Mono versus OSC

[AlistairHowie]

OK, I just can't decide between a mono or OSC Atik 314l+ for my first foray into AP.

I know the mono wins for flexibility and sensitivity, and to an extent is a future proof purchase since it offers the potential for a move into NB imaging. So no need to revist those arguments.

What I am trying to understand is the additional "work" involved, if any, in capturing mono images (processing less of a concern). With a 314L and a FLT 98 with focal length of 618mm, will it be possible/advisable to bin 2x2 the RGB images to save time, or will that just wreck the final image? Would I need more pixels for binning to make sense with my set up, and if so, how many?

Also, what's the situation with flats? Do they need to be taken for each LRGB "channel" as well (I guess so as there could be dust on the filters?), and could they be binned as well? Lastly, what length exposure do the flats need to be (I've read Every Photon but can't see it mentioned) and why do such a large number,(Steve suggest 20 ish) need to be taken? If vignetting isn't an issue for a particular set up, why wouldn't a single flat adequately show up dust?

So many questions, thanks for your patience!

[swag72]

Often what folks do with LRGB images is 2x2 bin the RGB and 1x1 the Luminance. The way I look at it is that the luminance is the detail of the image and the RGB is more of a colour wash that is added to the detail - I hope that makes sense, that's how I see it anyway!

Some folks will be along soon to say with your set up what would be OK for binning purposes, as sometimes depending on pixel scale it's just not worth doing. I don't really get the figures and stuff on that.

Flats do need to be taken for each filter and yes the recommended amount is about 20. I think if you did take 20 and stacked them you'd probably be quite surprised by the vignetting and possibly dust bunnies that would be there. These will all be there in a stacked image, so flats take that away. Don't forget, as you stack each image, even if there's a really small vignette that you can't see, it will get multiplied up in the stacking process to sometimes quite awful levels.

For exposure length, that depends on your camera, but for the 314L as it has no shutter you can use short exposures. You basically take an exposure of a constant light source (you may have read about the sky at twilght, or the sky with a Tshirt over the scope - Many people have light panels that they put in front of the scope, I used to use my laptop screen). You then take an image and exposure it for the length of time that gets you about 20-30K ADU in your image. For example, for my camera that is as short as about 0.3s. You will need to check on whatever software you use (I use Maxim) for the ADU value. The 314L will go to a maximum of 65500 ADU, this is a fully saturated pixel. You are aiming for pixels that are between 1/3 and 1/2 saturated.

Hope that helps and I'm sure someone will be along soon to give a far better and probably accurate explanation!

BTW, flats really are a really important calibration frame. They make life so much easier and they always seem a mystery, giving people bother. They are not as scary as you imagine!!

[AlistairHowie]

That's helpful, thank you. 80 flats sounds ominous, but not if they last less than a second or so. Little to no difference from taking 1 flat once set up, I guess.

I dont know what ADU is, but I'm guessing its well depth? How do you figure out how long to expose for to achieve the right pixel fill? Does Artemis or similar display fill levels? Sorry if that's a dumb question.

And I'll wait to see what others say about binning!

Many thanks again.

[swag72]

I use Maxim for determing the ADU figure. I take a picture in Artemis and then open it in Maxim and the info button as you move it around the screen will give you an ADU figure - Easy peasy! Sorry don't know what ADU stands for, to me I just know I need to get between 20-30K of them!!

I think that at the bottom right of the Artemis screen there's a figure that is the ADU figure. So you'd take an image in Artemis and then run the cursor over the image and the figure at the bottom will tell you what that pixel figure is.

80 flats? ............... You don't need that many!! I use 30 - once they're set up to take, they're very quick and easy.

[Uranium235]

Youre better off getting the mono version for a very good reason. The work you produce will be of a higher resolution with mono becuase there is no bayer matrix on the chip. To produce a colour pixel on a colour camera in one shot requires 4 pixels (RGGB), with a mono camera its just one becuase you are shooting the colour channels separately.

Personally, I'd love 3 Atiks - each doing a channel at the same time but that would work out to be a wee bit expensive!

Binning at 600mm(ish) might be ok on large targets, but to be honest youre better off doing it unbinned as you can stretch and NR the RGB layer to death. Thats becuase youre only after the colour data when blending with the luminence (which carries the contrast, depth and detail).

PS: If you want to save time, then you can just use bicolour and synth the green, example:

R/B/B

or

Ha/OIII/OIII (works very well)..... which reminds me. If you get an OSC camera, narrowband imaging will be inefficient to the point of not worth bothering with.

[ollypenrice]

I did a shoot out article for Astronomy Now and use the OSC and mono versions of the Atik 4000 routinely.

You cannot bin OSC - or if you do you lose the colour information.

The loss of resolution is greatly exaggerated. You just need to take more exposures in the OSC.

Flats are the glory of OSC. You take one set instead of four. You have one additional task which is to debayer the flats but the time gain is significant.

In general preprocessing is a bit quicker but post processing is the same. One thing you don't have to worry about is aligning the colour channels. On the other hand you are likely to have a weaker luminance channel from the OSC (the software produces a synthetic L channel from OSC.)

Quite honestly there is a lot of 'blah blah blah' in this discussion. My own reality is that mono is faster for a given quality, has some additional tasks but, at a more advanced level, makes other tasks easier.

What the theoreticians don't discuss is that it depends an awful lot on the target. That's the reality. Galaxies seem to be harder work in the OSC and nebulae come out more easily.

