Jump to content

Banner.jpg.b89429c566825f6ab32bcafbada449c9.jpg

A short guide to my solar image processing: from raw data to single pane


Recommended Posts

I have put together a few images which demonstrate in a nutshell how I get from raw input data to the panes I throw into AutoStitch64.

It all starts with the AVI, I typically gather 1000 frames per pane. With the Solar Spectrum filter, even a single frame can show a good deal of detail. Carefully tweaking focus until sharpness is optimal is essential. A frame obtained today is shown below.

1frame.thumb.png.d2cf85acc81b9ba95de0496a3debf1c8.png

Although it looks decent, this is not a good starting point for mosaic stitching. One reason is the fact that there is a distinct uneven illumination (or shading) visible. By taking flats this can be corrected. Various approaches can be used, I tend to defocus, take an AVI of the same length as the ones for the panes, and create a master stack in AS!2 (under image calibration). I often apply a smoothing filter (100 pixel diameter flat averaging often works) to get rid of residuals of sunspots or bright plage. A caveat is that dust bunnies are not well corrected for if you smooth the flat.  Fortunately, the camera shows no sign of them (yet). Loading the master flat into AS!2 ensures all frames are corrected. Not that if you take AVIs of an ROI rather than full frame, you must create flats for that ROI. The result of flat-field correction is shown below.

1frameFlatcorrected.thumb.png.9decc27108b20c54ab7b37856a068ed7.png

This looks more promising. I then stack 100 (typically) of the frames (letting AS!2 pick out the best ones). The result appears to be not much different from the previous, as can be seen below.

SunHalpha_103356_pipp_g3_b3_ap891.thumb.png.2976f1f4ca3d8d8db90cac80e2fc46bf.png

Looks can be deceiving, however. If we apply Lucy-Richardson deconvolution (sigma = 1.15, 50 iterations works for these data) and unsharp masking (sigma = 1.15, strength 2.75) in ImPPG to the stack of 100 images, we get a very decent result.

SunHalpha_103356_pipp_g3_b3_ap891_out_nocurves.thumb.png.d92901ad10dda293b18df105e3bbef3f.png

I frequently work with higher strength values (3.75) in ImPPG, as can be seen in this screenshot

post-5655-0-05366500-1430079913_thumb.pn

The parameter values used depend heavily on the optical system and camera used. I have recently found that the smaller sigma values work best for my current Solar Spectrum filter, whereas in my previous LS35THa and also SolarMax-II 60mm I tended to use much larger values of sigma for unsharp masking, and slightly larger ones for LR deconvolution. The reason for the small sigma value for unsharp masking in the new set-up might be that the contrast for large structures is good enough, and they do not need to be boosted, so I can focus on small detail. Just experimenting in ImPPG is the way to find what suits you best. The image above is the kind that I use for mosaic stitching. My procedure for that is VERY easy: open AutoStich64, click on the "open files" button, select all your panes, and sit back. Any tweaks of contrast after stitching is done in GIMP.

Applying the same settings for sharpening to the flat-field corrected frame yields this:

1frameFlatcorrectedDeconv.thumb.png.6d2961df7f6e3bd4f3ab94d63c91a1ad.png

Result ≠ good

The noise in a single frame just explodes. By stacking 100 frames we can increase the signal-to-noise ratio by a factor of 10 (square root of the number of frames stacked).

I might also want to use a single pane as a detail shot. In that case I like to apply contrast stretching or (partial) invertion of contrast in ImPPG, not GIMP. The reason is that AS!2 outputs 16-bit TIFF files, and ImPPG allows me to work in 16 bit mode with the curves. The results of applying a sigmoidal curve to stretch the contrast in the mid-tones look like this:

SunHalpha_103356_pipp_g3_b3_ap891_out.thumb.png.c3e54372f7e035b46d152dbc3e954c7e.png

The exact degree of contrast stretching is entirely a matter of taste.

One of my favourite contrast tweaks is partial inversion of the contrast. The aim here is to have the region outside the disk in positive contrast (but heavily stretched to bring out proms and spicule detail), and everything on the disk in negative contrast. To achieve this I create a convex curve with a single maximum near the grey level of the spicule layer. The original black and white values are both mapped to zero. I then insert an extra control point to set the apex of the curve at the grey level of the layer of spicules, or a bit above. This point is set to just below white (255) as output value to avoid saturation. The resulting curve is roughly a parabola, which I then correct with two additional control points on either side of the apex. I move these to make the curve on the right a bit concave, and on the left nearly straight. I then tweak until satisfied. The result looks like the one below, and has a 3D feel to it.

