Is there any way to brighten (enlarge) star images without flattening differences in magnitude?

[Dark Penguin]

My astrophotography has been fairly hit or miss so far. I have taken shots from parking lot of my apartment complex, located in a typically light polluted residential neighborhood, and successfully post-processed the images to the point where they resemble the night sky in a dark sky location. In the best examples, the stars show their proper colors, e.g. red for Betelgeuse, orange for Aldeberan, etc.

On the other hand, I have taken other photos from under a dark sky, and paradoxically encountered far greater problems in post processing. A recurrent problem is that the star images usually come out too small. Even in an unedited image, what appears at first to be nothing but washed out sky turns out reveal numerous stars, some as faint as 7th magnitude, Darkening the background sky as much as possible certainly helps a great deal, but the star images themselves are still too small.

I'm most familiar with GIMP though I do also have access to Lightroom and Photoshop. In GIMP, as I suppose is true of similar tools, you can "grow" a selected area by a specified number of pixels. If I select all of the star images in the photo, the grow tool will extend the radius of each star image by that many pixels, but a huge drawback here is that the magnitude range is flattened. For example, in an image of Jupiter the central bright area might be nine pixels in diameter, while the corresponding area for Ganymede is just a single pixel. If I grow all the images by two pixels, then Jupiter becomes only about 1.5 times brighter than it was before, perceived brightness being based on the area of the planet's image as it appears in the photo. Ganymede, on the other hand, now becomes more than 10 times as bright times as bright as was in the unedited photo. You wouldn't think it matters that much since the overall image still accurately shows that Jupiter is brighter than Ganymede, or that Sirius is brighter than Wezen. But it still gives the photo an artificial and faked aspect. What's worse is that familiar constellations now become difficult to identify since the differences in the brightness of stars is important in telling them apart. Orion becomes lost in the background when the three stars in the belt appear almost as bright as Rigel and Betelgeuse, and the shield stars are almost as bright as the belt. Still another problem is that star colors, which tend to show up around the edges of a stellar image, go completely missing, making the stars all pure white.

I have a Nikon D5600 and for astrophotography I use the 18-55mm lens that came with it.

 

[Astrozoid]

You can use a fog filter on the lens when you shoot the images.

Jerry

 

[Dark Penguin]

You can use a fog filter on the lens when you shoot the images.

Jerry

But wouldn't that also tend to filter out the fainter stars?

Since posting this topic I have found a third party action set for Photoshop that solves much of the problem, besides making it much easier to process astrophotos generally. (Thanks, Astronomy Tools!) You can also add diffraction spikes and ring flares for that classic astrophoto look.

Here's an unaltered photo of some stars mostly in Boötes and Virgo:




Unedited photo of night sky



And the altered photo:




Edited photo of night sky

The one drawback to enhanced star images is that it may reduce or eliminate the ability to discern fine detail, the trade-off being a much more appealing overall image.

 

[sharkmelley]

You can use a fog filter on the lens when you shoot the images.

Jerry

But wouldn't that also tend to filter out the fainter stars?

Since posting this topic I have found a third party action set for Photoshop that solves much of the problem, besides making it much easier to process astrophotos generally. (Thanks, Astronomy Tools!) You can also add diffraction spikes and ring flares for that classic astrophoto look.

Here's an unaltered photo of some stars mostly in Boötes and Virgo:


Unedited photo of night sky

And the altered photo:


Edited photo of night sky

The one drawback to enhanced star images is that it may reduce or eliminate the ability to discern fine detail, the trade-off being a much more appealing overall image.

It looks like a useful technique for emphasising constellations etc. Useful when the stars themselves are the subject of the image.

Mark

--
Takahashi Epsilon 180ED
H-alpha modified Sony A7S

Mark Shelley Astrophotography

 

[Lyle Aldridge]

Have you tried using the inverted image itself as a layer mask? GIMP and Photoshop both have this capability.

I don't do much wide-field imaging of entire constellations, but I've found this usually works, with varying levels of success, to control stars that otherwise bloom and brighten too much when stretching images like Alnitak and the Horsehead Nebula. In the inverted-image mask, the brightest stars are completely black, and fainter stars merely gray. So stretching the main image reveals no change in the bright stars, and brightens the dimmer stars by amounts that correspond to how light their gray masking is.

This technique seems to work slightly better for me in GIMP, where the layer mask dialog includes "Grayscale copy of layer" as an option when adding the mask. There is a check-box on the layer mask dialog to invert the image. In Photoshop, I add the layer mask first, then use "Image>Apply Image" while the layer mask is active. The Apply Image dialog includes the option to invert with a check-box. In either program, I often find I need to increase the contrast of the mask to confine effects to the dimmer parts of the image, and apply a Gaussian blur to keep the effect from showing too sharply.

 

[Dark Penguin]

Have you tried using the inverted image itself as a layer mask? GIMP and Photoshop both have this capability.

This sounds like it would be ideal, but I'm not sure now to do it. The stars themselves have subtle differences in color so I'm not sure how I could select them by color; the same goes for the background sky because, at the ISO levels required, it's as noisy as bubble wrap and has numerous tiny color blotches.

is there another way that I could select the stars than by color, for example by general brightness? Because that's the one thing that would reliably differentiate the stars from the rest of the image.

 

[Lyle Aldridge]

Have you tried using the inverted image itself as a layer mask? GIMP and Photoshop both have this capability.

