M106. Eternal question: will more data improve it?

[Datalord]

Hey all, I need some input. I tend to grab data from a successful night and process it before the next. If I like the result, I abandon the target and move on. However, I question all the time if that is sane.

I took this Friday night. Does the esteemed inhabitants of this site think I should add a night or two more on it?

171*30s, 59*120s, 12*300s. QHY, RASA and CGX.

 

[Allinthehead]

If you look at the best images routinely they will have gathered 15-30 hours. So in a word, yes.

[ollypenrice]

Before being able to answer your question we'd need to see all the original data. In the processing above, you've inadvertently discarded a lot of it, as the histogram shows:

There should always be a flat line to the left of the rising histogram peak, even if only a very short one. This is your faint signal from the most tenuous nebulosity. The reason for shooting more exposures is to add to that faint data. There is no point in doing so if you are then going to black clip it (discard it) in post processing.

In short your present data may contain more than you think.

Olly

 

 

[Hallingskies]

Depends what you are trying to achieve.  You can never have enough data I guess.  But if I had a result like that, I'd call the job a good 'un...

[Datalord]

23 minutes ago, ollypenrice said:

Before being able to answer your question we'd need to see all the original data. In the processing above, you've inadvertently discarded a lot of it, as the histogram shows:

There should always be a flat line to the left of the rising histogram peak, even if only a very short one. This is your faint signal from the most tenuous nebulosity. The reason for shooting more exposures is to add to that faint data. There is no point in doing so if you are then going to black clip it (discard it) in post processing.

In short your present data may contain more than you think.

Olly

 

 

I guess part of that answer is that I have intentionally clipped the black through a mask to give it a bit more contrast. I'll look out for this in future iterations!

36 minutes ago, Allinthehead said:

If you look at the best images routinely they will have gathered 15-30 hours. So in a word, yes.

Alright, I'll give it another night, process it (in similar way) and post the results here.

[ollypenrice]

8 minutes ago, Datalord said:

I guess part of that answer is that I have intentionally clipped the black through a mask to give it a bit more contrast. I'll look out for this in future iterations!

Alright, I'll give it another night, process it (in similar way) and post the results here.

You can give it more contrast without clipping. The method you used has clearly reduced  the contrast between the faintest signal and the background sky since the faintest signal has gone. I don't like masks because they are too hard to control. Of course, if you can control the blessèd things go for it! I'm no good at it! I prefer to work on contrast by pinning the background sky where it is at a healthy level and working above that. Many ways to skin a cat. 

Olly

[Allinthehead]

4 minutes ago, Datalord said:

Alright, I'll give it another night, process it (in similar way) and post the results here.

I should add that i don't practice what i preach. I gather 3-8 hrs on most targets and then move on knowing i should add at least double the time. Next season my goal is to stay on one target until i'm happy with the data.

[gorann]

Generally a very nice image of the main galaxy but in addition to having black cllipped it as Olly points out you also have dimond shaped stars (no idea how that happened), and star colour is quite extreme (either very blue, very red or white). Maybe these things are just the result of very quick processing, but then it is difficult to say if it would get better with more data. But if you now have about 4 hours, then as Allinthehead says, more would most likely improve it.

[StargeezerTim]

Personally, I'd move on to another target. If you are wanting to produce professional images for publication etc, then by all means get more, but I also think it important to really hone your processing skills to get the best possible results from the data you have. I've just had a two and a half month break due to persistent bad weather and cloud and I couldn't return to targets anyway as they are long gone! 

[ollypenrice]

17 minutes ago, Allinthehead said:

I should add that i don't practice what i preach. I gather 3-8 hrs on most targets and then move on knowing i should add at least double the time. Next season my goal is to stay on one target until i'm happy with the data.

You need to talk to your countryman Mr O'Donoghue! He really nails a picture with an insane amount of data - then doubles it. When I let him. 

Olly

[Datalord]

29 minutes ago, ollypenrice said:

You can give it more contrast without clipping. The method you used has clearly reduced  the contrast between the faintest signal and the background sky since the faintest signal has gone. I don't like masks because they are too hard to control. Of course, if you can control the blessèd things go for it! I'm no good at it! I prefer to work on contrast by pinning the background sky where it is at a healthy level and working above that. Many ways to skin a cat. 

Olly

This one in particular was actually done almost exclusively with StarTools and then a bit of correcting in PS in the end. I usually use PS exclusively with your method, StarTools just have a few very nifty tools for galaxies.

23 minutes ago, StargeezerTim said:

 I've just had a two and a half month break due to persistent bad weather and cloud and I couldn't return to targets anyway as they are long gone! 

