Jump to content

Banner.jpg.b89429c566825f6ab32bcafbada449c9.jpg

Improving the attached image of M99


Recommended Posts

Hi folks

I have attached on image of M99 Pinwheel Galaxy. This has been developed using Photoshop CS2. It was stacked in DSS and was basde on 23  300s subs, together with dark, flat and bias frames.

I am slightly disappointed with the result and am asking expert imagers on here to use Photoshop on my stacked image to show what is possible. I have only a few months experience of Photoshop and would welcome feedback that showed there is or is not more in my stacked image than I have so far managed to draw out.

The link to the stacked image is https://www.dropbox.com/s/ljervrh6rl6zap8/M99_23x300s_32bit.tif?dl=0

Thanks for looking

Alec

post-36789-0-89124600-1427637300_thumb.j

Link to comment
Share on other sites

  • Replies 25
  • Created
  • Last Reply

I think it's getting there. The focus looks a bit off to me. The stars look a bit bloated and are oval so it's a bit of bad tracking which if your guiding most likely means the level,alignment and or balance of the mount could be off a bit. I'd start fresh from the start and keep at it! 

Link to comment
Share on other sites

Hi Alec, 

It is a better image than I managed with just a few months experience!

As Leveye said, the tracking is a little off and you will gain some resolution in the image with better alignment / guiding.

The bloated stars are difficult to deal with, partly due to the focussing being just a little off and having to stretch the image hard to reveal the galaxy.

In the attached image from your TIF I created two layers, one pushed hard for the galaxy core and the other gently stretched for the stars, adding some blur, then blended together using the galaxy layer as a background and the stars layer as a foreground, carefully wiping away the weak galaxy core from the foreground star layer to reveal the more prominent galaxy core in the background layer, then flattening the two layers together to make one final image.

The final step was to run an inexpensive multi-function plug-in tool for Photoshop called "Astronomy Tools Actions Set", or more commonly, "Noels' Actions", using a tool from the plug-in called " Increase Star Colour" to restore the star colour to the burnt out stellar cores and then the gradient removal tool.

If I had a bit more time to play I would have scrapped the first attempt and redone the layers, raising the black clipping a little more for the star colour layer as it has revealed itself a little to prominently on the right hand side of the image.

Noels Actions can be found here:  http://www.prodigitalsoftware.com/Astronomy_Tools_For_Full_Version.html

There is a lot of information hidden in your data to play with, so keep at it and don't be disheartened, it is steep learning curve and has taken me years to get this far and still learning something new with each image processed.

 

Link to comment
Share on other sites

Thanks Leveye points taken. Seems so much to get right. In terms of balancing the scope, it never seems very sensitive, Obvious when it it well out of balance, but seems to balance almost too easily because of a lack of sensitivity.

In terms of tracking I am using the Jason Day method of Polar aligning and then doing a 1 Star alignment on Procyon and then a goto to Polaris and aligning this with the AZ/Alt mount adjusters, then a goto back Procyon and aligning. Then back and forth until l both are in the centre of the image.

On the assumption that this puts Polaris in the right spot and Polar Alignment is spot on and drift alignment is not required.

I use PHD for guiding.

Thanks Oddsocks for taking the trouble to have a look at the Tif, much appreciated. The stars on the image you have produced look vastly improved, lots of colour and they look rounder. I will look at the suggested plugin. Thanks for setting out your actions on thee image. I am still trying to get to grips with the use of Layers etc. I seem to dive in with Levels and Curves and don't know where to go from there. Doug German's tutorials are ace, but putting them into practice is another thing!

A few nights ago I attempted more subs on M99 but having got a few more usable ones, the guiding seemed to run into trouble. I was taking 300s subs and it seemed that during every other one. The PHD graph of the DEC axis seemed to fall off the cliff, with the guiding trying to get it back on an even keel. Needless to say, the completed sub had the stars trailing and were therefore deleted. I haven't noticed this happening before. Could it have been the scope out of balance? Ideas would be appreciated.

