I am in trouble with this one.... need some help!

[Sedat]

Hi folks,

Could you please give me a hand and process my data to show what can possibly be done with it ? I will really appreciate this.

https://rapidshare.com/files/1051544027 ... LA_HDR.rar (64 bit fits)
https://rapidshare.com/files/3163770842 ... LA_HDR.zip[/url] (16 bit tiff)[url]

I am a relatively new PI user and still have a long way to go to be able to use it properly. I must confess however, contrary to everyone's belief in the PS camp, my progress is much speedier and more fruitful compared to my PS efforts. Kudos to the developers and supporters once again.

I still have some problems with DBE, color calibration and noise reduction. For some reason MMT is not very effective on my pictures. I am suspecting the culprit to be my shooting with a DSLR. There must be an inherent noise due to the missing pixels, blocked by CFA. Whenever I apply MMT ot ATWT the background doesn't look silky smooth at all. I am sure I must be pushing some wrong buttons at some stages as well.

I have been playing with this Bode's Nebula data for the last couple of weeks. The stack I  share with you is an HDR composite of the following:

Scope: FSQ106ED @f/5
Mount: Vixen New Atlux with Starbook
Focuser:
Camera: Canon 550D - Hutech modified
Filters: None
Guide: Vixen A80SS, Orion SSAG

Exposures:
First set: 24x360 second Light @ISO1600, Master Darks, Bias, 24x Flats, 24x Flat Darks
Second set: 11x360 second Light @ISO800, Master Darks, Bias, 24x Flats, 24x Flat Darks
Shooting Date: 26.02.2012
Orientation:
Acquisition: BackyardEOS
Guiding: PHD
Stacking: Images Plus
Processing: Pixinsight
Location: Marmaris - Turkey
SQM: 21
Temperature: 9 C

This one is a quick process of the ISO1600 stack



and this is the result of the HDR composite...There are many errors I know but I guess I need a break and go back to basics before further attempts.


Can you see my White Flag waving ?

Thanks for reading and your help.

Cheers

Sedat

[Sedat]

In case you had difficulties with Rapidshare services, I have uploaded the files to Pixinsight file server as well;

Fits file --> http://endor.uv.es//data/public/7b488b41ae5dfe28fe5bfc0c69561efc.php?lang=en
Tiff file --> http://endor.uv.es//data/public/c9d49969a32ff239655770cd07687d72.php?lang=en

Hope some of you can find the time to fiddle with this data.

Best wishes

Sedat

[ruediger]

First set: 24x360 second Light @ISO1600, Master Darks, Bias, 24x Flats, 24x Flat Darks
Second set: 11x360 second Light @ISO800, Master Darks, Bias, 24x Flats, 24x Flat Darks
[...]
and this is the result of the HDR composite...

I did not look into your data yet, but I'm curious to know, why you want to make a HDR from the two sets that basically carry the same information. An ISO 1600 lightframe is just a up-scaled version of an ISO 800 one and it should be much easier for you to throw all lights into one stack and process them together.

What was your intention in changing the ISO-settings?

Rüdiger

[Sedat]

Hi Rüdiger,

At the time of shooting, it suddenly occurred to me  that 6 minutes sub exposure might have been to much at ISO 1600 which would lead to saturated galaxy core and I switched to ISO 800 for the rest of the session. When I shoot with my C11 Edge+Hyperstar, this can easily happen. This FSQ belongs to a friend and it was my second attempt to shoot with it. So I am not very familiar with its outcome yet. There was another scope mounted on the side bar for testing and comparison purposes, coupled to my Canon 450D camera. Due to this, I was not able to dither properly. I believe some of the problems are related to this.

At the first attempt, with the help of dithering, I managed to obtain a nice Leo Trio:


Leo_Triplet2 by bilgebay, on Flickr

If you are suggesting that I should stack the outcome of the 2 sets via Pixel Math or any other tools, I can do that too. Please let me know.

Thanks for taking the time to look at my data.

Sedat

[ruediger]

If you are suggesting that I should stack the outcome of the 2 sets via Pixel Math or any other tools, I can do that too. Please let me know.

.. ImageIntegration needs a minimum of three input files and therefore you tried HDRComposition.

My suggestion is, that you take all your available 35 registered files (24 with ISO 1600 and 11 with ISO 800) and put all of them again into ImageIntegration with noise evaluation as weighting method. You can work with a less restrictive outlier restriction and might get a slightly better overall result than avaraging the outcome of the two sets with PixelMath.

Rüdiger

[Sedat]

Hi Rüdiger,

I am not very comfortable with PI image integration yet. I can use Images Plus to do the integration of these 2 files.

When I mix the ISO800 images with the ISO1600 ones, what happens to Dark frame matching ? Can I work without Dark frames, is this possible in PI ?

Sorry for too many questions but I am a newbie indeed.

