Data: How Much

[Hillrat]

If you look at the images the good astro-photographers post, one of the first things you'll see is how many hours on target they had. I've always wondered how many GB of raw data it takes to put together a good image.

It seems a majority of the best images are captured with CCD/CMOS cameras, the remainder being DSLR. How the data is captured and processed into the final product is a different as the look of the camera. This is the limit of what I know about good images. I have a CCD, CMOS and DSLR. I am aware of some of the techniques and software to process a image.

I can't help wonder how much total data, how much lost data (dumped in processing) a very good image takes. I realize site, sky conditions, equipment quality and experience all have to be factored in. I guess the other consideration is if a mono camera is used vs a color camera. The time on target would grow exponentially if using a monochrome camera and 3-6 (or more) filters!

Thought I'd throw this out there, hope I"m not the only one wondering about this.

[s24man]

Hi, that's the age old question. How much data is enough? And you already kind of answered it--the answer is "it depends".

It's all a question of the quality of your system, how many subs you anticipate dumping, your light conditions, how long your subs are, how many filters, etc. etc. etc.
My avatar, the Cocoon, is about 20 hours of data, maybe more. My general (very general) rule of thumb is 10 minute subs for each filter and I usually try to get at least 20 subs of each. Sometimes more. If I'm doing narrowband I'll do 20 minute subs. But each target is different. If I'm doing something with bright stars, say a globular cluster, I'll do no more than 2 minute subs to avoid saturating the well, so it will be far fewer. I still go for 20 subs to maximize the S/N (according to Stark, not much more is gained by going over 20 subs). But also, I'm pretty ruthless at dumping subs I don't like so I always try to go for more rather than fewer. And it obviously takes me several days of data acquisition.

That all written, I know people who have had great results with just 2 hours of data. They're better astrophotographers than I am, though.

[CamelHat]

You are definitely not the only one. Having recently upgraded my mount I am looking into upgrading from a DSLR to an astro camera.

Astrobin lists all the details about equipment and exposure time and they are scary.

I want to optimize my time so instead of doing LRGB I will mount two scopes in parallel one doing mono the other OSC. That would save me about half the time because the RGB is done at the same time as the L for the resolution. Check out astrovetteman's images to see what can be had that way.

I'm afraid that this setup won't even get me a good image of the Veil. For faint targets you still need special filters each of which cost quite a bit and extra time. Alternatively, stars can be killed digitally as well with PixInsight.

Maybe I will start out with targets that don't need that, just LRGB. Check out jthommes' images for that.

Thirdly, I intend to set up at a dark site at 10,000 feet to get a much better signal to noise ratio. To make it worth the trip I need practice. It will be fun but time on target is harder than with imaging from my driveway with the luxury of being at home.

The cameras I am thinking of are the ATIK343L+ and ASI294MC.

I figure the final price of my images will be $1K a piece, for the ones that are really nice. But it will be fun.

[Hillrat]

LOL Henk, a grand per image that's not bad! What are you using for a capture program for the Atik 383L, and in what format are you capturing in.

Stuart, what is the average file size per sub?

The other interesting thing I'm aware of but ignorant about the "How it works" is binning. I have a low end Atik Infinity, I use the Atik capture software (cheating I know) and have seen binning in action, yet don't understand how this adds to the image. It sure speeds light capture in Black and White.

[jaetea]

I approach this question slightly differently. I look at what is the minimum amount of time I need on a particular class of targets to even think about starting the post-processing routine? For me I use a minimum of 30 minutes for star clusters - open and globular. Then the remaining faint fuzzies: nebulae (diffuse and planetary) and galaxies all get at least an hour of data. Sometimes I have to collect 50% more subs than that total just to get to that minimum amount of time. For me, every minute after the minimum is gravy. And we all know how much everyone loves gravy!

Cheers,
JT

[CamelHat]

LOL Henk, a grand per image that's not bad! What are you using for a capture program for the Atik 383L, and in what format are you capturing in.

Stuart, what is the average file size per sub?

The other interesting thing I'm aware of but ignorant about the "How it works" is binning. I have a low end Atik Infinity, I use the Atik capture software (cheating I know) and have seen binning in action, yet don't understand how this adds to the image. It sure speeds light capture in Black and White.

I forgot to tell, the price is what it costs me averaging over the equipment and images not what someone would pay for it (zero).

I want to capture images on a Raspberry Pi 3. ATIK drivers for Linux are available I believe, they advertise them on their site. ASI I would hope, haven't checked yet. I will be looking at INDIlib.org and associated links. I like the Pi because it's just a little box hanging off the USB port of my battery and runs forever.

Format? I presume there's only one.

Obviously this needs some work and convincing. If Linux does not work heck I will drag out my laptop ASCOM ball and chain. Installing on the Pi from source will be non trivial just slugging through it.

[jimminCT]

I'm a PC guy since DOS 3.1, but the thought of that little Pi is just so cool, Henk...
I hope you get it all sorted out and in a nice little project breakout box, velcroed to the side of the mount.
Pictures when you do! (Maybe the forums will allow them normally by then... )

[KathyNS]

My "normal" shoot is 88 minutes of total exposure time in LRGB, totalling 352Mb of raw data. If I am adding Ha, I add another 80 minutes, for 168 minutes, and another 256Mb, for a total of 608Mb. Sometimes I shoot more.

My total time on target is limited by my cold tolerance right now. I am working on a project to complete the automation of my dome, which would allow me to shoot a lot more data. More is always better.

[Hillrat]

My current set-up is capturing 18.9 MB/30 second capture.

[Pauls72]

Binning is rather simple. Most CCD sensors are capable of hardware binning, while very few CMOS sensors are capable of binning.

What binning does is tie groups of photosites (pixels) together electronically, making them act like one large photosite. So when a photon hits any one of the binned photosites it adds to the group charge value. Then when the binned group is read, only one value is read and amplified for the whole group.

2x2 binning
P, P
P, P

3x3 binning
P, P, P
P, P, P
P, P, P

4x4 binning
P, P, P, P
P, P, P, P
P, P, P, P
P, P, P, P

When you bin a sensor you gain sensitivity, you reduce read noise, you increase well depth and you decrease your read time. On the other side you increase your photosite size and you reduce your resolution.
This was real common with older CCD cameras. With newer CMOS cameras they have much lower read and amplifier noise and they read much faster, plus can't be hardware binned, so there is no advantage to binning.
If you bin a OSC (One Shot Color) camera the color data is lost. The group pixels is read as one, so there is no way of knowing what percentage of photons hit the Red, Green or Blue photosites. So we only get a single Luminance value for the whole group.

R, G
G, B


On most objects I go for a minimum of 1.5 to 2 hours of light subs.

This video is worth watching.
https://www.youtube.com/watch?v=IlcaLvKjDd8&t=2722s