Jared Willson:
Jared Willson:
Dave Rust:
I'm in Bortle 5 in Indiana. My EDGE-HD 925 is 2350mm at f10 and shoots way-oversampled to an ASI2600, Seeing is usually 3.5 arc sec.
"Oversampled" doesn't seem to matter. Plus, I can always reduce to BIN2 in post if no detail will be lost. But most of my BIN1 images have just a tad more detail than a reduction would have, so I generally publish the full size result.
I shoot on a color camera and don't use a reducer. Dark targets are difficult to catch at such magnification, but galaxies and Ha nebulae turn out terrific, if one is patient (lots of subs).
I have tons of fun and, in the end, images show few compromises.
Go go forth and image without worry.
What gain do you shoot at? I just picked up 9.25 Edge and plan on using my ASI2600MC on it, but I'm wondering about exposure settings. I haven't had a chance to use it yet because I mistakenly bought a stellar mate pro to control the rig, and it's turned into a nightmare trying to get it to work, so I've switched to a Mele PC and Pegasus micro power box and now hoping for clear nights. lol
I'm not Scott, but I'll take a stab at answering here since I do own and image with an ASI2600 (mostly for EAA)...
The obvious decision point is whether you shoot at gain 0 or gain 100. Technically, gain 0 gives you your greatest dynamic range, but depending on how dark your skies are you might need fairly long sub exposures to swamp read noise. This is especially true for narrowband data. If you go by one of the rules of thumb--somewhere between 5x (rn)^2 and 10x (rn)^2--and have an f/7 to f/10 scope, then you could be looking at subs under dark skies needing fifteen minutes or longer. Under light polluted skies shot noise swamps read noise much faster, so five minute subs might be ample.
The second option is to shoot at gain 100. This is where dual-gain mode kicks in, so read noise drops precipitously. You get almost the same dynamic range (since both read noise and full well capacity drop in a similar manner), so you can shorten the sub exposures and get substantially the same result in a given amount of integration time. Basically, the noise floor drops, a good thing, but stars start to saturate faster, a bad thing. They offset. I believe most people end up using gain 100 with this camera especially with slower optical systems such as those you are likely to use for galaxies. With a moderate focal ratio scope and reasonably dark skies gain 100 will leave you with moderate sub exposure lengths. You won't have so many subs that data storage and processing requirements are huge, or a very small number of very long subs where you might not have enough data to effectively address satellites, clouds, airplanes, guide errors, etc.
I don't see a reason for using any other gains. Higher than 100 and the full well capacity drops faster than the read noise, so no real benefit. Between 0 and 100 and the same thing happens--you lose full well capacity without actually gaining much read noise reduction. That's why gain 0 and gain 100 have the two highest dynamic ranges. With a 9.25 Edge, even with the usual reducer, gain 100 is almost certainly the best choice.
- Jared
Thank you for the explanation. So, I will sound like an idiot here, but I still struggle to figure out my sub-length. I'll give a couple of examples. Most of my shooting is a Bortle 5 sky. If I use my 4" refractor at f7, I have to shoot ten-minute subs to get the histogram away from the far left with NB filters(I found gain 300 to work well, but now I'm second-guessing that gain). Using my Sharpstar sca260 at f5 on an ASI294mm (gain 120), I still find I'm shooting ten-minute subs with NB filters to get the histogram away from the left. Does that make sense? For the most part I look at the histogram and fiddle until it seems "ok."
Looking for some separation between the extreme black point in your data and the left edge of the histogram is not a bad first approximation on effective sub exposure duration, but it's not optimal, mostly because it's not quantifiable and not precise. There are better approaches, but even the better approaches won't give you a one-size-fits-all result that will work for every subject. For example, I have tried to "standardize" on sub exposure lengths where my sky background comes out at 10x read noise squared. This is a nice safe value for drawing out the faintest details in one's data, but it would be a very poor choice if I were shooting the Double Cluster since I'd end up with several thousand saturated stars that have lost their color! So, here is what I would recommend...
For objects like most star clusters, anything where the stars are really the "star" of the show--keep the exposures short enough that you have, at most, a few hundred saturated pixels. These are likely to be very pretty short exposures, and your black point may still be piled up along the left edge of the histogram. Your acquisition software hopefully tells you how many stars are at max value. I use NINA, and it reports this for every exposure. Even this rule of thumb, though, won't apply to all star clusters. The Pleiades, for example, are going to saturate almost no matter what you do. If you want to pull out any nebulosity at all, you're going to have thousands of clipped pixels. No getting around it, at least not without combining a range of exposure lengths, and that is beyond the scope of my general advice.
For most deep sky astrophotography, though, the "stars" of the show are really the faint background objects, not the stars. Dust, gas, galaxies, etc. For these objects, I would lean towards 10x read noise squared as a good spot to land. Some would give a lower value like 5x read noise squared. Whatever the value, the goal is to strike a balance between faint detail and star saturation.