Thanks again Oddsocks and Leveye, I am spurred on by your feedback. To improve the image would you simply take more subs and add to those already obtained? (Obviously taking more care over tracking, guiding and balance).

Is there a need for drift alignment with my current method of Polar Alignment?

Best regards

Alec

Link to comment
Share on other sites

When you combine images in a stack you are averaging the data so adding great data to not-so-great you end up with a half way house, the balance shifting one way or the other in response to the ratio of good to bad and the total number of subs in the stack.

The best subs will be degraded by adding in poorer ones, the poor ones will be improved by adding good ones, but the end result will never be as good as only combining the very best subs in the stack.

This is one of the advantages of layers in Photoshop, if you have say a dozen great subs and the same number of slightly poorer ones then make a stack of the best and use this as the luminance layer to reveal the best level of detail, use the poorer subs and best subs together to make a stack for the chrominance layer, boost the colour saturation and then blend the layers giving more weight to the luminence layer, the possible combinations are enough to keep you occupied for weeks when it is too cloudy to get out with the telescope.

The problem with the dec guiding is probably backlash, this is common with the EQ5, the effect becomes more noticeable with perfect balance and when the telescope is pointing near the zenith as the gears bounce backwards and forwards in response to guiding commands and shifting balance point.

When imaging away from the zenith you can reduce this by deliberately making the front or rear of the tube heavy so the drive is always pushing against the gears, near the zenith this becomes less effective as the lever of out-of-balace weight becomes shorter.

My solution with my EQ5 is to use a piece of shock cord, thin elastic cord, about 6mm diameter, used for luggage nets, yachting etc, and sold by the mtr in B&Q, tied around one end of the tube, the other tied to one leg of the tripod, or something heavy on the ground if the angles are a bit tight, with the equivalent pull of about 0.5 - 0.75 kg always acting on the DEC axis irrespective of how high the tube is pointing, it's a bit crude but works well as long as I remember to unhook it before before slewing to the other side of the mount!

Link to comment
Share on other sites

Thanks Oddsocks very helpful advice re the use of Photoshop layers - I must get to grips with these.

Your point about backlash is probably spot on as when I first started the imaging that night the scope was pointing SE, by the time that I began getting trouble I scope was looking higher in the sky and S. I will invest in a piece of elastic as you suggest.

With regard to Photoshop plugins. I've had a look at your suggestion and these look to be very helpful additions to sorting out better images. Will be downloading later today and applying them to my DSS stack. Thanks again for your inputs.

Regards

Alec

Link to comment
Share on other sites

Hi Oddsocks

Looking for further advice. Have downloaded Noel's actions and had a quick play around. I can get the action on framing the image and star diffraction spikes working but I'm having problems with some of the others. I ran the Increase Star Colour action, but it seemed to make no difference to the image. As it  ran through the list of actions, the image on the screen remained the same. I ran "As Layer On Top" at the end so I could compare the image before and after - or I thought I was!.

Am I missing something? As we agreed earlier it's a steep learning curve!

Presumably if the action is run with the "Dialogue" toggled off each stage of the action seems to run from start to finish? When I toggle this 'on' the actions can be seen on the screen as they progress. They stop at various stages and require OK to be ticked for the process to continue. Is this so parameters can be changed? Or is it simply for information? The image in the photoshop window stays the same throughout the process - is this right. 

Advice would be very welcome.

Thanks

Alec

Link to comment
Share on other sites

Hi Alec,

When you use the Increase Star Colour tool it works best if you use it incrementally during the whole processing sequence with the dialogue toggled on.

First step, make sure the image is being edited in 16 bit mode, the default output from DSS is 32 bit TIF and you can not use all the PS tools in 32 bit mode with TIF images, once you open the image in PS click "Image" in the top tool bar, "Mode" and choose 16 bit then save the image with a new name, close the current open image without saving, now open the image with the new name, you can always return to the original image again if you don't like the way the process is going (and don't yet know how to use the "History Tool" to get back a few steps when necessary).

An important point is to keep the sky background neutral at all stages during the star colour enhancement otherwise any star colour enhanced will be biased towards the sky background at that moment.