This got me thinking more than anything. I think I will stay on this as long as we have good weather and accumulate as much data as our skies permit.

2 hours ago, gorann said:

Generally a very nice image of the main galaxy but in addition to having black cllipped it as Olly points out you also have dimond shaped stars (no idea how that happened), and star colour is quite extreme (either very blue, very red or white). Maybe these things are just the result of very quick processing, but then it is difficult to say if it would get better with more data. But if you now have about 4 hours, then as Allinthehead says, more would most likely improve it.

Yeah, it is the StarTools colouring that does it. It is one of the main reasons I dropped it in the first place. I'll use PS for the subsequent attempts.

You mention the diamond shaped stars. I don't see it in the small stars, so my best guess is that they are coming from the diffraction, which then bleeds into the pixels. Does that sounds reasonable?

[gorann]

Yes, diffraction is a possibility although I did not expect it from a RASA, but maybe it is caused by the cables from the camera. You could look for it in other RASA images, for example on Astrobin. Good luck with further data capture and processing! PS is a very good choise.

PS. If you want to fix those stars in PS, then this method from Olly (described a bit down in this thread) works wonders:

https://stargazerslounge.com/topic/299515-m33/?tab=comments#comment-3277659

[ollypenrice]

Thanks. I must have a look at Star Tools.

Olly

[Oddsocks]

Looking at the high resolution forum image and zooming in to pixel peep at the stars, all the stars seem to suffer some form of chromatic astigmatism with a roughly 45 degree red-blue displacement.

Since this is a colour camera there should be no mis-registration due to filter shift. 

Is this an integration artefact with the subs not perfectly registered, is the RASA not collimated accurately, maybe an example of under sampling or has this crept in during one of the post processing steps?

Extract from near the image centre is attached below that shows the issue though the effect is consistent right across the image.

Just curious....

[Datalord]

2 hours ago, Oddsocks said:

Looking at the high resolution forum image and zooming in to pixel peep at the stars, all the stars seem to suffer some form of chromatic astigmatism with a roughly 45 degree red-blue displacement.

Since this is a colour camera there should be no mis-registration due to filter shift. 

Is this an integration artefact with the subs not perfectly registered, is the RASA not collimated accurately, maybe an example of under sampling or has this crept in during one of the post processing steps?

Extract from near the image centre is attached below that shows the issue though the effect is consistent right across the image.

Just curious....

Ah well, glory to yet another problem.  :-(

StarTools does some wonderful things with colour, which in this case exacerbates the problem.

I got CCDInspector to look at my images and clearly I have a huge problem:

Something is seriously borked. I checked pictures taken with the previous camera and it is the same problem, so the camera is not an issue.

I even went back to images I took with the RASA, but my Canon 6D mounted on it:

Less tilt, but the same pattern. Any ideas what to make of this?

[Oddsocks]

31 minutes ago, Datalord said:

Less tilt, but the same pattern. Any ideas what to make of this?

Wow, 34% tilt with the QHY and 10% tilt with the Canon, both in the same direction, that would suggest something is off with the set up of the RASA, but I know next to nothing about these types of modified Cassegrain OTA's.

I was looking at investing in the Hyperstar / Celestron c11 combo at one time but was put off when reading how finicky they were to get just right.

[Datalord]

1 hour ago, Oddsocks said:

Wow, 34% tilt with the QHY and 10% tilt with the Canon, both in the same direction, that would suggest something is off with the set up of the RASA, but I know next to nothing about these types of modified Cassegrain OTA's.

I was looking at investing in the Hyperstar / Celestron c11 combo at one time but was put off when reading how finicky they were to get just right.

I have a sneaking suspicion it is related to the mirror lock. Going to test that tonight. 

[tomato]

Are you combining LRGB IN StarTools? 

It looks like a scope issue, but I have had some instances where the channels have not aligned 100% in ST. I’m currently experimenting with combining in AstroPixel Processor first, then moving to StarTools, I don’t have PI.

With the UK skies, I am terribly impatient, the best total integration time I have achieved is 4.5 hours, and the quality of my final images clearly reveal this.

When I have a permenant set up and more time, I’ll definitely be after longer integration times.

[Datalord]

42 minutes ago, tomato said:

Are you combining LRGB IN StarTools? 

Nope, the QHY247 is a one shot colour. And I usually do my work in PS, but this one somehow got me better results in ST.

[Datalord]

Good news, it is not the mirror lock! That would have rendered the scope totally unusable. 

Bad news, I have no idea what it is. I'll have to disassemble the entire image train and put it back together to see if I can find a culprit. Any suggestions very much welcomed. 