So, how do you measure 10x read noise squared? It's not too hard, but it does require a bit of effort. Pick a dark night--no moon up and few clouds/little haze. Point your scope at a section of sky that isn't filled with nebulosity (stars won't really matter for this.) Pick an altitude that is typical for most of your imaging, maybe 50 degrees or so--most of us rarely image much below 30 degrees, and few objects get above 70 from any given location, so 50 seems like a decent compromise. Take a bunch of images at different exposure lengths, perhaps from one minute to fifteen minutes in one minute increments. If you already know you are going to be somewhere in the 2m to 10m range for broadband and 5m to 15m range for narrowband you can just cover those ranges. Also, make sure you have a bias frame available or a dark frame--any bias/any dark will do, but they need to be taken at the same offset, gain, and binning as your lights.
Now, go to your camera manufacturer's website. Look up the gain in e- per ADU for the settings you were using on your camera. Also look up the read noise in e- for the settings you were using on your camera.
Look at the median ADU (brightness in analog "counts" where 65,535 is "white" and zero is "black" for a 16 bit camera, and even most 12 and 14 bit cameras will convert to a 16 bit scale) for one of your lights. We look at median rather than mean so that stars won't skew the result. Let's say your software reports a median brightness of 500 ADU. Now do the same thing for your dark/bias frame. Let's say the software reports a median brightness of 350 ADU. Cool, so your background level (the median reports essentially background level since so much of an astronomy image is background) is 150 ADU above your bias/offset. Now, using the manufacturer's data, convert that to e-. For the ASI 2600 at gain 100, it looks like about 0.24e-/ADU, so that count of 150 ADU above bias is actually about 36 electrons. The read noise, per ASI, is about 1.4 electrons at gain 100. So, 1.4 squared is 2. Our 36 electrons above offset is about 18x rn^2, probably overkill. We are clipping more stars than we need to for my example exposure. We could use shorter exposures, theoretically one that is a little over half as long. Mind you, there is nothing "magic" about the 10x read noise squared value. It's just a good point to choose for faint detail since read noise is guaranteed to be much, much less than shot noise.
You would repeat the above process for each filter and come up with a different exposure length for each filter. This could then be your standard value for that filter on a dark night from that location. If you want to be more precise, you can measure the actual gain and read noise for your camera, but I have found the manufacturers' specifications to be close enough that I don't see any particular value in that, at least for this purpose. Variations in background brightness from night to night and from object altitude changes are going to be much larger than any slight variation in gain/noise from one sample camera to the next. You might find there is enough difference from one filter to the next that you can't really standardize on one exposure length. Not only will narrowband and continuum filters be different, but even RG&B might be different enough to matter somewhat. This is just because skyglow is not "white" in color, and camera sensitivity varies with wavelength. Totally normal. If the values are close, pick one since it will simplify your dark library. But my red and blue values, for example, vary by almost 2x. That's enough that I use different sub exposure lengths for red and blue, not just for RGB and Ha.
The problem, of course, is if you don't image from the same location all the time. The "best" sub length from my home is much, much shorter than from my dark sky location just because the light pollution is so much worse at home. Where I need three minute exposures from my dark sky location (even with an f/3.8 scope), optimum from home would be somewhere closer to 10s at gain 100. So what do you do if your conditions vary? I suppose I would recommend doing the above exercise one time, and seeing just how much separation you are getting from the left edge of the histogram. That will give you a general visual guideline. That will probably be enough for you to use anywhere on any night. If you are getting more separation you are probably clipping more stars than you need to. Less separation than that reference, and you may not be getting as deep as you might.
If you don't want to do the above work (though I would recommend doing it at least once just so you understand what's going on), then lots of software tools will do it for you. I know Sharpcap, for example, can do these calculations on the fly if you know your read noise and gain. I've run it before and it's pretty straightforward. You may need the "pro" version of sharp cap--I can't remember for sure. It's pretty inexpensive, though.
Last piece of advise--don't stress about this too much. Integration time is vastly more important than sub exposure duration. Finding dark skies is vastly more important than sub exposure duration. Most astrophotographers really like having a rubric to follow to get best results, but you quickly get to the point of diminishing returns, and since the "optimum" value will often vary depending on what part of the subject you are trying to emphasize in your image (stars, spiral arms, faint dust), it's not like there is an absolute, categorical correct answer.
Sorry this post was so long.
Your post is perfect. I've mostly "winged" it when it came to exposure and gain, despite the fact I've been doing this off and on for a few years now. I'm going to print your post and put it in my observatory and it will be something I do because I really do need to understand it better than I do.
|
11.99
0.00
6.05
0.90
6.45
0.90
1.51
