boxerman
Senior Member
I got up early (about 2 hours before sunrise) a little while ago to try to capture images of the Eta Aquarid meteor shower. When I went out, it immediately seemed this trial was going to be a bust. There were lots of clouds, some high and a little thin, but more were quite dense—and, it happened, dense just exactly in the direction from which the meteor streaks should originate (Aquarius Constellation). I could see only a very few stars at all, and they were also “in the wrong place.”
What I did was set the camera on burst, and just let it run for about 40 minutes. (I used starry sky focus, which was trivial and flawless.) I assembled the resulting frames into a couple of time-lapse videos, mostly to create an easy way to look for actual meteors. They were of moderate interest, showing movement and change in the clouds. I also set about to play with one still that appeared to show a meteor streak. It isn’t from a meteor. It was a slow-moving object crossing the sky, but my 20 sec. shutter time made it appear to be a “perfect” meteor streak, rather than what it was: a bright speck that lazily traversed the sky. (Well, meteor streaks are seldom consistent; most often, a quick flash that has a build-up and decay in brightness along its length.)
So, here it is. The triangle of bright stars just at the tail of the streak were, essentially, the only ones I could see with my naked eye.
Not a stunning success (C&C are invited), but the postprocessing was interesting. I think this might be an opportunity for people to chip in with their experiences and expertise (I’m not an astro expert, though I’ve been out several times, now).
POSTPROCESSING
Here’s where I started, SOOC. I used Olympus Workspace, for all its faults.
Even this begins to show some interesting phenomena. As mentioned, to my eye, there were almost no stars in the sky at all. But, even SOOC, there were many to see. I have been used to the fact that my eyes are far less good than my camera sensor, but this was extraordinary. On my previous star excursions, perhaps it was true that the camera saw more, but there were always plenty of stars “to my eye.” Especially in my first astro experience, in the Kalahari in Africa. Here, the haze and clouds definitely hindered my eye much more than the camera. And it's not transparently "low-light sensitivity" that's the issue. It seems like low-contrast sensitivity.
So, here’s a sketch of my process (from memory, probably not in order). Readers might chip in with their own experiences, relative to any or all of these points. I’d especially like to hear from relative experts, which I am not.
Overall exposure. I made a mistake and underexposed the image by about a stop compared to what I intended. The mistake came about because I have set up my modes (C1-C4) to maintain their initial settings, even if I make situational changes. So, if I, for example, turn off the camera, things like ISO, even shutter speed, etc. are “reset” to the settings that were in effect when I saved the C-mode originally. (I knew what I was doing when I chose this option. I had not, however, anticipated this particular case.) In fact, I had chosen ISO 400 for this shot, but then that got reset to 200 when I turned the camera off while I set the tripod up.
When adjusting the exposure in post, I noticed that the salience of the streak was surprisingly dependent on exposure compensation. That was quite a surprise. I don’t really have a theory of why it’s so sensitive. I settled on a boost of .1 stop, which was far less than recovering my estimated—but not enacted—setting.
I tried de-haze. Seemed to me a good idea. It didn’t make remarkable difference, but I left it on…full. Since then, I heard an expert astro-photographer say that he always uses de-haze.
Probably the biggest adjustment has to do with “ditching” the background in favor of the stars (or streak). A lot of people use black-level to do that. In this case, I used the tone curve to essentially shift the black level, but also to make the local contrast around the streak maximum. So, I measured pixel brightnesses in the region right around the streak, including the streak, and made the tone curve steep just there.
After the tone curve adjustment, I checked to see that the whole range, both background local to the streak and streak itself, was, in fact, within the steep range of the tone curve. (I’m used this technique to process a number of other images, to bring out detail exactly where I want it. I seldom use selective processing, such as Lightroom or Photoshop allow. That’s just my state of expertise…or, rather, lack of expertise.)
I know that some people use “highlights and shadows” to enhance star images, but both the sky around the streak and the streak itself were in the mid-tone range. So, while I did a little tweaking of highlights and shadows, it didn’t really help much.
I had previously some really good success cleaning out “sky background” (a shooting session on a night with a full moon and high clouds) by selectively adjusting color channels. In particular, on that occasion the sky appeared (both to the eye and on the SOOC) relatively blue, and turning down the blue channel did a remarkable job of clearing out the background to the stars (leaving an orange light-painted rock in the foreground pretty much intact—sometimes you get lucky).
This was not an immediately successful strategy, here. Turning down the blue also shut off a part of the image that I thought was interesting, the blue down toward the low clouds and ground. So, in the end, I only turned down blue, and adjacent “aqua” sliders just a little. Here’s the setting:
What was remarkably more successful was just fiddling with the white balance. I had read somewhere that 4000 degrees Kelvin makes for good astro shooting, and in my previous attempts, this worked very well. But, by this time in the morning (sunrise) and after my initial PP, the dense clouds and the area just above them had turned pretty garish, and I was looking for a way to maintain the interest, but tone-down the garish. Desaturating worked a bit, but not well and also had the side-effect of taking some interest out of other parts of the image. So, I just fiddled the white balance to see what would happen. Remarkably, it worked very well. It nearly completely removed the yellow and aqua tones that were the most problematic for me. What I did was turn up the white balance temperature from 4000, to about 5300, which is close to “sunlight,” which was, indeed, the real source of the light in and just above the dense cloud areas. Obviously, sunlight is not the source of light for astro images, generally.
One last thing. While adjusting, I wondered what noise reduction level I had set on the camera. I just went to the adjustment of noise reduction in Workspace. While there, I just fiddled the noise reduction to see what happened to the image. To my surprise, this was a remarkably effective adjustment of just how many stars appeared in the image. I settled for a "modest" level (some noise reduction) rather than "all the stars possible" (noise reduction off) even though noise was not a particular concern of mine.