[ruediger]

Hi Sedat,

you integrated the two sets with ImagesPlus and I assume you have different MasterDarks for ISO800 and ISO1600? You can use them in PI also. After calibrating and registering the two ISO sets, you can throw all files together in one ImageIntegration process, even if exposure times are different. The noise weighting will take care of proper integration weights.

If you really want to try PixelMath to combine the two results you already have, you should linear fit them against each other and then try something like "0.6 * set1 + 0.4 * set2". The optimal factors however will depend of a noise analysis of the two images and before I say something wrong of how to compute them, I will leave the computation up to you :-)

When trying to download your files, I only get "Cannot find file!".

I'm really interested to have a look at your files, because I've the same problems like you to smooth out the background (using a spectrum modified Canon EOS 500D). So far I've found no working noise reduction process at linear stage, but after stretching the data both ATWT k-sigma noise thresholding (supported by inverse luminance mask) and ACDNR works good.

Rüdiger

[Sedat]

When trying to download your files, I only get "Cannot find file!".

Oh! Either I am doing something wrong or the system is not up and running yet to produce the correct link. I have tried to download the files myself and got the same response.

The files are under Forum Shared Files, within the folder named "Sedat". You can reach the files at http://endor.uv.es/

I hope you can download the files this time. If not, please try the rapidshare links I gave in my original post.

Cheers,

Sedat

[Sedat]

Any volunteers ? No one ? 

[ruediger]

Any volunteers ? No one ? 

I'll have a look this weekend.
Rüdiger

[ruediger]

I still have some problems with DBE, color calibration and noise reduction.

Hi Sedat,
finally I downloaded the HDR fits and tried on my own. Your unprocessed DSLR data looks as ugly as mine, so I'm sure now, that I make no major mistake in the calibration process

DBE is not necessary with this data. I started with ABE and decreasing function degree from 4 to 2, always inspecting the modelled background. I applied ABE with function level 2 and the outcome looked very natural to me. DBE with automatically placed background points very easily produces wrong looking background models.

My processing steps:
- DynamicCrop (some pixels at each side)
- SampleFormatConversion (from 64Bit float to 32 Bit Integer), only to speed up processing on my notebook and keep swap files small
- ABE
- BackgroundNeutralization (Black preview covering nearly the upper third of the picture without the galaxies)
- Colorcalibration (white preview covering M81 and surrounding)
- Deconvolution (following Manuel's tutorial, but using artificial PSF)
- HistogramTransformation
- Denoising with AtrousWaveletTransformation supported with Luminance Mask:
   - Dyadic 8 Levels Gaussian(7), k-sigma noise thresholding, Threshold 10.0, Amount 0.8, Soft thresholding, Multiresolution Support

I have the same problem like you that most of the other denoising techniques tend to "clump" together noise to larger structures.
Maybe someone else can try noise reduction on this data?

- SCNR
- ColourSaturation. I used the luminance as mask, but applied pixel math "max(0.1,$T)" to it to really protect the background from getting color
- UnsharpMask, same maske like before
- CurvesTransformation to bring down background level somewhat

I applied no HDRWT because I didn't like the look.

There is still residual noise in the picture, but the transition from background towards the galaxy looks smooth. Your leo triplet picture in highres has an artificial smooth background, but your galaxy cores are noisy and oversaturated

Inside the bright parts of the galaxy cores ringing artefacts are visible, that come from deconvolution. I don't know how to avoid them. I see similar rings in the Centaurus A pic presented here, so I'm not alone with this. Any tips?

Hope you like my result somewhat. I tried not too overcook it.

Regards,
Rüdiger

[Sedat]

Hi Rüdiger,

Thank you for taking the time to work on my data. I guess I was expecting too much from this data My flats are not perfect I believe. This results in that terrible looking data. Denoising methods are not helping as you say.

Our friends are all CCD guys I think and they don't want to spend their times with the worries of us, the poor mortals Maybe we should move to CCD cameras

[Alejandro Tombolini]

Hi Sedat,
I have procesed your image and it was very interesting. Let me time until the end of next week and i will post it, (i do not have the data with me).
Saludos, Alejandro.

[ruediger]

Hi Sedat,

what did you expect from the data? There was e.g. not the slightest hint of the flux nebula visible, even when terribly overstretching the image. Has anybody made much better/deeper pictures with cooled CCDs at the same day?
There are many good DSLR astropics out there, M81/M82 e.g. here (not mine): http://www.ried-sternwarte.de/deep-sky.html, showing that optimal weather conditions, dark sites and processing skills are even more important than just spending money for an expensive CCD 

Regards
Rüdiger

[Sedat]

Hi Rüdiger,

Well, maybe it's only me but I am seeing hints of the flux. Please compare my version below , however ugly looking it is, with Harry's image. I am sure you will see what I mean. If I had one tenth of Harry's skill I am sure I would make mine much better.



and this is Harry's

http://www.harrysastroshed.com/Image%20html/m81%20and%20m82.html

Cheers

Sedat