Use the histogram view window, look at the composite RGB channel display and using the levels tool, select each colour channel in turn and adjust black level and mid point sliders so that the left side of each channel coincides and overlaps the others, each time you make an adjustment click the refresh / recycle icon, top right in the histogram window, use the mid point slider to give more or less width to each channel so that looking at the individual and composite histogram view all the channels are the same width, perfectly aligned on the left and as near perfect on the right side as possible, don't black clip the image too early on in the processing sequence, keep the background light until almost the last steps in the processing sequence, the left side of the channel histograms should be at least an eighth to quarter of the way across the window from the left side until the very last final stretch.

As you make deeper and deeper stretches return to individual channel adjustment in levels whenever you notice the background colour moving away from neutral, it is important to get this right early on as during the end stages of stretching and curves just one step adjustment in any channel will move the background colour strongly towards blue, red, or green and you may find it difficult to restore the correct overall colour so little and often works best before you get that far.

Starting from the raw unprocessed image, make a series of small levels stretches in PS rather than just one big one, after each small stretch zoom the image and examine the star core for colour, as the core begins to move towards white after successive stretches then run the Increase Star Colour tool again.

Most of the time the default dialogues that pop-up can be left alone but the important one to look at is the Gaussian Blur window.

When that one appears look in the preview window, zoom out first and grab the image in the window to drag it about until you find the brightest, biggest star visible, then zoom in to maximum magnification, keep dragging the image to stay centred in the window, when the stellar core covers about 10 or so pixels move the radius pixels slider to the right until you see a dark shadow or donut appear in the core, make a note of the pixel size indicated, subtract one pixel count and move the slider to the left until that count is displayed, i.e, if the darkening occurs at 4.4 pixels radius, move the slider to the left until the pixel count says 3.4 pixels.

Now click OK to continue.

Depending how burnt out the core was to begin with then it is unlikely that one integration of the tool will be sufficient, for your source image I ran the tool about ten times in succession, each time you run the tool in a sequence, one after the other, reduce the pixel radius by one pixel until you reach ∼0.3 pixels and leave it there for any subsequent iteration.

Once the star colour is to your liking return to stretching with levels and curves and as soon as you notice the star colours are fading stop at that point and run the tool again, as many times as necessary, until the colour returns.

For a deep space image of a faint fuzzy I would expect to make as many as ten individual levels stretches,  just one curves adjustment and run the Increase Star Colour tool at least four or five times between each stretch.

You can make the star colour deeper before the final stretch as it will become paler in the final image, or just keep on with the same target colour depth in each stretch right up to the end result, you will soon find the method that works best for you.

As long as you use the tool early on before the star cores become burnt out then the final result should be near perfect but if the star cores were max white before you began the tool may introduce colour artefacts into the core, to deal with these in the final image zoom in until you see the individual pixels, examine the stellar cores for bright colour artefacts and use the colour replace tool in PS to sample the correct colour from the stellar airy disk, the boundary between the star and the background sky then brush the colour over the artefacts to removing them and replace them with the correct colour.

Once you get the hang of way the tool works on a single image and are confident then swap to using layers, you can set the output of the Increase Star Colour tool to a "Layer on Top" and then blend the colour layer back in after working on the core in the background layer, I prefer the flexibility of working in multiple layers but to begin with, working on a single flat image is good to learn the capabilities of the various tools available in the actions set.

HTH...

Image of the Gaussian Blur window below...

 

Link to comment
Share on other sites

Hi Oddsocks

Looking for further advice. Have downloaded Noel's actions and had a quick play around. I can get the action on framing the image and star diffraction spikes working but I'm having problems with some of the others. I ran the Increase Star Colour action, but it seemed to make no difference to the image. As it  ran through the list of actions, the image on the screen remained the same. I ran "As Layer On Top" at the end so I could compare the image before and after - or I thought I was!.

Am I missing something? As we agreed earlier it's a steep learning curve!

