Introduction and request for help

[bianry]

Hi all.
I have been lurking here for about two weeks so I felt it was about time to step forward. And I need help....

A bit of background:
I am from Finland, 52 years, work as an IT-professional, write programs as a hobby and have been photographing my whole life. I even have my analog darkroom up and running.

Have been contemplating deep sky photography for a year but the cost of good equipment has been a hindrance. And I live in Finland. In the summer there is light around the clock and in the winter the sky is cloudy. There is a 2 month window in the autumn where you can really enjoy the stars.

A month ago I found iTelescope.net where you can rent a scope and just pay for the time. Just what I was looking for.
Have been taking some pictures just for practice.

A fellow astrophotographer recommended Pixinsight and I downloaded a trial license 2 weeks ago. Since then I have done nothing else it seems than learning this software. I consider myself fairly good at learning new programs but this one must have the steepest learning curve I have ever come across. And I have enjoyed every minute of it. I like the fact that it is NOT a set of procedures for doing this or that but a whole toolbox full of new, fantastic tools with names that no one can pronounce. Or remember. Thanks to Harry's Astroshed and Herr Wechselberger I have started to climb the slope. One cm at a time.

To my question.
Last weekend I took a series of lum-pictures of NGC 5033 with this telescope and camera. http://www.itelescope.net/telescope-t21/ A total of 17 frames. Bin 1. 180 sec each. Dithered.

From the site I can download both calibrated and uncalibrated frames. I will return in a later post to the calibration procedure since I am running in to the dreaded 'Warning: No correlation...' issue when trying to calibrate with raw darks from the site.

I have now used the calibrated frames in the following procedure:
- Cosmetic correction with a master dark.
- Star alignment
- Image integration

The problem I am having is that there is a coulmn of darker pixels at x-pos ~2060. It appears on some frames but not all. See attached pic.
The frames are integrated using Linear fit.

Code: [Select]

Integration of 18 images:
Pixel combination ......... average
Output normalization ...... additive
Weighting mode ............ noise evaluation
Scale estimator ........... iterative k-sigma / BWMV
Range rejection ........... range_low=0.000000 range_high=0.980000
Pixel rejection ........... linear fit clipping
Rejection normalization ... scale + zero offset
Rejection clippings ....... low=yes high=yes
Rejection parameters ...... lfit_low=5.000 lfit_high=2.500

Code: [Select]

Total :   5279729   4.662% (  4142036 +   1137693 =   3.658% +   1.005%)
MRS noise evaluation: done
Computing noise scaling factors: done

Gaussian noise estimates  : 8.9316e-005
Scale estimates           : 1.2412e-004
Location estimates        : 6.5516e-003
SNR estimates             : 9.2485e+003
Reference noise reduction : 1.1752
Median noise reduction    : 1.3163

94.30 s
If I use Winsorized Sigma clipping and pull Sigma low down to 1.000 the column disappears. But the number of rejected pixels go up to 50%.

I am sure there is some way to remove this other than throwing away about 1/3 of the frames where this pattern exists. HELP!!

I have tried a lot of different combinations with Cosmetic Correction but to no avail. I am also not completely clear on how to use the preview on that one.
- use a master dark
- open a frame
- make a Preview
- click Real time preview
- adjust Sigma sliders

How do I know when to stop adjusting? The count of hot pixels keep rising until all turns white. Is 3179 hot pixels a lot on a 6Mpixel frame??? What exactly does Show map do?

If some kind soul would like to take a look at the calibrated, registered, cosmetically corrected and integrated frame there is a link below.
https://www.dropbox.com/s/uwch9l71xd416zs/ngc5033.fit

I hope I am making some kind of sense with this ramble and look forward to learning a lot more.

Best regards

Mats

[sreilly]

My first guess is that there are bad columns that usually a dark would eliminate. It's hard to tell from a combined dithered image. a single raw image would show defects more clearly. I'd start by posting an uncalibrated raw image to see what you are starting with. Is the dark master the same length/temperature as the raw image?

