Gradient removal on Solar disk images

[Photosbykev]

hi folks,

I'm using the full version of PI for my astrophotography work and I'm very pleased with it. I've recently started shooting solar images using a PST solarscope and getting some reasonable results, however the PST solarscope has a sweet/hot spot when imaging with it.

Is there a routine in the full version of PI that could reduce the gradient on a solar disk image caused by the sweet spot on a PST?

An example of the hotspot can be seen in these images http://www.photosbykev.com/wordpress/2011/07/14/pst-solar-imaging/

best regards
Kev Lewis
http://www.photosbykev.com

[Philip de Louraille]

If this spot is due to parts of the Ha filter overheating, then while you should be able to get rid of it (color wise), the region imaged behind will still be recorded at (a) different wavelength(s) than the rest of the solar disk. So yes I am pretty sure the gradient can be removed but the solar granulations/features might still look different than the rest.
Note I am aware I am not telling you how to remove it, (I would have to play with your data in tiff or fit format) and a PI function to correct for that is not coming to mind at this moment as I have not processed this type of picture before.

[Photosbykev]

If this spot is due to parts of the Ha filter overheating, then while you should be able to get rid of it (color wise), the region imaged behind will still be recorded at (a) different wavelength(s) than the rest of the solar disk. So yes I am pretty sure the gradient can be removed but the solar granulations/features might still look different than the rest.
Note I am aware I am not telling you how to remove it, (I would have to play with your data in tiff or fit format) and a PI function to correct for that is not coming to mind at this moment as I have not processed this type of picture before.

hi Philip, thank you for the prompt response. the 'hot' spot is a function of the optics and is unaffected by heating, it is like a vignetting effect.

I do have a Raw CR2 format file online at Mediafire which can be downloaded from here http://www.mediafire.com/?ri8b3uazrokg963 . Normally I would use two files of differing exposures and then split the images to R, G, and B images as all of the data is in the red channel. After enhancing the detail in the disk I paste it over the image with good prominence details. After that it is a simple job to recolour the image.

best regards
Kev Lewis

[Carlos Milovic]

Hi Kev

Do you want to achieve a flat solar disk, or just correct the efect and get a natural gradient? If the first, you may try DBE or ABE, inverting first the image. This way, you may increase the shadows tolerance and allow the samples to be valid across the solar disk. Then, get the model, invert both images, and divide.

[Photosbykev]

Hi Kev

Do you want to achieve a flat solar disk, or just correct the efect and get a natural gradient? If the first, you may try DBE or ABE, inverting first the image. This way, you may increase the shadows tolerance and allow the samples to be valid across the solar disk. Then, get the model, invert both images, and divide.

Ideally a reduction in the hot spot would suffice as the remaining variation might add to the spherical appearance. I will have a go at your suggestion and see where it goes, I did try DBE but I didn't consider inverting the image to get valid samples.

regards
Kev

[Carlos Milovic]

If you are not afraid of doing dirt stuf... I would follow the DBE path, and try to get a flat result. Then, you may add back the natural gradient with a sinthetic one. I think Juan published a PixelMath tutorial somewhere that explained how to create radial gradients. If you are not in a hurry, I may search for old code I had, and write a gradient generator process (with polynomial functions). If you want to try that, let me know and I'll put it on my short to do list (right now I have 2 pending processes with high priority).


Another way to go may be to design a way to create accurate flat field calibration frames through hardware (i.e. shooting some targets)... or, at least, a way to predict the specific correction function that has to be applied, leaving the natural gradient intact. Not easy...

[Photosbykev]

If you are not afraid of doing dirt stuf... I would follow the DBE path, and try to get a flat result. Then, you may add back the natural gradient with a sinthetic one. I think Juan published a PixelMath tutorial somewhere that explained how to create radial gradients. If you are not in a hurry, I may search for old code I had, and write a gradient generator process (with polynomial functions). If you want to try that, let me know and I'll put it on my short to do list (right now I have 2 pending processes with high priority).


Another way to go may be to design a way to create accurate flat field calibration frames through hardware (i.e. shooting some targets)... or, at least, a way to predict the specific correction function that has to be applied, leaving the natural gradient intact. Not easy...

If you have the time, and I can wait, I'd love to see a gradient generator process. It would certainly benefit myself and has the potential to help other PST scope users. If you need reference images I can supply them.

I do intend to try shooting some flats but the PST only allows through a very narrow bandwidth of light and only the sun can produce the light required. Maybe a diffuser in front of the scope will produce a usable flat, something to try out next time the sun appears

best regards
Kev Lewis

[Enzo De Bernardini]

Kev, check the attachment. The icon process will generate a radial gradient of 1000px in diameter. You can adjust all parameters from 'Symbols' field in PM (see description notes, click in the little red button inside process icon) It is a manual work, but it can work. 

I hope this helps.

Greetings,

Enzo.

[zvrastil]

Do you want to achieve a flat solar disk, or just correct the efect and get a natural gradient?

Hi,

if I'm not too much mistaken, there should be no "natural" gradient, because Sun as a Lambertian surface has the constant apparent brightness. Of course, you may want to add gradient to add "spherical" look to the disc, but I'd not call it "natural".

regards, Zbynek

http://en.wikipedia.org/wiki/Lambert%27s_cosine_law

[Carlos Milovic]

The sun is not a lambertian surface. It shows a natural gradient from the limb darkening. I can't remember exactly the phenomena, but since it is a physical property, related to different opacities and temperatures of the sun layers, there may be an analytical function that models that gradient.

[georg.viehoever]

The sun is not a lambertian surface. It shows a natural gradient from the limb darkening. I can't remember exactly the phenomena, but since it is a physical property, related to different opacities and temperatures of the sun layers, there may be an analytical function that models that gradient.

A formula for this can be found in http://en.wikipedia.org/wiki/Limb_darkening.
Georg

[zvrastil]

if I'm not too much mistaken...

So, I was mistaken :-). Thanks for explanation. At least, we have now correct formula for solar gradient...

[Photosbykev]

Thank you for the process Enzo, I'll have a play this evening and see what it produces.

best regards
Kev Lewis

[Photosbykev]

I managed to get the icon process loaded and can generate a circle with gradient but I'm struggling after that. Sorry for any hassle I'm causing

regards
Kev Lewis

[Enzo De Bernardini]

No problem Kev!

You need to create a new image, same size of your solar image, grayscale. To do this: CTRL + N in Windows, or menu Process > Image > NewImage.

Apply the icon process in this new image to create your mask, and apply this as a mask in your solar image (you can drag and drop the mask over the vertical identifier of the target image) Then, you can adjusdt mask properties (from PM process) and see the changes (with 'Show Mask' option enabled, see mask toolbar)

(apologies for the probable errors in my english )

Regards,

Enzo.