Presumably if the action is run with the "Dialogue" toggled off each stage of the action seems to run from start to finish? When I toggle this 'on' the actions can be seen on the screen as they progress. They stop at various stages and require OK to be ticked for the process to continue. Is this so parameters can be changed? Or is it simply for information? The image in the photoshop window stays the same throughout the process - is this right. 

Advice would be very welcome.

Thanks

Alec

just a thought, and without knowing anything about Noel's actions, but I've found before that if I'm trying to make changes in somethign in photoshop and nothing seems to be happening, then it's because i'm focussed on a wrong layer, usually a hidden one or one with little/no data on it.  Could that be the case ?

Link to comment
Share on other sites

Oddsocks

Very many thanks for taking time out to provide all that detail. Wow, a lot to do and learn. Before It set out on trying to work through your advice, I have 1 further, probably stupid question, How do I recognise "keep the sky background neutral"?

Regards

Alec

Link to comment
Share on other sites

Matt

Thanks for having a go at my image. I am encouraged to find, as your stretching shows, that all that information is in there somewhere. All it needs is for me to gain experience of manipulating the software - no mean feat from what I have seen so far.

PS. I like the image. I have wondered since I started to develop the image, what the smudge is down the bottom right hand corner? I had assumed it was an artifact rather than some cellestrial body?

Regards

Alec

Link to comment
Share on other sites

There are two methods, by eye, the background sky should look grey when not black clipped, use the brightness adjustment to temporarily raise the image brightness level, there should be no overall colour cast in the background, photographers often keep a neutral grey colour card next to their display to refer to, or a neutral grey reference image downloaded from the web, open on the desktop and then flip between them to see how their processing is going. A reference image is quite useful as it takes into account how your individual monitor is calibrated where a reference card does not, but this is a whole other can of worms.....

The other second method is more accurate but more time consuming.

With the image open on the desktop select the channels view, deselect the RGB "eye" and view each colour channel in turn by selecting only the "eye" symbol for a single colour channel then move the cursor in the image to a set position of the background sky, from the bottom of the PS desktop the current cursor postion is displayed as X:xx Y:xx and lastly the pixel "value".

For the background sky to be neutral the pixel value in each channel should be the approximately the same value at the same set of coordinates.

Then move to the brightest object in the image that is not saturated star or the core of a galaxy, the spiral arms are mostly neutral and can be used, now measure here in the three channels and the pixel counts should be same in each channel.

When the background and foreground counts are similar for each colour chanel then the image is neutral, the background will be grey with no colour cast and the galaxy arms will look grey-white at a large scale.

This is not a general rule though as some galaxys do have strong colours in the arms and disks so it depends on the object, to be super precise you need to use a star map to find a similar star to our sun in the image, type G2V, and measure the average white pixel value in the airy disk for each channel.

Using these numbers as a guide the levels adjustments are made until each colour channel has the same pixel count at the same coordinates in the background and the same numbers in the same coordinates of a G2V star or galaxy arm in the foreground, for example 23/25/24 in the background and 7000/6900/71000 in the foreground would indicate the image had a good colour balance and the sky background was neutral with no colour bias, the actual numbers are dependent on the image and your own preferences and not to be used as an aim, just as an indication that the levels adjustment for each colour channel had been carried out correctly and the final image colours would be as close to true-to-life as possible.

For most of us, by eye, a grey card or grey reference image to refer to while stretching the astro image will produce a sufficiently accurate final result.

If you want to assemble and publish your images on a web site, in the press, or just in an album then at least aim for a common sky background pixel count so that when you flip through your collection on screen or send them off for publishing then there is some consistency of appearance.

For now, just raise the brightness, adjust the individual RGB channels in levels so that the background looks plain grey, go back to brightness and lower it back to where it was and continue with your processing, that should be sufficient, you can get more technical later if you want to...

Link to comment
Share on other sites

Hi folks

I have attached on image of M99 Pinwheel Galaxy. This has been developed using Photoshop CS2. It was stacked in DSS and was basde on 23  300s subs, together with dark, flat and bias frames.

