Lunch time Ha 2021-12-03 - Images, and an FPS problem.

[tombardier]

The seeing was wonderful yesterday lunch time.  The air was cold and crisp.  I thought the seeing was so fantastic at the eyepiece, that I'd get the camera out and try to capture some images.   There actually wasn't much going on at all.  I'm not sure you'd have seen more than a couple of tiny black dots if you were looking in white light, but in Ha it was still captivating.

Now, I've got to say, I haven't been blown away with these captures and I think I could see more at the eyepiece (albeit without the same contrast and so on!).  After getting a bit disheartened with the first couple, I just applied the same Deconvolution, MultiscaleLinearTransform (sharpening), and TGVDenoise (very low strength) settings to all of them (in PixInsight). 

There are a couple of problems.  Firstly, I haven't sorted out my little issue, chronicled here: https://stargazerslounge.com/topic/387166-only-solar-imagers-know-this-struggle/, and I didn't even get the towel out (mistake), and so I don't think I managed to achieve perfect focus. 

The second thing is that in there captures, I've only taken 100 frames when previously I've always gone for 500 frames. On my ASI290MM it would take 6-7 seconds to capture 100 frames at 80FPS in 12-bit mode.  Since I bought an ASI1600MM, using 12-bit mode, with an ROI that excludes the vignetting from the Quark, I can only get 7FPS.  The previous images I've produced with the ASI1600MM have been much better, even with 500 frames taking over a minute to capture, so I might go back to that.  I do wonder why my FPS is so low though.  I'm using Firecapture, in USB 3 mode.  I've upgraded my cable to a Lindy Chromo-line, which is really thick, and has stopped me ever seeing glitchy half-frames or lags, but hasn't changed my frame rate at all.  I'm sure I should be able to get 14FPS at full resolution in 12-bit mode!  I might go back to capturing more frames anyway.  Not sure if anyone here has a rule of thumb for how long you can capture before details start to visibly change in the chromosphere?

I think I should have got my 290MM out too, because the small pixels really do pick up huge amounts of detail, and I think it's easier to focus with it too!

Anyway, on to the images!

[vlaiv]

1 hour ago, tombardier said:

The second thing is that in there captures, I've only taken 100 frames when previously I've always gone for 500 frames. On my ASI290MM it would take 6-7 seconds to capture 100 frames at 80FPS in 12-bit mode.  Since I bought an ASI1600MM, using 12-bit mode, with an ROI that excludes the vignetting from the Quark, I can only get 7FPS.  The previous images I've produced with the ASI1600MM have been much better, even with 500 frames taking over a minute to capture, so I might go back to that.  I do wonder why my FPS is so low though.  I'm using Firecapture, in USB 3 mode.  I've upgraded my cable to a Lindy Chromo-line, which is really thick, and has stopped me ever seeing glitchy half-frames or lags, but hasn't changed my frame rate at all.  I'm sure I should be able to get 14FPS at full resolution in 12-bit mode!  I might go back to capturing more frames anyway.  Not sure if anyone here has a rule of thumb for how long you can capture before details start to visibly change in the chromosphere?

Try doing some in door testing.

There are a few settings that you need to tweak in order to get best FPS performance. I'm using SharpCap so I'll point out settings in that software, but FireCapture should have something similar. If not - give SharpCap a go?

- first, be sure to select exposure length that can provide wanted FPS. There is relationship between FPS and exposure length that goes FPS = 1 / exposure in seconds. This really means that you can't achieve more than 5fps if your exposure is 200ms. Since solar is in essence planetary type imaging - you should really restrain your exposure length to 5-6ms regardless of what your histogram or brightness of image say. Go for 10ms in very good seeing. Any exposure longer than that - blurs the image additionally as you won't be freezing the seeing and seeing induced shimmers will add motion blur to the image.

- You probably don't need 12bit precision. For that we actually might need to do some math, but if you are sampling at critical sampling rate - you can easily limit yourself to 8bit because short exposure help. In planetary - one often can't even saturate 8bit with needed exposure lengths (I think that signal is about 100e on critical resolution for 5-6ms exposure for Jupiter).

- Turn on high speed readout and USB boost:

Turbo USB is usually set at 80, but you should set it as high as you can without starting to have issues with image - like black frames (no image) or dropped frames / freezing and such.

In the end - get as many frames as possible. Don't limit your run to certain time interval - do research and see what sort of time span you can use before things start to change and get max frames you can in that time frame.

To figure out how fast things are changing - look up time laps images of solar activity that have time stamps on and judge by that (or do internet search on the term?). I never looked that up before so I can't be much of a help at the moment, but will look it up later when the time allows and post it here in case you don't manage to find info.

[vlaiv]

According to this video - you have about 60 seconds to gather data.

https://www.youtube.com/watch?v=u2bXZqASi1c

I would actually advise you to experiment a bit and capture longer video. It is easy to split video into chunks if you want to process shorter recordings.

AS!3 is capable of aligning features that displace few arc seconds - this means that any motion that happens on such small scales won't actually produce motion blur.

Additional benefit is that you can create animations by recording longer video and splitting it into chunks.

If you can - try to get large frame rate - like at least 100fps. That way you'll be able to get ~ 20000 frames total for 3-4 minute recording. That is good number of frames for planetary type stacking (you can keep 5-10% and still have significant SNR improvement of over x30).

[tombardier]

Thanks @vlaiv, for the comprehensive reply.  I'm going to give your suggestions a try, especially the bench-test, haha!

