Problem with DSS - What Am I Doing Wrong?

[fatwoul]

OK, I've had this happen once before, long ago, and I know it's something that's been brought up on here, but I can't remember how I fixed it.

Last Friday, I pointed the scope at NGC7380, and collected a bunch of lights (x5 900s), darks (x5), flats (x10) and bias frames (x10). I shoved them all through DSS (still using that, I'm afraid ), and all the nebulosity visible in the light frames disappeared! It was as if the dark and bias frames made the image WORSE!

Here are some screenshots to show what I mean. Obviously the adjustments I've made aren't done properly, but were just rushed to see if anything was visible in the resulting images.

Just the light frames:

Lights and flats:

Lights, flats and darks:

Lights, flats, and bias:

Lights, flats, darks and bias:

So what's awry? As you can see, the histograms fizzle away into little vertical lines as soon as either the dark or bias data is added.

I can't think of anything in my workflow that is different to previous sessions which haven't presented the problem. All I can think is maybe I fiddled with something when I was using DSS to stack Garradd subs a fortnight ago, and messed something up.

Any ideas, gang?

Thanks for looking.

[fatwoul]

GAH! Same thing just happened with some NGC7023 subs I was doing - L,R and G were OK, but B did the same as above! What is going on?!

[AndyUK]

Hi Robin - I don't know, but if you can upload your 7380 subs / calibration frames into one dropbox folder, and the 7023 subs into another, I'll happily put them through my DSS and see if there's any difference(?). If there is, I'll give you a call and we can talk through the screen settings...

[JimStan]

Does it have anything to do with the fact that he is using square root of LOG for his stretching ? I am just a beginner at processing with DSS, but I have played around with those different stretching functions, and they DO create a significnt difference in the final stack.

[AndyUK]

Hi JimStan - I must admit, I've never played with that setting in DSS myself, but my setting's exactly the same...

[fatwoul]

Hi Robin - I don't know, but if you can upload your 7380 subs / calibration frames into one dropbox folder, and the 7023 subs into another, I'll happily put them through my DSS and see if there's any difference(?). If there is, I'll give you a call and we can talk through the screen settings...

A gent as always, Andy. The files are in DropBox whenever you'd like to have a look. Cheers.

Does it have anything to do with the fact that he is using square root of LOG for his stretching ? I am just a beginner at processing with DSS, but I have played around with those different stretching functions, and they DO create a significnt difference in the final stack.

Thanks Jim I'll give that a try.

[fatwoul]

OK just tried all the other stretching modes with no change.

It was a good idea, though. Anyone got any others?

[AndyUK]

Hi Robin - I'll send you an email (or we may need to talk...!). Your bias and dark frames are quite different from mine

[Paxo]

Can you post up a dark and bias frame, they seem to be the culprits here.

Steve

[AndyUK]

Well, fits headers aren't exactly my specialty(), but for comparison purposes, my 600s dark and bias frames both have cblack and cwhite values of 0 and 65535 respectively.

However, the dark frames here appear to vary from cblack 211-2899 and cwhite 325-4197, and the bias frames have cblack at about 740 and cwhite about 1130.

I can see from the fits header that the darks were all taken at 600s and the flats at 0.001, but I think these cblack and cwhite values is where the problem lies - Of course It SHOULD be easy to resolve (ie take some more darks / bias frames ), but it is rather odd... I always thought that flats were the tricky b*ggers!

[Jgw]

Hi Fatwoul.

Send me please your subs & Darks, I would like to have a look at your pictures.

I had the same prob at the beginning...

Ps... I cant find your Subs in Dropbox..

Cu John

[fatwoul]

Thanks everyone. With some help from Andy, I think I've worked out where I was going wrong.

I thought the options In Artemis' "Display" dialogue box referred to options of how the images were displayed and nothing more. Afterall, that's where you switch on/off the reticle, but that reticle isn't saved as part of the image.

However, if "Auto Stretch" is selected in that dialogue box, that DOES get saved into the final image, and does affect the dark and bias files.

I have since shot some replacement bias and dark files, but of course the original subs are also damaged by the auto stretching that happened when they were taken, so the new calibration frames can't do much to remove the excess noise in the light frames.

Still, I think that solves the problem, so the next session will (I hope) be more successful.

Thanks again, everyone.

[Freddie]

Hi,

If you have a look at page 7 of the quick start guide for the camera, it says that the auto stretch will only affect the screen image and not the saved image. Must admit only affecting the screen image is my experience as I always have the auto ticked, can see plenty of detail in the screen image but not in the saved subs.

Freddie

[fatwoul]

Hi Freddie,

I shot two sets of bias and dark frames. First set with auto stretch still switched on, second set with it switched off. I didn't change anything else. The resulting FITS images were completely different when viewed in Maxim.

I would have expected the "display" options to be precisely that, but it wasn't my experience when I tried it. Andy can confirm that the bias and dark subs I sent him were very heavily clipped at both ends. I'll see if I can put a couple of examples on here.