-Steve

[bianry]

Here are 2 raws. One out of the camera and one calibrated by their system

https://www.dropbox.com/s/wy1ybuzypkxktgn/Calibrated-T21-bianry-NGC5033-20140222-031947-Luminance-BIN1-W-180-012.fit
https://www.dropbox.com/s/4jb0gc3wqyikrpm/T21-bianry-NGC5033-20140222-031947-Luminance-BIN1-W-180-012.fit

Regards

Mats

[jkmorse]

Mats,

Take a look at Gerald's tutorials on PixelMath.  He describes one routine for eliminating dark pixels in an image that may help here.  I had a problem with my Apogee Alta F16M that created dark horizontal bands stretching left and right from saturated stars.  Apogee has since repaired the issue, but Gerald's PixelMath solution at least helped somewhat in the meantime.  Its at least worth a try.

If you haven't found them already, you can find his tutorials here: http://www.werbeagentur.org/oldwexi/tipstest.html

And check out all of the other tutorials in his set.  They are great (my favorites along with Harry's).

Best of luck,

Jim

[bianry]

Yes I have looked at all his videos. Along with Harry's.

Great that we have such helpful people aboard.

regards

mats

[sreilly]

mats,

If you open all your calibrated images I think you may find that some have partial column defects which is perfectly normal. Where the problem comes in is when they aren't each corrected before combining especially if the images are dithered. If the dark frame is the same temperature and duration as the lights I would not use a bias frame for calibration. The dark should eliminate any defects. If it doesn't then you would need to address those before combining. It takes some extreme screen stretching sometimes to see them clearly but become very evident when combined and stretched. I haven't grasped the "Cosmetic Correction" using Batch PreProcessing so someone with a better understanding of its settings would need to help you with those. What I have found is , because I already own MaxIm, I use it to do my defect map corrections and then process the rest in PI.

I'd also check the quality of all the frames and only use the best. One or two not so good can make a big difference.

[bianry]

Thank you Steve. Yes I have identified that these dark columns appear in about a 1/3rd of all the calibrated frames. I have access to the uncalibrated frames also but the problem is that I can not get the calibration process in PI to function with the frames and darks. I have sent a message to PI support and hope to recieve some answer from them.

regards

mats

[pfile]

this is the "classic" 'calibration masters from another program' problem.

if you want to save pulling your hair out, just use the calibrated frames they provide you. if there are residual column defects, just use CosmeticCorrection to define the bad (partial) columns.

otherwise you are faced with the problem of getting the masters created by the other software to work in PI. it's not impossible but you have to carefully inspect the masters and figure out what scheme was used to write out the file. pixinsight always writes f32 images with values in the range 0.0-1.0. other programs do other things, so when reading one of those f32 files into PI you need to figure out how to scale it on your own.

rob

[bianry]

Rob.
I have access to the raw bias,dark,flats also but PI is rejecting them with the 'classic' Warning: No correlation.... message

The calibrated frames I get from iTelescope are OK. It just bugs that I can not use all the tools in my new arsenal

regards

Mats

[pfile]

are the bias/dark/flats in i16 format? if not you will run into this problem.

if they are master frames, then they are likely in f32 format. if you are talking about bias/dark/flat subs, then if they are not in i16 that's kind of a mistake on iTelescope's part. there's no reason to represent raw camera data as anything other than 16-bit integer data.

rob

[bianry]

This is a dropbox link to a dark sub from iTelescope
https://www.dropbox.com/s/vv6abkzcjivzdzt/2014-01-24--1213_Darks_180_077.fit

I have examined it and to my understanding it is 16 bit.

But, I am very new to FITS files and easily confused

regards

mats

[sreilly]

I'm going out on the limb here but I'm pretty sure all but the stacked image is 16 bit just based on the file size of 12,297KB. The image titled ngc5033.fit is 24,582KB, 32bit?

-Steve

[bianry]

Yes Steve, the ngc5033.fit is processed in PI and saved as 32 bit. The rest come straight from the camera.

regards

mats

[Harry page]

Hi

Can you supply one each of the following to look at

Light - bias- flat -dark

And we can then see what's going on

regards

Harry

[bianry]

Here is a Dropbox link to a folder with all the frames I have tried to feed in to the calibration. Both BPP and as separate processes.
https://www.dropbox.com/sh/u4y8lzb01lxo1qw/TxOWnfBSZ0

Thanks to all for helping.

regards

mats