SunHalpha_103356_pipp_g3_b3_ap891_out_inv.thumb.png.b2d1a752bfe74167d4fcdb16d12a8031.png

A screenshot of ImPPG shows the curves used:

post-5655-0-29447400-1430079968_thumb.pn

Finally, I might want to turn the image into a pseudo-colour version. To do this I open the image in GIMP, and apply a series of curves for red green and blue to achieve this:

SunHalpha_103356_pipp_g3_b3_ap891_out_col.thumb.png.3ebd1ebbcafa9fef53f4f25616ddf1c2.png

The latter is done with curves that look more-or-less like this

post-5655-0-25974800-1429781330.png

I will add more detail (including screenshots) in due course.

  • Like 32
  • Thanks 5
Link to comment
Share on other sites

About time too Michael :evil: , should be a sticky.

Once I get a few more Quark sessions under my belt I may give this a try as well.

Now looking at a new ccd to give me either FD's or maybe something so I only need 6-8 panes as I do like the hi res images.

Link to comment
Share on other sites

Michael, your timing is impeccable! I downloaded ImPPG just last night and was playing around with it. I gave up in the end as it was getting late but your tutorial will give me a great starting point. Adding screen shots etc would be the icing on the cake so to speak.

Just one question regarding taking flats, I use the defocus method through my PST but I struggle to fill the field of view of the QHY with the solar disk if I take the flat at the same focal length as the lights. I assume I need to fill the field of view or take a flat with the same portion of the solar disk in the field as the light pane?

Many thanks for taking the time to write this up and any guidance regarding flats much appreciated.

Link to comment
Share on other sites

Michael, your timing is impeccable! I downloaded ImPPG just last night and was playing around with it. I gave up in the end as it was getting late but your tutorial will give me a great starting point. Adding screen shots etc would be the icing on the cake so to speak.

Just one question regarding taking flats, I use the defocus method through my PST but I struggle to fill the field of view of the QHY with the solar disk if I take the flat at the same focal length as the lights. I assume I need to fill the field of view or take a flat with the same portion of the solar disk in the field as the light pane?

Many thanks for taking the time to write this up and any guidance regarding flats much appreciated.

In my scope the disk does fill the entire FOV when centred. If this is not the case, you need to use some diffuser (like cling film or a white plastic bag) to create a flat. That does often mean increasing the exposure time.

  • Like 1
Link to comment
Share on other sites

That's some useful information there. Thanks for posting. Would be interesting to see how wavelets compare to deconvolution. I believe you say you get finer detail with deconv. that wavelets wash out?

Thanks Freddie. Registax wavelets gets pretty decent results

file.php?id=13235&mode=view

The very fine detail is better in LR deconvolution, I feel

Link to comment
Share on other sites

  • 3 months later...
  • 6 months later...
  • 3 months later...
  • 4 weeks later...
On 21 June 2016 at 20:21, SiriusDoggy said:

Very helpful for someone like myself that's just getting into Solar astrophotography.

Thanks you very much.

 

On 20 April 2015 at 20:52, michael.h.f.wilkinson said:

I have put together a few images which demonstrate in a nutshell how I get from raw input data to the panes I throw into AutoStitch64.

It all starts with the AVI, I typically gather 1000 frames per pane. With the Solar Spectrum filter, even a single frame can show a good deal of detail. Carefully tweaking focus until sharpness is optimal is essential. A frame obtained today is shown below.

file.php?id=13153&mode=view

Although it looks decent, this is not a good starting point for mosaic stitching. One reason is the fact that there is a distinct uneven illumination (or shading) visible. By taking flats this can be corrected. Various approaches can be used, I tend to defocus, take an AVI of the same length as the ones for the panes, and create a master stack in AS!2 (under image calibration). I often apply a smoothing filter (100 pixel diameter flat averaging often works) to get rid of residuals of sunspots or bright plage. A caveat is that dust bunnies are not well corrected for if you smooth the flat.  Fortunately, the camera shows no sign of them (yet). Loading the master flat into AS!2 ensures all frames are corrected. Not that if you take AVIs of an ROI rather than full frame, you must create flats for that ROI. The result of flat-field correction is shown below.