[fatwoul]

OK, here are examples of the bias and dark frames respectively, taken with Auto Stretch switched on:

Now here are the bias and dark frames taken with all the settings identical, the only exception being that Auto Stretch was switched off:

The original images are obviously the same size. I suspect the reduction in size in Photobucket for the first two images is due to them being more memory intensive/less compressible as jpgs.

I can't explain it, but that's what I'm looking at.

I agree with you, Freddie, that this seems counter-intuitive, but I can't find any other explanation.

[AndyUK]

Hi both - I can indeed confirm that the cblack and cwhite settings were most definitely awry in the first bias / dark images... (and my darks / bias look exactly the same as the 2nd set!). Most odd...

[Freddie]

Interesting. Andy, do you find the same i.e. the autostretch does affect your darks and bias. Do either of you find that autostretch affects lights?

Maybe Vince from Atik has a view on this? I'll send him a PM.

Freddie

[AndyUK]

Hi Freddie - To be honest, after my initial experiences with the "pre" option, I try to leave as much as possible on the display and exposure screens well alone, but I do have the auto-stretch option checked.... BUT, the black slider is always set to 0 and the white slider to 65535.

[vince]

Hi guys,

I had a word with the boss! (Steve) and he reinforced what I told Freddie by PM earlier, the autostretch is just for the displayed image on the screen at the time of capture. None of the setting in the manual stretch or autostretch are saved with the image in Artemis capture.

fatwoul have you tried using our Dawn software for processing?

Hope this helps

Vince

[fatwoul]

Thank for replying, Vince, and thanks for consulting with Steve, too. Unfortunately, Steve's assertion that the AutoStretch settings are not saved into the image just isn't my experience.

Having read your reply, I decided to go out to the obsy again, and from scratch shoot some bias frames (because they're the quickest to do but still demonstrate the effect), using my Atik 314L+ Mono.

This time, I also took screenshots of Artemis at the time of capture, so there can be no doubt that the settings have been altered in any other way.

Here is a shot of Artemis cycling images of 0.001sec with AutoStretch switched off:

Now, a shot of Artemis shortly afterward, with AutoStretch switched on:

I won't post the 60 with/without shots, but I here is a screenshot of the resulting saved .FIT images, as thumbnails:

Those thumbnail previews are available as the computer recognises .FIT files as MaximDL images, and uses Maxim to automatically open them. I haven't used Dawn to process any images, nor have I yet tried Maxim for anything other than viewing subs, etc.

However, the software used afterwards shouldn't matter.

My question at this point is this: If AutoStretch settings are not saved as part of the files saved in Artemis, and AutoStretch supposedly has no effect on those images, how is it possible for any other programme to show the images differently? If the AutoStretch settings are not saved into the image, no other software would know those settings had ever been applied.

Lesser questions would be why it should matter what software I use in processing the images, and why Andy's AutoStretch doesn't seem to have as dramatic an effect on his subs.

I'd be happy to put these files onto DropBox for Vince or Steve to inspect the individual subs for themselves, but as you can see from the images above, nothing other than AutoStretch was altered, but the effect on the resulting images taken is clear.

Sorry if I'm coming across as a little frustrated at this point. I appreciate everyone's help in this matter. It just feels like I'm being told that Dulux don't make red paint, whilst holding a tin of Dulux red paint!

Andy - If it's not too much of a faff, would you mind maybe firing up Artemis, with your 314 plugged in, and trying something similar. In particular, I'm interested about your mention that whilst AutoStretch is switched on, the sliders remain at 0 and 65535. Screenshots of Artemis at the time would be good to see.

I should also mention that I am using Artemis Version 3.25.0, as provided on the discs that came with my cameras.

[vince]

Hi

I think I confused myself there for a moment! I should of said the screen stretch doesn't effect the data in the saved image, you still have the full dynamic range of the camera saved. The values of the applied screen stretch are stored in the Fits header for use by all programs (sorry been a long day!) If you want to send me a couple of Fits files I'll be happy to have a look. PM me and I'll send you the email address.

Regards

Vince

[fatwoul]

Thanks Vince. I'll PM you now.

[AndyUK]

Hi Robin - Vince's response aside, I did promise I'd do some quick bias tests of my own with bias on / bias off... which surprised me, although given Vince's response, maybe not quite so much now.

With stretch on, I did get the "grey" image (ADU cblack=215 cwhite=317).

I then took stretch off and didn't change anything and the next set of bias frames were.... grey (same ADU values).

I then manually changed black to 0 and white to 65535 and took them again with stretch off, and got the black frame (ADU cblack=0 cwhite=65535).

I had a quick look at all 3 in Maxim, and, as Vince said, the cblack / cwhite ADU settings were contained in the fits header (as above), and they all looked like the above, but when I looked at them all in DSS, they were all BLACK - Here's a screendump of the image with stretch ON:

(I'm not sure that this really explains how your original darks / bias frames looked grey in DSS )

[Freddie]

Hi Andy,

What version of Artemis Capture are you running?

Freddie

[AndyUK]

Hi Freddie - V3.35...