stretched differently (with flats is stretched more) - enough to show gradients in each: both are linearly stretched (just setting black and white point to certain value while keeping pixel values linear). Neither has been altered in any way (no gradient removal).

My conclusion is that there is some sort of issue with light leak Here I have outlined where two stacks show differences (it might be better to look at original image without markings - markings are distracting, just take note of areas they are pointing to): Top image is brighter in two upper circles and darker in bottom circle. Here is what I read from this: 1. there is signal in the image that moves depending on where scope is pointing (issue with calibration files will be always in the same place - same calibration files used) 2. It is not due to high altitude clouds as gradients are too similar between first bunch of lights and second bunch of lights. Clouds tend to be blown away by wind rather quickly (they stay the same few subs at most) My guess is that this is signal that is captured while scope was tracking and it is external to scope as it changes places in the image. Not all of it behaves this way - but it points to the light leak of sorts. If there is light leak - some of it will change as the scope moves and some will stay the same (ambient illumination vs point sources). It has nothing to do with calibration files as these stay the same between first and second batch.

Well, they are more similar than different - but there is one feature that makes clear distinction between the two: Notice illumination in bottom right corner - top image is dark and bottom is bright. This is with LP gradient removed from both images so it is not due to that. Difference of two images clearly shows that there is difference in overall illumination: (images are not aligned - that is why we have "emboss" effect on stars and target - but look at background only - it should be even if two have same background - but it is not - right lower corner shows great difference).

And why flocking and baffling is very important. If scope is flocked and baffled well - then any ray that reflects more than once or twice is severely attenuated compared to principal rays and its signal falls below noise floor - and hence flats work. However - if you have anything shiny in path of the light - it will create problems.

On the other hand - I'm seeing something completely different: Zones of "over correction" You can't have both bright, medium and dark parts of flat over correct Over correction looks like inverse of flat so it should have following pattern: very white right edge and corners (as it is very dark in flat itself) and dark central part. In above stretched image - no corners are lighter then the rest of the image.

This image is wiped by StarTools - why don't you examine original stack that is still linear - distribution of bright and dark patches is rather different.

Why do you insist that this is the case? I just posted images above showing that gradients don't resemble vignetting on flats even a bit.

Here is an idea you can try in the mean time. Take your luminance files and split them into two groups - first half and second half Use same calibration masters on each group, but stack each group separately. After finishing - post both results here (just for inspection) - but more importantly look if they look differently when heavily stretched. If problem is with calibration files - both stacks must show same pattern of these strange gradients (when LP is subtracted), but if things are different after LP removal - then cause is something that changes due the course of the evening. Light leak on OTA can be both stationary and moving. Stationary will be light source moving together with OTA as it tracks the sky and that is less likely. In any case - we will narrow things down a bit if we examine two stacks.

Out of interest I calibrated master flat green with master flat luminance and got this: Apart from far corners (different level of vignetting for filters) - all I can really see is dust difference. In my opinion - flats work. I don't think that flats are over correcting in traditional sense. When I wipe LP gradients (linear ones) - this is what remains. I don't really see how that corresponds to flats. What we see here is simply not related to flats in my view - remember - flats have significant vignetting. For that reason - flats don't need to be long - these work just fine. It is scope? I would say 99% yes - some sort of internal reflection or light leak or something strange.

This has been common theme with CMOS sensors and ASI (and probably other vendors as well) - results seem to be heavily impacted by driver / firmware versions. I would not be surprised that this has actually been fixed in drivers in the mean time.

In addition, @BrendanC posted master flats - which show the same thing: This is ~90ms master flat for luminance. And Red channel at 0.34s There is no significant tilt in flat exposures.

I did not properly pay attention to this post - but I have several issues with it. This shows what appears to be influence of "global shutter" on flat exposure rather than non linearity of camera response. X axis is in pixels and not exposure length so above graph does not represent non linearity of camera - but apparently some sort of shutter artifact. Problem is that ASI1600 has a rolling shutter - not global. It is electronic rolling shutter and exposure length can't impact its operation (unlike with mechanical shutters - where you have to ensure that exposure is long enough versus shutter speed). I simply can't believe that above graph is true - it's certainly not true for my camera. Here are three examples of master flat files. First two are from RC 8" scope, third is from 80mm F/6 refractor. Third also suffers from slight OAG shadow in the top of the image (I did not pay full attention to completely avoid casting shadow) All three are made with very short exposures. In fact - it takes much more to download file in SGP than what exposure is (for some reason dowload in SGP lasts for about second although camera is capable of higher FPS). I shoot 256 flat exposures and associated flat darks in about 10 minutes. Never had issues with flat calibration of my images, and only gradients I had are from either passing high altitude clouds (which can easily be detected by doing frame animation) or linear LP gradients that are easily removed (and also change with the course of session as target is being tracked).

I'll see what I can do - my flat panel is extremely bright and I use millisecond exposures - like literally few milliseconds for luminance and 30-40ms for NB filters. Not sure if I'll be able to significantly decrease its brightness (order of x1000) in order to shoot 5s exposures.

I haven't had issues with flats on this camera with even very short exposures (few milliseconds). My model is one of early models - not pro version. Both of my scopes are well baffled - refractor and RC. I can see how newtonian could be bit of a problem with respect to flat calibration since it is not well baffled design. If you want to check linearity of your flats - there is simple way to do it. Take flats with few different exposures and calibrate them one against another. They should produce flat even images regardless which exposure you used to calibrate another.

Ideally, you want some sort of planetarium software that can direct telescope to desired location - for example Stellarium (sorry, I don't know if there is Mac version, don't use Mac, but I'm sure there will be something similar - like Sky Safari) and you'll need ability to plate solve. That usually comes in form of plate solve library that you install to use along software used to capture images (different software for capture can connect to different plate solve libraries).