This sounds like it would be ideal, but I'm not sure now to do it. The stars themselves have subtle differences in color so I'm not sure how I could select them by color; the same goes for the background sky because, at the ISO levels required, it's as noisy as bubble wrap and has numerous tiny color blotches.

is there another way that I could select the stars than by color, for example by general brightness? Because that's the one thing that would reliably differentiate the stars from the rest of the image.

The method I described doesn't use any selection tools. Instead, the "selection" is done by the layer mask, which contains an exact positive or negative monochrome copy of the main image. Where the pixels of the layer mask are white, full effect of an adjustment will show, and where the pixels of the layer mask are black, the adjustment will have no effect. Where the pixels of the mask are middle tones of gray, the adjustment will show more where they're light gray and less where they're dark gray.

As an example, I quickly threw together a couple of screen shots from Photoshop, processing an image of a couple of well-known galaxies:



Notice the Hue-Saturation adjustment layer "circled" in both shots. The thumbnail on that layer shows that a positive monochrome copy of the main image has been applied as the layer mask of the adjustment layer. In the top screenshot, the adjustment layer is inactive (no eyeball) and in the bottom shot, it's active. In that bottom shot, the galaxies and stars show color from the saturation adjustment, because they contain the brightest pixels on the layer mask. The dimmer stars show less color because they are not as bright on the layer mask, and the black sky shows no effect from the saturation adjustment at all.

 

[Dark Penguin]

Have you tried using the inverted image itself as a layer mask? GIMP and Photoshop both have this capability.

This sounds like it would be ideal, but I'm not sure now to do it. The stars themselves have subtle differences in color so I'm not sure how I could select them by color; the same goes for the background sky because, at the ISO levels required, it's as noisy as bubble wrap and has numerous tiny color blotches.

is there another way that I could select the stars than by color, for example by general brightness? Because that's the one thing that would reliably differentiate the stars from the rest of the image.

The method I described doesn't use any selection tools. Instead, the "selection" is done by the layer mask, which contains an exact positive or negative monochrome copy of the main image. Where the pixels of the layer mask are white, full effect of an adjustment will show, and where the pixels of the layer mask are black, the adjustment will have no effect. Where the pixels of the mask are middle tones of gray, the adjustment will show more where they're light gray and less where they're dark gray.

As an example, I quickly threw together a couple of screen shots from Photoshop, processing an image of a couple of well-known galaxies:



Notice the Hue-Saturation adjustment layer "circled" in both shots. The thumbnail on that layer shows that a positive monochrome copy of the main image has been applied as the layer mask of the adjustment layer. In the top screenshot, the adjustment layer is inactive (no eyeball) and in the bottom shot, it's active. In that bottom shot, the galaxies and stars show color from the saturation adjustment, because they contain the brightest pixels on the layer mask. The dimmer stars show less color because they are not as bright on the layer mask, and the black sky shows no effect from the saturation adjustment at all.

I can certainly see the difference. I'm going to have to learn more about how layers work. In your example, if a pixel isn't pure white nor grey but tinged with some other color like yellow or read, will this technique still work?

 

[Lyle Aldridge]

Have you tried using the inverted image itself as a layer mask? GIMP and Photoshop both have this capability.

This sounds like it would be ideal, but I'm not sure now to do it. The stars themselves have subtle differences in color so I'm not sure how I could select them by color; the same goes for the background sky because, at the ISO levels required, it's as noisy as bubble wrap and has numerous tiny color blotches.

is there another way that I could select the stars than by color, for example by general brightness? Because that's the one thing that would reliably differentiate the stars from the rest of the image.

The method I described doesn't use any selection tools. Instead, the "selection" is done by the layer mask, which contains an exact positive or negative monochrome copy of the main image. Where the pixels of the layer mask are white, full effect of an adjustment will show, and where the pixels of the layer mask are black, the adjustment will have no effect. Where the pixels of the mask are middle tones of gray, the adjustment will show more where they're light gray and less where they're dark gray.

As an example, I quickly threw together a couple of screen shots from Photoshop, processing an image of a couple of well-known galaxies:



Notice the Hue-Saturation adjustment layer "circled" in both shots. The thumbnail on that layer shows that a positive monochrome copy of the main image has been applied as the layer mask of the adjustment layer. In the top screenshot, the adjustment layer is inactive (no eyeball) and in the bottom shot, it's active. In that bottom shot, the galaxies and stars show color from the saturation adjustment, because they contain the brightest pixels on the layer mask. The dimmer stars show less color because they are not as bright on the layer mask, and the black sky shows no effect from the saturation adjustment at all.

I can certainly see the difference. I'm going to have to learn more about how layers work. In your example, if a pixel isn't pure white nor grey but tinged with some other color like yellow or read, will this technique still work?

You need to distinguish between the image and the mask. The image can have lots of color. But no matter how much color is in the image, or what colors they are, the image copied to the layer mask will always be grayscale - just black, white or some gray in between. So, if the stars in your image are bright with yellow hue, for example, the corresponding pixels of the layer mask will be a light gray. Or if you invert the image on the mask, they'll be a very dark gray. (And if you want, you can do a curves adjustment of the mask to make them black.)

So an inverted image on the mask can be used to bring out the dimmer stars while minimizing "bloom" of the brighter stars. A positive image on the layer mask can be used - as in my example - to increase color saturation of the brighter parts of the image without affecting the dark background sky.