file.php?id=13154&mode=view

This looks more promising. I then stack 100 (typically) of the frames (letting AS!2 pick out the best ones). The result appears to be not much different from the previous, as can be seen below.

file.php?id=13155&mode=view

Looks can be deceiving, however. If we apply Lucy-Richardson deconvolution (sigma = 1.15, 50 iterations works for these data) and unsharp masking (sigma = 1.15, strength 2.75) in ImPPG to the stack of 100 images, we get a very decent result.

file.php?id=13156&mode=view

I frequently work with higher strength values (3.75) in ImPPG, as can be seen in this screenshot

Schermafbeelding (2).png

The parameter values used depend heavily on the optical system and camera used. I have recently found that the smaller sigma values work best for my current Solar Spectrum filter, whereas in my previous LS35THa and also SolarMax-II 60mm I tended to use much larger values of sigma for unsharp masking, and slightly larger ones for LR deconvolution. The reason for the small sigma value for unsharp masking in the new set-up might be that the contrast for large structures is good enough, and they do not need to be boosted, so I can focus on small detail. Just experimenting in ImPPG is the way to find what suits you best. The image above is the kind that I use for mosaic stitching. My procedure for that is VERY easy: open AutoStich64, click on the "open files" button, select all your panes, and sit back. Any tweaks of contrast after stitching is done in GIMP.

Applying the same settings for sharpening to the flat-field corrected frame yields this:

file.php?id=13168&mode=view

Result ≠ good

The noise in a single frame just explodes. By stacking 100 frames we can increase the signal-to-noise ratio by a factor of 10 (square root of the number of frames stacked).

I might also want to use a single pane as a detail shot. In that case I like to apply contrast stretching or (partial) invertion of contrast in ImPPG, not GIMP. The reason is that AS!2 outputs 16-bit TIFF files, and ImPPG allows me to work in 16 bit mode with the curves. The results of applying a sigmoidal curve to stretch the contrast in the mid-tones look like this:

file.php?id=13157&mode=view

The exact degree of contrast stretching is entirely a matter of taste.

One of my favourite contrast tweaks is partial inversion of the contrast. The aim here is to have the region outside the disk in positive contrast (but heavily stretched to bring out proms and spicule detail), and everything on the disk in negative contrast. To achieve this I create a convex curve with a single maximum near the grey level of the spicule layer. The original black and white values are both mapped to zero. I then insert an extra control point to set the apex of the curve at the grey level of the layer of spicules, or a bit above. This point is set to just below white (255) as output value to avoid saturation. The resulting curve is roughly a parabola, which I then correct with two additional control points on either side of the apex. I move these to make the curve on the right a bit concave, and on the left nearly straight. I then tweak until satisfied. The result looks like the one below, and has a 3D feel to it.

file.php?id=13158&mode=view

A screenshot of ImPPG shows the curves used:

Schermafbeelding (1).png

Finally, I might want to turn the image into a pseudo-colour version. To do this I open the image in GIMP, and apply a series of curves for red green and blue to achieve this:

file.php?id=13159&mode=view

The latter is done with curves that look more-or-less like this

snapshot7.png

I will add more detail (including screenshots) in due course.

Hi Michael how do you deal with different contrast/levels between panes?

Link to comment
Share on other sites

  • 1 month later...
On 7/20/2016 at 15:28, Owmuchonomy said:

 

Hi Michael how do you deal with different contrast/levels between panes?

Actually, I tend to take all panes at the same exposure settings, and process them identically. This means that panes towards the edge of the sun are darker than those in the centre. With good flats, the panes end up pretty consistent. Automatic gain correction in AutoStitch and MS-ICE generally handles the rest well. Sometimes I do have to fiddle around with the brightness levels a bit, but rarely very much. Differences in sharpness due to varying seeing are far more difficult to deal with.

Link to comment
Share on other sites

  • 5 months later...
  • 5 months later...

Michael, I want to thank you for this thread.  I've overlooked it previously, but decided to read it carefully yesterday.  I had never been successful with ImPPG previously, usually ending up with "junk" when trying to set the compound curves and the sharpening settings.  Your simple explanation and the screenshots finally pounded it through my thick skull.

I "played" with it all day yesterday.

Thanks Again!

Clear Skies

  • Like 1
Link to comment
Share on other sites

  • 3 months later...

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • 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.