I took the images above at unity gain of 139, and 4ms exposure time.  I think I've already set the high speed mode, and tried upping the USB readout, but I will double check that again, because every time I've tried to tweak it in the past, I've been squinting in sunlight, and I probably haven't been rigorous!

I do have Sharpcap installed, but I've not used it at all! 

I think I started using 12-bit mode because I noticed banding in the colours on my lunar images, especially on the smoother looking maria.  I'll do some testing, but I'd certainly be happy with the performance boost if I could use 8-bit mode!

With my 1600MM, 1x1 binning, through the quark, I'm sampling at 0.28"/px. 

I started using Windows on an older laptop recently, after having been using my main linux machine for capturing.  I only started using it at all because I bought an old ATIK filter wheel for £75 ( a really old one with a separate control box), and ATIK decided to abrubtly drop support for it in Linux!  I was quite pleased to find that in addition to being able to use the filter wheel, I got a literal doubling of my frame rate (with my ASI290MM)!  I gather ZWO rely on some misuse of the USB system in Windows to get maximum performance, and the Linux kernel is too pedantic to allow it because it violates the USB specification!

I'll report back with my findings.  Thanks again!

[tombardier]

18 minutes ago, vlaiv said:

If you can - try to get large frame rate - like at least 100fps. That way you'll be able to get ~ 20000 frames total for 3-4 minute recording. That is good number of frames for planetary type stacking (you can keep 5-10% and still have significant SNR improvement of over x30).

I can easily get frame rates like that, and more with my ASI290MM, but the ASI1600MM has a much larger sensor.  I love being able to fit the full disc of the moon on the sensor in my 254mm f/4.7 newtonian, but the frame rate just dives off a cliff!  To get 100FPS, I think I'd only be using a small fraction of the sensor!  Ideally I'd have an ASI174MM too, haha!

[vlaiv]

2 minutes ago, tombardier said:

With my 1600MM, 1x1 binning, through the quark, I'm sampling at 0.28"/px. 

That might be way too much.

What scope are you using?

If you are using Quark - it has integrated x4.3 telecentric lens. Using scope that is say F/6 with it will result in F/25.8 system. I know that for quark to operate at its best - you need to be at F/25-F/30 - but that is very far from camera best operation.

ASI1600 has 3.8µm pixel size and since you are imaging at 656nm - optimum F/ratio is F/11.6 - you might be oversampling by factor of x2 or even x3.

Try preprocessing your images with PIPP and selecting 16 bit output (do calibrate with at least darks if not flats as well) and maybe even bin your data to be closer to optimum sampling rate. That will improve your SNR further (which makes it much easier to properly sharpen and there is no need for denoising).

[vlaiv]

1 minute ago, tombardier said:

Ideally I'd have an ASI174MM too, haha!

Well yes - for use with Quark - you actually want sensor with large pixels.

Only drawback is high read noise of ASI174mm

ASI482 seems like really interesting camera to be used with quark - but it is only color for now.

[tombardier]

I'm using an Explore Scientific AR102 102mm achromat, at f/6.5.  The 290MM has much smaller pixels than my 1600MM, and it produces much nicer fine detail, although it possibly was a little more suited to my 80mm f/7 ED doublet, thinking about it.  It samples at 0.21"/px in the AR102!  I chose the 290MM over the 174MM originally because of the read noise, and because I was only interested in planetary imaging at the time, so the sensor size was of no concern. 

[tombardier]

And sorry, yes, it's a Quark, so the 1600MM in the AR102 with the Quark is at a focal length of 2850mm

[vlaiv]

1 minute ago, tombardier said:

The 290MM has much smaller pixels than my 1600MM, and it produces much nicer fine detail,

On solar with quark?

[tombardier]

1 minute ago, vlaiv said:

On solar with quark?

Yes.  I think this is the last image I took with this system, using the 290MM.

[tombardier]

Of course, it could just be that I'm getting more than 10x the frame rate with the 290MM!

[vlaiv]

Let me show you something. I've taken crop of one of your images - one that seems sharp enough, and here it is (view this on screen that will display these images without scaling - so not on phone):

Then I resized this image down x3 - or down to 33% of original size (equivalent to x3 larger pixel size)

And now I'm going to enlarge it back again to 100% as you've captured it:

So this last image - is resized small image - enlarged. Can you spot any difference between original and it?

If you do that with any image that is properly sampled - difference will be obvious. With over sampling - you simply don't loose anything when you resample down - simply because there is no detail at that resolution. It cannot be due to physics - your telescope simply can't resolve in Ha at that scale.

[vlaiv]

Ok, here is the same thing done with that image done with ASI290:

One of them is original crop - other has been scaled to 1/3 of its size and then enlarged back to match original size - can you tell which is which?

[vlaiv]

My point is - neither of the cameras is used optimally as both grossly over sample.

ASI290 will create better image - as it can produce more frames, and more frames means better SNR of the stack - which you can sharpen more without bringing the noise out.

Properly sampling will give you even better SNR - and you'll be able to sharpen even more.

[tombardier]

I certainly take your point with the shot taken from the 290MM.  I can't tell which is which!  The shots in my original post are all a bit mediocre anyway, what with the focus issues and what not, so I guess they're not good examples to use.  I think on my next imaging session, I'll certainly try binning.  I might even try 3x3 and see how that looks!

Thanks for all your help and advice!

[Luke]

Nice images. I also limited captures to 60 seconds usually.