So, there you have it. A failure (by eye) turned into a modest success with post-processing. Lots of surprises along the way. More than anything, this was a good learning experience for me. Comments, please.
--
The BoxerMan
What I did was set the camera on burst, and just let it run for about 40 minutes. (I used starry sky focus, which was trivial and flawless.) I assembled the resulting frames into a couple of time-lapse videos, mostly to create an easy way to look for actual meteors. They were of moderate interest, showing movement and change in the clouds. I also set about to play with one still that appeared to show a meteor streak. It isn’t from a meteor. It was a slow-moving object crossing the sky, but my 20 sec. shutter time made it appear to be a “perfect” meteor streak, rather than what it was: a bright speck that lazily traversed the sky. (Well, meteor streaks are seldom consistent; most often, a quick flash that has a build-up and decay in brightness along its length.)
So, here it is. The triangle of bright stars just at the tail of the streak were, essentially, the only ones I could see with my naked eye.
Not a stunning success (C&C are invited), but the postprocessing was interesting. I think this might be an opportunity for people to chip in with their experiences and expertise (I’m not an astro expert, though I’ve been out several times, now).
POSTPROCESSING
Here’s where I started, SOOC. I used Olympus Workspace, for all its faults.
Even this begins to show some interesting phenomena. As mentioned, to my eye, there were almost no stars in the sky at all. But, even SOOC, there were many to see. I have been used to the fact that my eyes are far less good than my camera sensor, but this was extraordinary. On my previous star excursions, perhaps it was true that the camera saw more, but there were always plenty of stars “to my eye.” Especially in my first astro experience, in the Kalahari in Africa. Here, the haze and clouds definitely hindered my eye much more than the camera. And it's not transparently "low-light sensitivity" that's the issue. It seems like low-contrast sensitivity.
So, here’s a sketch of my process (from memory, probably not in order). Readers might chip in with their own experiences, relative to any or all of these points. I’d especially like to hear from relative experts, which I am not.
Overall exposure. I made a mistake and underexposed the image by about a stop compared to what I intended. The mistake came about because I have set up my modes (C1-C4) to maintain their initial settings, even if I make situational changes. So, if I, for example, turn off the camera, things like ISO, even shutter speed, etc. are “reset” to the settings that were in effect when I saved the C-mode originally. (I knew what I was doing when I chose this option. I had not, however, anticipated this particular case.) In fact, I had chosen ISO 400 for this shot, but then that got reset to 200 when I turned the camera off while I set the tripod up.
When adjusting the exposure in post, I noticed that the salience of the streak was surprisingly dependent on exposure compensation. That was quite a surprise. I don’t really have a theory of why it’s so sensitive. I settled on a boost of .1 stop, which was far less than recovering my estimated—but not enacted—setting.
I tried de-haze. Seemed to me a good idea. It didn’t make remarkable difference, but I left it on…full. Since then, I heard an expert astro-photographer say that he always uses de-haze.
Probably the biggest adjustment has to do with “ditching” the background in favor of the stars (or streak). A lot of people use black-level to do that. In this case, I used the tone curve to essentially shift the black level, but also to make the local contrast around the streak maximum. So, I measured pixel brightnesses in the region right around the streak, including the streak, and made the tone curve steep just there.
After the tone curve adjustment, I checked to see that the whole range, both background local to the streak and streak itself, was, in fact, within the steep range of the tone curve. (I’m used this technique to process a number of other images, to bring out detail exactly where I want it. I seldom use selective processing, such as Lightroom or Photoshop allow. That’s just my state of expertise…or, rather, lack of expertise.)
I know that some people use “highlights and shadows” to enhance star images, but both the sky around the streak and the streak itself were in the mid-tone range. So, while I did a little tweaking of highlights and shadows, it didn’t really help much.
I had previously some really good success cleaning out “sky background” (a shooting session on a night with a full moon and high clouds) by selectively adjusting color channels. In particular, on that occasion the sky appeared (both to the eye and on the SOOC) relatively blue, and turning down the blue channel did a remarkable job of clearing out the background to the stars (leaving an orange light-painted rock in the foreground pretty much intact—sometimes you get lucky).
This was not an immediately successful strategy, here. Turning down the blue also shut off a part of the image that I thought was interesting, the blue down toward the low clouds and ground. So, in the end, I only turned down blue, and adjacent “aqua” sliders just a little. Here’s the setting:
What was remarkably more successful was just fiddling with the white balance. I had read somewhere that 4000 degrees Kelvin makes for good astro shooting, and in my previous attempts, this worked very well. But, by this time in the morning (sunrise) and after my initial PP, the dense clouds and the area just above them had turned pretty garish, and I was looking for a way to maintain the interest, but tone-down the garish. Desaturating worked a bit, but not well and also had the side-effect of taking some interest out of other parts of the image. So, I just fiddled the white balance to see what would happen. Remarkably, it worked very well. It nearly completely removed the yellow and aqua tones that were the most problematic for me. What I did was turn up the white balance temperature from 4000, to about 5300, which is close to “sunlight,” which was, indeed, the real source of the light in and just above the dense cloud areas. Obviously, sunlight is not the source of light for astro images, generally.
One last thing. While adjusting, I wondered what noise reduction level I had set on the camera. I just went to the adjustment of noise reduction in Workspace. While there, I just fiddled the noise reduction to see what happened to the image. To my surprise, this was a remarkably effective adjustment of just how many stars appeared in the image. I settled for a "modest" level (some noise reduction) rather than "all the stars possible" (noise reduction off) even though noise was not a particular concern of mine.
So, there you have it. A failure (by eye) turned into a modest success with post-processing. Lots of surprises along the way. More than anything, this was a good learning experience for me. Comments, please.
--
The BoxerMan