If you are worried about the complexity of it all I would say this; it's all complex but you can do it! There is not a lot to choose between the two and I have developed a gut feeling about what will suite the OSC and what won't.

If I only had one... mono for certain. i couldn't live without Ha on the emission nebulae.

Olly

[AlistairHowie]

80 flats? ............... You don't need that many!! I use 30 - once they're set up to take, they're very quick and easy.

I only thought 80 on the basis of 20x each LRGB filter. Is that not right? If you can get away with, say 5 per filter, giving 20 total, then presumably an OSC would only require 5 flats? That doesn't sound quite right.

If you are worried about the complexity of it all I would say this; it's all complex but you can do it! Olly

Not worried about complexity, just trying to understand the additional work in capturing mono images (not processing, I can do that at my leisure) since I don't have the clear skies that you do, Olly I know that mono wins hands down and is pretty much the choice of most.

So ... if someone can tell me how many flats, typically, per filter, that would be great.

And I guess binning RGB on my set-up is not desirable? If not binning, how much less time can be spent on RGB subs compared to L? Or do they need to be broadly the same? Just trying to see where time might be saved. I know the answer is probably "it depends", but any rough idea would be a help.

Thanks again

[Stephen]

80 flats? ............... You don't need that many!! I use 30 - once they're set up to take, they're very quick and easy.

20 per filter = 80?

[swag72]

Sorry, misunderstood - 80 flats in total with 4 filters, 20 a piece - Yep that's about right. I thought you meant 80 with 1 filter!!

[r3i]

You have one additional task which is to debayer the flats but the time gain is significant.

Olly

Olly, could you explain a bit more what you mean here. I was under the impression that you applied flats prior to debayering the lights (to convert them from raw to colour images) so I don't understand why you'd need to debayer the flats. Would really appreciate you filling in this hole in my knowledge

[AlistairHowie]

... and thoughts on whether binning RGB on my set-up is desirable, or whether I can spend less time on RGB subs compared to L ... would be much appreciated.

Thanks again.

[dph1nm]

You cannot bin OSC - or if you do you lose the colour information.

I believe some Kodak OSC chips can be hardware-binned 'correctly' (i.e. red, blue & green pixels are binned independently before read-out).

NigelM

[ollypenrice]

Flats straight from the camera have the outline of the matrix visible as a chequer pattern so that would be transferred to the lights if the flats weren't debayered. I do it in AstroArt but I guess all the software products for image integration will do it. In AA I average combine the flats using a master bias as the dark frame. This throws up a flat with the matrix visible. I then go Image-Shift and shift by 0.5 on both axes to get a smooth flat. That's all there is to it. I always prepare master flats because I can use them throughout a single spell of dark time on any images I take. I very rarely rotate the cameras so this isn't a problem.

I tend to do more than 20 flats, by the way, because when integrating lights I see a huge difference in noise between a stack of 20 and a stack of 40. I may be missing something but I prefer a stack of 30 to 40 flats. There was a discussion recently about whether there was any need to take more than about 25 lights. 25 are good but 40 are better. I'm often in a position to compare shorter and longer stacks and, sad but true, 40 are better than 25!

Should you bin with your setup? In an ideal world, no. You won't get quite the same quality in the stellar images. I like to remove the luminance, where possible, from the stars because it tends to blow out the colour. But binned RGB stars won't look so good. They tend to be a bit blocky. However, the pressure of time might make binning worth it. Your FL isn't too short. When you are working at very short FLs the price of binning is high in terms of quality.

Olly

[AlistairHowie]

Thanks, Olly. So if binning isn't the greatest idea in the world, is another possible route to speeding up imaging reducing the time spent on the RGB subs versus L subs? To my simplistic way of thinking, the L should be the longest because it's the detail, but the RGB might be able to be of shorter duration as they're "only" contributing the colour. If they can be less, by what sort of amount? (My last question, honestly).

[ollypenrice]

Keep asking! I used to go and and shoot more colour if I found I had enough L to wash it out. I don't do that now, necessarily, because there's a trick;

Add the L to the RGB at about 30 percent opacity. Boost the saturation and denoise by using a noise filter or Gaussian Blur. Treat this image (call it LRGB1) as an RGB layer and add more L to it, say 60 percent. This is LRGB2. Up the saturation, blur it and then try the full luminance on that. In a recent Sky At Night I think Pete Lawrence explained another way of doing it but I haven't read up on it yet.

I wouldn't give up on binned colour at your FL. It may well give a fine result on some targets. Others, like clusters, are often best in just RGB but unbinned. You can't make hard and fast rules, I don't think, because different images behave differently. All part of the charm of making pictures.

Olly

[AlistairHowie]

Thanks, Olly. That pretty much went over my head, but that's fine. I've taken a copy of your reply and will keep it safe, and will refer to it once I'm up and running and know one end of a camera from the other!

I'm minded to place my FLT98, iEQ45 and 314L+ mono order with Ian King tomorrow. I've reached the analysis paralysis stage now. I was just having a last look at other Atik cameras like the 428EX, but I've realised that the gain in pixel numbers is offset by their smaller size ... so the FoV is near identical to the 314L. I can't afford the forthcoming 460EX, so that just leaves the 383L+ to consider. But that's more expensive and I'd need to buy the WO flattener IV, so yet more expense. Even if money were no object, which it is, my main concern is time requried to capture (not process) images, and I think I'm right in saying that I might get away without darks on the 314L+, but not the 383L+ and its Kodak chip. Is that right? (so there was another question in there after all) :-)