Under sampling

[astro mick]

Hi.

I must admit I have never paid any attention to the importance of this but now I think I will have to.

Although in focus my images always seem soft and lack fine detail.

I use mainly scopes around F7.5 or F 6.0 with a reducer coupled to an Atik 314L+.

Using Astronomy tools calculator,i am under sampling significantly.

I would have to switch to an F9 focal length scope to get the sampling right.But who images at F9 except planetary imagers.

Must admit this is causing me confusion,as to what to do.

Any advice on this.

Mick.

[CraigT82]

Have a look into drizzle... works well on harpening up undersampled images.

Plenty image at F9 or longer... galaxy imagers with SCTs for example.  Just have to be a little more patient! 

Edited March 29, 2020 by CraigT82

[ollypenrice]

The F ratio isn't relevant. For a given camera the sampling rate is determined by the focal length. The key unit for resolution is arcseconds per pixel and this is defined by just two variables, the focal length and the pixel size. F ratio doesn't come into it. (With one minor caveat: it is harder to find and maintain critical focus in fast F ratio scopes because the steep light cone reduces the depth of field. This really only matters much with ultra-fast astrographs.)

What image scale are you working at when you say you're under sampled? Our widefield images are taken at a very coarse 3.5"PP so we cannot resolve fine details but there is no reason for the images to look soft or for the stars to look 'blocky' at this scale.  It's a matter of choosing the right targets but is this 'soft' at 3.5"PP?

To be able to present small, highly detailed subjects like galaxies at large screen sizes you need to be approaching 1"PP. However, the danger of getting a 'soft' result greatly increases at this kind of scale because the seeing has to be good enough to support it and so does the guiding. Coarser pixel scales are far more tolerant of poorer seeing and guiding.

It would be helpful to see an example of an image with which you're not happy. I suspect that your sampling rate may not be the culprit and would look at focus and guiding. Beware of the 'round stars' test of guiding. It is a flawed test because equal errors on both axes, even quite large ones, will give you round stars while detail is being blurred out all the same.

And, regarding F ratio, remember that more aperture means more object photons. (The 'F ratio myth' is a volatile subject!) There is no reason to fear imaging at F9 if the entire object fits on the chip and you have decent aperture.)

Olly

 

Edited March 29, 2020 by ollypenrice

[astro mick]

11 hours ago, CraigT82 said:

Have a look into drizzle... works well on harpening up undersampled images.

Plenty image at F9 or longer... galaxy imagers with SCTs for example.  Just have to be a little more patient! 

Thanks Craig.

Yes I have heard of drizzle but have never used it.Something I will have a look at.

Mick.

[astro mick]

11 hours ago, ollypenrice said:

The F ratio isn't relevant. For a given camera the sampling rate is determined by the focal length. The key unit for resolution is arcseconds per pixel and this is defined by just two variables, the focal length and the pixel size. F ratio doesn't come into it. (With one minor caveat: it is harder to find and maintain critical focus in fast F ratio scopes because the steep light cone reduces the depth of field. This really only matters much with ultra-fast astrographs.)

What image scale are you working at when you say you're under sampled? Our widefield images are taken at a very coarse 3.5"PP so we cannot resolve fine details but there is no reason for the images to look soft or for the stars to look 'blocky' at this scale.  It's a matter of choosing the right targets but is this 'soft' at 3.5"PP?

To be able to present small, highly detailed subjects like galaxies at large screen sizes you need to be approaching 1"PP. However, the danger of getting a 'soft' result greatly increases at this kind of scale because the seeing has to be good enough to support it and so does the guiding. Coarser pixel scales are far more tolerant of poorer seeing and guiding.

It w2ould be helpful to see an example of an image with which you're not happy. I suspect that your sampling rate may not be the culprit and would look at focus and guiding. Beware of the 'round stars' test of guiding. It is a flawed test because equal errors on both axes, even quite large ones, will give you round stars while detail is being blurred out all the same.

And, regarding F ratio, remember that more aperture means more object photons. (The 'F ratio myth' is a volatile subject!) There is no reason to fear imaging at F9 if the entire object fits on the chip and you have decent aperture.)

