Strange background color after stacking with pixinsight?

[redfox1971]

Hello this is a image stacked in pixinsight batch proccesing script of 30 lights,a master bias of 100 and master dark of 50 with 15 flats taken with my QHY8l once opened in pixinsight and stf unlinked applyed i get a lot of pink red color to the sides whitch if i try to dbe the Tolerance has to be set way above 2000 to turn the samples white can anybody explain what is happening to my images before i go mad

[ollypenrice]

A few things jump out from this image.

1) You have inverse vignetting. This could have two obvious sources. Maybe the flats didn't calibrate properly with the bias subtraction. This causes them to over correct, so the outside becomes too bright and the middle too dark. Or (my bet is on this) you had too many DBE markers and some of them were much too close to the galaxy. Any background marker lying on what is really a bit of outlying galaxy will record a slight brightening over the background sky (which is fine because it's galaxy!) and then drag it down because it has been told that it's really background sky. I use few background markers and keep all of them miles away from the galaxy.

2) You have a huge imbalance in colour, weighted towards green. A dose of SCNR green should help no end, but first I'd re-do the DBE with fewer markers. Personally I'd hand-place just 9 markers on this image. It's vital to get DBE to read gradients and not image detail.

Olly

[redfox1971]

Hi Ollly this is how I see it before any work is done to the image,like you say it could be inverse vignetting as I don't get the problem if I don't stack flats and bias but then I see the dust bunnies on the image?.

[Erquy]

Hi,

So this is invert vignetting. I have been in the same situation. Do you have the histogram of you stacked flats? i did manage to adress the problem (not totally, but you can finish then with dbe) by manipulating a bit the master flat before stacking: move/transform the histogram of the master flat (keep the original safe!) and stack (use masters, and do trials and errors with fewer subs). If your imging set up has not moved, then try take different flats and find the right ones (or check around for flats methods)

My learning: keep your optics/sensor as clean as possible and do not go for flats if you do not have too severe gradient -dbe can take well care of it- (depends how bad is your gradients of course). I have not used flats for long, only pay very much attention not too have dirt on sensor glass/fr and that s make life easier.

[nucdoc]

This is my best attempt, I used DBE and selected 9 samples, I tried both subtraction and division, subtraction worked best.  Then used SCNR for the green, stretched the histogram a bit and used curves to darken the edges and emphasize the Galaxy, still needed to crop it to completely remove the vignetting.

Mark

[redfox1971]

Thanks for having a go Mark.I have reset my gain and offset once again on my qhy8l and had to up the offset to 113 whitch gave me rms value of 1038.5 just to get rid of the warning message offset to low,got a figure of 64189.1 with saturated image,since then i took a image of NGC 6995 and seems to be a lot better the first image straight histogram transfer in pixinsight and the second cropped and processed.Just got the problem of amp light on my darks of 300seconds whitch i see i need to go back to a older driver version as the latest made it worse and qhy dont seem to wont to solve this problem.

[ollypenrice]

I can't agree with not taking flats. I think they are vital and the harder your target is (ie the more faint signal it contains) the more important it is to have flats to distinguish between signal and uneven illumination. However, flats can sometimes be devils and I've had my share of trouble getting them right! Good as it is, DBE cannot distinguish between Integrated Flux and uneven exposure. Indeed Rogelio Bernal Andreo advises against DBE on very faint targets. I always use it on colour but rarely on luminance and never on Ha.

Firstly I would make a master flat independently of the rest of the image. Just put your set of flats into the box normally used for lights and a master bias, or set of bias, into the box dedicated to darks for your lights. Use average combine. The software will just treat the flats as images and this removes the possibility of something going wrong during the more complete stacking and calibrating routine.

If flats made this way still over correct and give inverse vignetting I would try a lower brightness for them (shorter exposure.)

Olly

[Robp]

Sorry to slightly derail the thread but do you combine all your flats in that way Olly?

I usually take 99 flats per filter/session and add them in to the normal stacking procedure with lights and bias frames. The way you suggest is something I've never came across before so wondered if it was worth adding to my workflow.