Removing a bright centre without a flat?

[bloodhound31]

Hey guys, is there any way I can remove this bright centre while preserving the details of the star field? Screen shot below.

I have Photoshop elements 6 so please keep it simple.

Cheers!

Baz.

[h0ughy (David)]

simple DBE in pixinsight

[naskies (Dave)]

It depends upon whether this is a light pollution gradient (you have extra light that needs to be subtracted) or vignetting (you're missing light in areas that need to be multiplied to compensate). I'm going to assume that it's the latter - vignetting.

Since there isn't any nebulosity in the image, you can just remove the stars from the image and use it as a flat. Here are instructions for PS (I assume they'll work in Elements - but I've never used it before):

1. Open the image in PS.
2. Convert to 16-bit mode
3. Duplicate layer.
3. Gaussian blur the new layer 100 px (depends upon image size - you want to remove all stars and other interesting features, and keep the image gradients smooth).
4. Open Levels and stretch the white point left until it reaches right-most edge of the histogram (53 with your web sized image).
5. Change the blend mode of the layer to Divide.
6. Flatten the two layers.

Your vignetting is extremely severe - there was no detail around the edges in your image, so this won't help with that. You can see that the central region becomes more evenly illuminated (this will work much better with full sized images - convert from RAW directly to a 16-bit image for editing if you can).

Hope this helps.

[bloodhound31]

Originally Posted by naskies

It depends upon whether this is a light pollution gradient (you have extra light that needs to be subtracted) or vignetting (you're missing light in areas that need to be multiplied to compensate). I'm going to assume that it's the latter - vignetting.

Since there isn't any nebulosity in the image, you can just remove the stars from the image and use it as a flat. Here are instructions for PS (I assume they'll work in Elements - but I've never used it before):

1. Open the image in PS.
2. Convert to 16-bit mode
3. Duplicate layer.
3. Gaussian blur the new layer 100 px (depends upon image size - you want to remove all stars and other interesting features, and keep the image gradients smooth).
4. Open Levels and stretch the white point left until it reaches right-most edge of the histogram (53 with your web sized image).
5. Change the blend mode of the layer to Divide.
6. Flatten the two layers.

Your vignetting is extremely severe - there was no detail around the edges in your image, so this won't help with that. You can see that the central region becomes more evenly illuminated (this will work much better with full sized images - convert from RAW directly to a 16-bit image for editing if you can).

Hope this helps.

Awesome and working well until I got to the blend mode bit....

where is that?

[naskies (Dave)]

Doh... looks like PSE doesn't have the "Divide" blending mode - only PS does.

Okay, an alternative way is to do the Gaussian blur step, but then save that blurred (but don't apply Levels) image out as a separate file. You can then use that image as a flat frame in DeepSpaceStacker or whatever stacking program you use.

[bloodhound31]

Originally Posted by naskies

Doh... looks like PSE doesn't have the "Divide" blending mode - only PS does.

Story of my life man.... every time.

[bloodhound31]

Originally Posted by naskies

Okay, an alternative way is to do the Gaussian blur step, but then save that blurred (but don't apply Levels) image out as a separate file. You can then use that image as a flat frame in DeepSpaceStacker or whatever stacking program you use.

Nope, can't do that either. DSS won't stack RAW files with JPEG or TIFF files.

[naskies (Dave)]

Originally Posted by bloodhound31

Nope, can't do that either. DSS won't stack RAW files with JPEG or TIFF files.

Maybe not JPEG (avoid using JPEGs for any kind of stacking - it ruins the fine detail that stacking relies on), but it definitely will with TIFF - master calibration frames in DSS are written out as TIFF by default.

Just load in the RAW as a light frame, and your artificial flat frame as a 16-bit TIFF.

[bloodhound31]

OK I'll try that. Thanks Dave.

[multiweb (Marc)]

Originally Posted by naskies

It depends upon whether this is a light pollution gradient (you have extra light that needs to be subtracted) or vignetting (you're missing light in areas that need to be multiplied to compensate). I'm going to assume that it's the latter - vignetting.

Since there isn't any nebulosity in the image, you can just remove the stars from the image and use it as a flat. Here are instructions for PS (I assume they'll work in Elements - but I've never used it before):

1. Open the image in PS.
2. Convert to 16-bit mode
3. Duplicate layer.
3. Gaussian blur the new layer 100 px (depends upon image size - you want to remove all stars and other interesting features, and keep the image gradients smooth).
4. Open Levels and stretch the white point left until it reaches right-most edge of the histogram (53 with your web sized image).
5. Change the blend mode of the layer to Divide.
6. Flatten the two layers.

Your vignetting is extremely severe - there was no detail around the edges in your image, so this won't help with that. You can see that the central region becomes more evenly illuminated (this will work much better with full sized images - convert from RAW directly to a 16-bit image for editing if you can).

Hope this helps.

Very cool tip. Gotta try it on my color shots from last session.

[bloodhound31]

It's a lost cause. I'm going to throw the data away and shoot it again with flats.

[Octane (Humayun)]

DPP, Peripheral Illumination Correction.

H

[Octane (Humayun)]

http://www.iceinspace.com.au/forum/s...4&postcount=14
http://www.iceinspace.com.au/forum/s...5&postcount=15
http://www.iceinspace.com.au/forum/s...6&postcount=16

H

[Octane (Humayun)]

If your images weren't taken with a Canon lens, ignore my posts above.

H

[bloodhound31]

Originally Posted by Octane

DPP, Peripheral Illumination Correction.

H

Thanks H. You da man. Fixed.