Olly

 

Hi Olly.

Thank you so much for this detailed answer.

At 3.5"PP the image looks absolutely fine,areas are sharp and defined,and makes me wonder if part of it is the quality of equipment used,as well as expert processing.

My Luminance is taken with a Celestron ED80 with an F Ratio of 7.5 giving a F/L of 600mm.This with an Atik 314L+ mono ccd.

Now according to the the calculator that gives me 2.22"/Pixel when again the calculator says this is significant under-sampling.Ideally it should be 0.33-1"PP.(As you say)

This suggests to me that I cant produce fine detail in an image even with Photo-shops tricks.

I enclose this recent photo of M101 If you look carefully at the dust lanes they are not sharp but soft,even the overall image.Now this might well be focussing and tracking,but I would be surprised if it was + of course my woeful processing.

I would be interested in your views Olly.Your the man.

Mick.

[ollypenrice]

17 minutes ago, astro mick said:

Hi Olly.

Thank you so much for this detailed answer.

At 3.5"PP the image looks absolutely fine,areas are sharp and defined,and makes me wonder if part of it is the quality of equipment used,as well as expert processing.

My Luminance is taken with a Celestron ED80 with an F Ratio of 7.5 giving a F/L of 600mm.This with an Atik 314L+ mono ccd.

Now according to the the calculator that gives me 2.22"/Pixel when again the calculator says this is significant under-sampling.Ideally it should be 0.33-1"PP.(As you say)

This suggests to me that I cant produce fine detail in an image even with Photo-shops tricks.

I enclose this recent photo of M101 If you look carefully at the dust lanes they are not sharp but soft,even the overall image.Now this might well be focussing and tracking,but I would be surprised if it was + of course my woeful processing.

I would be interested in your views Olly.Your the man.

Mick.

I would never advocate an image scale of 0.33"PP or anything like it. 0.33 is crazy, totally unworkable for long exposure imaging, though not for fast frame work. 0.33"PP would require a guide RMS of about 0.16 arcseconds and uncanny seeing. (Guide error mustn't be more than about half the sampling rate.)  Our excellent Mesus run an RMS of around 0.3 arcsecs.  I'm not convinced that I was getting any more useful detail out of 0.63"PP than I can get out of my present 0.9"PP so I would make 1"PP the minimum sampling rate.  But before you buy anything to give you that rate, what is your guide RMS? A good EQ5 or 6 can give about 0.5 arcsecs under guiding so that will support an image scale of 1"PP when the seeing allows. (If you give PHD2 your guide camera pixel size and guide scope focal length it will give your your guide RMS in arcseconds.) Do you know your guide RMS, by chance? 

With only 80mm of aperture you are also short of optical resolution so a scope with more aperture and more FL will help on this kind of target. However, many of us would be very happy to work at 2.2"PP on more diffuse targets. On many nights the seeing won't support much more than that.

I'd say your M101 was a creditable effort for 80mm/2.2"PP. A lot of the nice resolution you'll have seen in other images will come from good post processing, too. Any kind of sharpening needs lots of signal. You may have enough exposure to give a low noise image unsharpened but when you sharpen you will need twice to four times as much. Only sharpen selected areas of strong signal as well.

What I would do with that image is adjust the green-magenta axis in the colour balance. It's very magenta so upping the greens ought to help.

Olly

 

[astro mick]

Thankyou for all this Olly.

Having it explained makes it easier to understand,and I will not now be rushing out to buy new equipment.(If only)

Mick.

[newbie alert]

I wouldn't get too wrapped up in sampling other than to look at what I need to guide at to my image scale, I'm sampling at 2.24 atm and beforehand I was at over 3... I did drive myself partially insane as my stars looked wrong so was chasing flex, then I moved onto sampling and drove myself nuts, I had to change my scope as the focuser broke and also changed camera's.. I now think that my problems were focuser sag..

Olly has shown quite remarkable images sampling at 3.5... but I suppose with god like wizard processing skills who cares about sampling

Don't beat yourself up on trying to get your sampling down to the level that is mentioned to be optional.. I've never seen blocky stars, but then again I'm not blowing images up to poster size