DBE can't eliminate vignette

Ken Bates

I’m having a problem with background modelization, and can’t seem to get things right. The jpeg (stretched) straight out of integration is attached, and as you can see, there appears to be a significant vignette. I’ve tried ABE, as well as DBE, using both subtraction and division, and also placing the DBE points automatically, manually, and as an axial. All of these result in horrible splotches of color around the outer periphery that I can’t seem to get rid of.

The images (10+ hours total) were taken over three nights, and a random sample of a single image on each of the three nights shows a similar pattern when debayered. There’s no vignette that’s apparent (to me) un the un-debayered images, although that may be normal.

I’ve taken another series of flats on the assumption that my previous ones were bad, but the result is exactly the same; an obvious circular color pattern around the center of the image that I can’t seem to get rid of with DBE (or ABE). Any idea what might be causing this and/or what I could possibly do to correct it? This is the first time I’ve run into anything like this, and it’s very frustrating. After several hours of trying to fix it, I’m about ready to just do a severe crop and go from there.

The xisf file seemed to be a bit too large to upload, so it can be accessed from this Dropbox link.

BlockHead

Ooooh...I have seen this before. 

Pedestals man... over/under correction always means pedestals. 

You have a calibration problem. It is either because you are attempting to use optimization (and if you are watching MY videos...you know to do this with CAUTION)... otherwise... bad darks. 

Please see this video:

-adam

Ken Bates

Thanks Adam; I've watched all the Fundamental videos, but must have missed that one. Interestingly, I've been using the same darks and flats for earlier targets, and this is the first time this has happened. This is also the first time I've used WBPP, since I usually do the manual stacking steps (per your videos).

I just redid the processing using the manual steps, and although there's still a slight 'vignette', there's a huge difference between the manual and WBPP results in the apparent vignette, brightness, and background. I think what I need to do now is track down why WBPP doesn't work well for me when the manual steps do.

I do prefer doing things manually since I can examine the results of each individual step, but it would be good to know why there's such a difference between WBPP and manual.

Thanks again for your suggestions, as well as all your video tutorials.

BlockHead

Post the screen shot of your control panel with the files set up. 

There should be no difference in the results. 

-the Blockhead

Ken Bates

Thanks for looking at this. I agree, it's really puzzling that there is a noticeable difference between WBPP and what should be and the identical manual processing steps. I'm sure it's something that I'm doing incorrectly, but I can't figure out what it is.

A screen shot of the WBPP control panel is attached. All the settings should be default since I haven't changed anything except pointers to the files. I'll go over the individual WBPP settings today and see if they're any different than what I use in the manual processing flow.

Ken Bates

I'm also attaching jpegs of the unlinked auto stretched master flat, WBPP processed image, and the manually processed image. There's not only a difference between the processed images, but they also appear to be 180 degrees rotated with respect to each other (I didn't do that).

BlockHead

Hi Ken,

First things... In your screen capture it shows you are not calibrating your Flat field images. 

You do not have a matching dark frame for the Flats. 

Thus, it appears your intent is to use a bias frame to calibrate your Flats. This is fine as long as your sensor is the right kind to allow this.

Please click on your flat group. It should tell you a warning that there is no dark frame with a difference of less than 5 seconds compared to your flat exposure. Correct?

You need to uncheck the use dark matching..and this forces the use of the bias frame.

WBPP will communicate that the configuration is probably not what you intend when you click the Calibration Diagram. You will see that your flats are not be calibrated by anything at all.

-the Blockhead

Ken Bates

Thanks, Adam. AFAIK, using bias for works fine when I do a manual stacking, although I can certainly take some flat darks and see how that goes.

I did click on the flat group, but there was no warning (and diagnostics showed no issue). Also, where is the "dark matching" check box? I can see in the flat group that the calibration settings are blank, bit I don't see how to set them.nor can I find anything resembling 'dark matching' in any of the WBPP panes.

Thanks again.

BlockHead

Oh... my bad. You have loaded masters. 

So, I need to see the configuration you used when you loaded the RAW files in to create the masters. Yes, the flats will need to be bias subtracted if you do not have matching darks. 

Can you show *that* configuration. At this point, whatever happened is baked into your masters...so I can't tell. 

-the Blockhead

Ken Bates

The flats were created a couple of months ago, and I can't remember if they were created with WBPP (version 1 or 2?) or manually via image calibration and image integration processes. At this point I think you've narrowed the possibilities down sufficiently that I can take it from here.

I'll do some dark flats, as well as recreating the "light flat" masters via both WBPP and individual PI processes, and try most combinations to see what does and doesn't work. Cloudy skies and intermittent snow starting Monday for the next week, so I'll have plenty of time to experiment.

Thanks again for the help and advice. I think you nailed it with the flat issues, so now I just have to make sure that everything is in sync. 

BlockHead

Yeah...my suggestion is to start from raw data and run everything through WBPP.

This will ensure everything is consistent. Then, if you want, you can also run the data manually...

the answers will be identical if the configuration is identical. There is simply no magic math happening...

-the Blockhead