The Flea in the Pussycat's Ear: Trifid SHO 17 hrs

[irwjager (Ivo)]

Lovely deep data here! If I may, though, make a suggestion?
Utilise the enormous oversampling. The massive resolution is just going to waste showing "smeared out" detail. Instead bin your data so you can push it harder and/or use deconvolution to restore the latent detail (preferably a combination of both).
You can scale it up again with your favourite interpolation algorithm if - for whatever reason - you really like big images.
Clear skies!

[Placidus (Mike and Trish)]

Originally Posted by Bart

Very deep, lots of detail.

Thanks Bart!

Originally Posted by ChrisV

Beautiful, and my cat Romana (DvoraTrelundar) agrees.

Cheers, Chris! It is good to know that an elegant and discerning referee is on side!

Originally Posted by rustigsmed

excellent as usual M&T. I don't usually like a nb of the triffid but you may be changing my mind with this rendition!

cheers

russ

Thanks, Russ. Perhaps the middle position is to notice that the SII really isn't telling us so very much in this image, and to present it as Ha-OIII only. But as mentioned before, the fact that SII is on the quiet side is actually useful information.

Originally Posted by Atmos

Very nice!
A deep image with some great details in there MnT

Many thanks Colin

Originally Posted by irwjager

Lovely deep data here! If I may, though, make a suggestion?
Utilise the enormous oversampling. The massive resolution is just going to waste showing "smeared out" detail. Instead bin your data so you can push it harder and/or use deconvolution to restore the latent detail (preferably a combination of both).
You can scale it up again with your favourite interpolation algorithm if - for whatever reason - you really like big images.
Clear skies!

A thoughtful suggestion. The Nyquist limit for a 2-dimensional image (as opposed to a 1-dimensional signal) is 3 pixels to the smallest recordable detail. We usually get 2 sec arc seeing. So the Nyquist limit would be 0.67 sec arc per pixel. Unbinned, we are at 0.55 sec arc per pixel, just meaningfully inside the Nyquist limit. Binned, we'd be at 1.1 sec arc per pixel and throwing out information, so that it was mathematically unrecoverable. Dithering at image capture time and drizzling to reconstruct can recover some of the information.



Binning is usually done in order to increase the signal to amplifier readout noise. We have a 20 inch scope, we do long exposures, and lots of them. Amplifier readout noise is not a problem for us. Our images are not gritty. Hence, binning for us would just throw away information with no benefit.


(On that topic: we have a 16803 chip. On-chip binning would be meaningful for us, if the problem were signal to amplifier noise ratio. But the new ZWO cameras (we have two, which we use for guiding) have no on-chip binning. If you ask for binning with a ZWO 16000M for example, you get no improvement in signal to readout noise, only an increase in video frame rate. So there is NEVER any point in binning with a 16000M).


In conclusion, in our case, the hard cold mathematics is that binning would achieve little, and inevitably throw away some information.


What would help us a lot is moving to 2.5 Km up in the Atacama desert, but it would be hard to raise our chooks steers and veggies there.


Thanks muchly for the thoughts though.


Best,
Mike

[irwjager (Ivo)]

Originally Posted by Placidus

The Nyquist limit for a 2-dimensional image (as opposed to a 1-dimensional signal) is 3 pixels to the smallest recordable detail.

It's fantastic to hear you are across the conditions for when you'd want to bin (or not bin) and/or apply deconvolution. I'm hopeful then that you'd agree that the smallest resolved detail in your dataset is very, very far from 3 pixels.

E.g. a point-light at infinite distance (e.g a star) in your "Big One" image looks like this;

http://download.startools.org/Tutori...ts/MnT_PSF.jpg

This close-up of a star is very close to the point spread function (PSF - though a non-linearly stretched version of course) in your entire image. Every bit of real recordable detail is "smeared out" in this way. As you can see, the pixel in the middle (the center of the star) looks virtually identical in brightness to its surroundings. E.g. much of its energy bleeds into neighboring pixels. It's text-book oversampling and makes the scene blurry/soft for no good reason.

Binning is usually done....

Firstly, I think you're confusing one particular flavour/application of binning ("hardware" binning) with the general procedure itself, which can be done in software. If the read noise consideration you expertly lay out does not apply, then it is preferably done in software these days, as it gives you more flexibility (e.g. fractional binning/resampling or rejection algorithms) and avoids possible CCD well limitations too.

Atmospheric seeing is only half the story; your optical train and/or its configuration can equally have these effects (or worse). Perfect optics don't exist, and a circular opening comes with an innate PSF (the Airy disc) to begin with.

In conclusion, in our case, the hard cold mathematics is that binning would achieve little, and inevitably throw away some information.

All I'm saying is that when reality doesn't match up with the mathematics and assumptions, it's usually not a flaw in the fabric of reality

What would help us a lot is moving to 2.5 Km up in the Atacama desert, but it would be hard to raise our chooks steers and veggies there.

The animals can rest easy, as there is an easy (though not perfect) solution to this. Bin (if it benefits your signal) and apply deconvolution to restore the lost detail. PI should have modules/functionality for both. It is important to note that binning lets deconvolution restore more using the improved signal, hence the recommendation to combine measures of both.

Right now, you are leaving detail on the table, which is a shame!

http://download.startools.org/Tutori.../MnT_Decon.jpg(your JPEG binned to a quarter resolution, then deconned; applying decon restoration to an already processed lossily compressed 8-bit image is a big no-no of course, so you should be able to do a lot better with much fewer ringing artefacts when using the real linear data)

Hope this helps!

[Placidus (Mike and Trish)]

Ivo, you are very wise, and say many things that are true but irrelevant. Let us agree that my Trifid is oversampled, but that is not the point. There are two points that I am debating.

(1) You seem to argue that deconvolution can reverse the information loss caused by binning. This is wrong.

(2) You seem to argue that 2x2 binning in software can somehow increase the information in an image. This is wrong. It can only ever lose information. In a particular case, it might lose NO information, or it may lose a tiny amount of information, but it can never INCREASE the amount of information.

I will now explain and justify these two assertions.

The effect of seeing, diffraction, tracking, and optics can be modelled as convolving God's Own Image with a point spread function, and a good estimate of the point spread function is the actual image of a star.

What deconvolution does: it finds an image, which, when convolved with the point spread function, produces the photo that you actually took.

To that extent, deconvolution is an attempt at applying the inverse function to the convolution by the atmosphere.

What binning does: it irretrievably loses topographical information. THERE IS NO INVERSE FUNCTION.

Consider this: I am thinking of a number between zero and one million. I tell you that the number I am thinking, modulo six, is five. What number am I thinking of? There is no unique answer. The vast bulk of the information is irretrievably lost. There is no inverse function to the modulo operator, and there is no inverse function to binning.

Why I think you are confused:

If I take my published image and post-hoc bin it 2x2 or even 3x3, it LOOKS sharper. But that is a property of the human eye, not the image The eye is very forgiving and imagines the extra detail. No extra information has been added. It is an optical illusion.

Adding noise to an image makes it LOOK sharper, but again, that is the physiology of the human eye, a quirk of our visual perception. No extra information has been added. It is an optical illusion.

A third example: increasing the contrast of an image makes it look sharper. But in fact, increasing the contrast can at best only ever LOSE information. Again, an optical illusion.

So others publish images that are binned, have residual noise left in on purpose, have increased local sharpness, because it fools the eye, makes for a pleasing image, but there is in fact no more information in the image. These glosses are all cosmetic only.

I look forward very much to seeing your image of the Trifid taken from an altitude of no more than 660 meters, deserts excluded.