I am slightly disappointed with the result and am asking expert imagers on here to use Photoshop on my stacked image to show what is possible. I have only a few months experience of Photoshop and would welcome feedback that showed there is or is not more in my stacked image than I have so far managed to draw out.

The link to the stacked image is https://www.dropbox.com/s/ljervrh6rl6zap8/M99_23x300s_32bit.tif?dl=0

Thanks for looking

Alec

Very nice image! But don't think M99 is known as the pinwheel galaxy.  I believe M101 is referred to as the northern pinwheel and M83 is the southern pinwheel but I might be mistaken... :)

Link to comment
Share on other sites

Hi Scorpius

Lifted the name off Stellarium which calls it the Pinwheel Nebula and classed it as a Galaxy.

Regards

Alec

seems to be some name confusion out there.  Wikipedia has M101 as the pinwheel galaxy, but if i use my synscan goto for 'pinwheel galaxy' it sends it somewhere entirely differemt, triangulum glxy i think.

edit:

in fact the Wikipediaq see-also's says it all:

See also
  • Messier 74 – a similar face-on spiral galaxy
  • Messier 83 – a similar face-on spiral galaxy that is sometimes called the Southern Pinwheel Galaxy
  • Messier 99 – a similar face-on spiral galaxy
  • Triangulum Galaxy – another galaxy sometimes called the Pinwheel Galaxy
Link to comment
Share on other sites

Hi Oddsocks

Me again. Just been trying to understand the need for a neutral background during processing. I now realise why my assortment of 'final' images have all sorts of coloured backgrounds!

Just getting ready to reprocess the image using your helpful advice and I stumbled on web pages that explained neutral backgrounds and suggested that if the sky background wasn't neutral following DSS then simply load the image into PS and go to Image > Adjustments>Match Color in the Match Color dialogue box click the Neutralize checkbox. Using this it does change the "redness" of the DSS image to a neutral grey.

Is this and acceptable way forward? Does it lose information/data? It seems to easy.

Alec

Link to comment
Share on other sites

Hi Oddsocks

Started to use your procedure for increasing star colour. After setting the pixels in Gaussian Blur (7 pixels in the case I am working on), when it gets to the Gaussian Blur tool for the second, third and fourth time during the run through of 1 iteration of the star colour tool is it set at 7 pixels each time it stops at the Gaussian Blur tool?

Alec 

Link to comment
Share on other sites

Hi Alec,

The match colour, neutralise background works on images that contain just stars and a smallish galaxy but doesn't get it right if the image is a large galaxy or red emission nebula as it interprets the nebula as background, removes the red colour and biases the image towards blue, it will work on your current image but it won't work on all so I prefer the manual method with more control.

In your version of PS the pixel value for each colour channel can be seen in the info screen to the right of the processing area.

How the actions run depends on which version of PS you have, Windows or Mac, and when you download the actions set you would have found versions for CS, CS6 and PS6.

For me using the CS6 version on a Mac I only see the Gaussian Blur window once per iteration the others run in the background without a dialogue screen appearing, for you, in one iteration keep it set to the pixel radius you set in the first screen, just click ok each time it appears and you should be fine.

In your version of PS the pixel value for each colour channel can be seen in the info screen.

 

Link to comment
Share on other sites

post-36789-0-32839800-1427825443_thumb.jThanks again Oddsocks

I take your point on Match Colour.

I have re-run the DSS output using your advice and Noel's Increase Star Colour action. Although not perfect, and not as good as yours, I think it is and improvement on my previous attempt. You seem to have achieved rounder stars. Copy attached.

By way of update. Running the Increase Star Colour on CS2 does bring up Gaussian Blur several times during an iteration and it has to be manually set to the 1st occasion setting as it seems to put a different value in each time it opens up.

Once again many thanks for your help with all of this, certainly moved me along the steep, long learning curve. 

Great site, with great people on here, willing to take time out to help others.

Kind regards

Alec

Link to comment
Share on other sites

No, it was the right image. It just looked better before I posted it. Got some colour, but nothing anywhere near as good as yours! More practice required methinks.

Thanks again Oddsocks

Regards

Alec

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.