[Datalord]

I just had a thought. in the two images above, the tilt is virtually the exact same place on the sensor. Those are from pictures taken 8 months apart, with two different cameras and I'm quite sure I didn't set the camera in the same angle at both occasions.

What is going on?!

[Datalord]

What's indeed going on...?! I set the thing going on M106 for the third night, and the real pictures now shows this:

I'm beginning to think CCDInspector is pulling my leg...

[Oddsocks]

8 hours ago, Datalord said:

I'm beginning to think CCDInspector is pulling my leg...

CCD Inspector, like any piece of software, follows the well known software rule, GIGA (garbage-in garbage-out), and so it is important to use it taking account of it's foibles.

First, make sure that in the settings for Default Image Properties that Image Scale is set to 1.29 arc-sec per pixel, Sensor Type = Colour, Saturation Level = ( measure your camera's ADU), Pixel size = 3.91um x3.91um.

Make sure that CCD commander actually uses these figures and doesn't change them to something else when it opens the images, look for a pop-up message when the image(s) is/are opened where CCD commander will state the image scale it is going to use and gives you the opportunity to change it.

(Image scale above was derived from the RASA focal length of 620mm and the Sony IMX 193 spec for pixel size 3.91um x 3.91um using the formulae 205 x (pixel size in um / focal length in mm) )

For CCD inspector the image must not contain large scale galaxies, nebulae or clusters. It should only contain a rich star field, any large scale structures in the image will skew the results.

A single image may be subject to variations in seeing, tracking errors, vibration, etc. To improve reliability of the results at least eight to ten images should be loaded simultaneously, select all images loaded for analysis, check the scores for each frame loaded after measurement and deselect/delete any in the list that show big variations from the norm in FWHM, Aspect Ratio or Background ADU, then run the analysis again using the remaining frames and view the charts.

The colour contour maps are just a visual representation of the data and show a composite of all the measurements taken from the image(s) so the curvature map will show tilt superimposed on curvature, the range of colours shown doesn't help to identify which component, tilt or curvature, is contributing most to the colour range seen in the map, only when the contour map is read in conjunction with the calculated figures does the contour map make any sense.

For a SCT (or Newtonian) telescope, where mirror flop can be an issue, the telescope has to be pointed to three different locations in the sky and several images taken at each location.

First, the telescope should be pointed straight up to a rich star field near the zenith, at least 8-10 images taken and the telescope dithered slightly between each image. Second position is taken with the telescope pointing to a rich star field approx half way between zenith and the celestial equator and with the OTA on the east side of the mount, take 8-10 image here. Final position is with the telescope pointing to the same star field, or as near as possible, but with the OTA on the west side of the mount.

Looking at all three groups of images individually the tilt and curvature measured with the telescope pointing straight up to the zenith shows the system with the mirror unstressed and well balanced, this position will show the underlying basic collimation of the OTA. The image groups taken when the the OTA is pointed at a star field lower in the sky, on either side of the mount, will show how much collimation error is introduced because of mirror flop and stresses induced on the mirror because of poor mirror cell design, implementation or adjustment.

For a refractor, testing the OTA in the three positions can also be useful in showing tilt errors due to poor focuser, flexure in the OTA tube or lens-cell problems.

CCD inspector measurements for tilt and curvature are determined by the shape of stars airy disk. If the system being measured has very small spot sizes, or is a fast optical system, or is measured with a camera using large pixels or pixel binning then the star's airy disk will occupy a relatively few pixels on the sensor, this can lead to very large errors in CCD Inspectors calculations. Quite large variations in star shape may occur across only a few pixels and this is why seemingly wide fluctuations in curvature and tilt may be reported from one single image to the next, measuring a group of images simultaneously and averaging the measurements helps to improve accuracy.

It may also help with fast optical systems to defocus very slightly so that the stars airy disk is spread over more pixels, any defocus should be small enough that an OTA with a central obstruction does not begin to show a central donut in the stars disk, just a slight blurring of the stars profile is enough.

HTH.

 

[Datalord]

Oddsocks, that entire explanation deserves its own separate topic, so it isn't buried deep in a thread about M106. Thank you!

On a very different note, I continued to gather data and now have 12½h of data from the RASA on the target. I did a quick'n'dirty processing just like I did in the first one, although this one didn't really need any additional PS meddling after StarTools. Making the exact colour match was a bit hard, but I think the result is fairly good. You be the judge.

Old, 4h data:

New, 12h data:

EDIT: I should mention that I masked the colouring on the second picture, because I really don't like what ST does to stars when saturating hard. I guess that is blunder from my side, but I couldn't bear seeing the stars become all weird...

[Allinthehead]

The extra detail in the galaxies is great well worth it.