Processing under sampled stars

[Astroblagger]

Hi

iv been looking for a new imaging scope something widefieldish, been looking at either the zenith star 61 or SW 72ed, but going on CCD carc it says my Atik 460 will under sample the stars, how critical is this? Is it a deal breaker? As I want a fairly large fov And is it correct this can be sorted in post processing? 

[vlaiv]

SW72ED with Atik460 and no reducer / flattener will give you 2.23"/pixel.

While one can argue that this resolution is under sampled, in no way it will affect star shapes - you simply will not get "blocky" looking stars. Even with reducer / flattener this will not happen.

The smaller is the aperture you are using - more "blur" comes into image from light refraction. Couple that with usual seeing blur and blur that comes from pixel size (yes there is also a blur related to the fact that camera is not point sampling device, but rather integrates photon flux over some area - pixel) - there will be enough blur to avoid such sampling artifacts as pixelated stars.

Bottom line, don't worry about being under sampled, as it will not show as artifacts in the image - you will simply have image at lower resolution, but with wider FOV.

[Pompey Monkey]

On 04/09/2018 at 10:08, vlaiv said:

SW72ED with Atik460 and no reducer / flattener will give you 2.23"/pixel.

While one can argue that this resolution is under sampled, in no way it will affect star shapes - you simply will not get "blocky" looking stars. Even with reducer / flattener this will not happen.

The smaller is the aperture you are using - more "blur" comes into image from light refraction. Couple that with usual seeing blur and blur that comes from pixel size (yes there is also a blur related to the fact that camera is not point sampling device, but rather integrates photon flux over some area - pixel) - there will be enough blur to avoid such sampling artifacts as pixelated stars.

Bottom line, don't worry about being under sampled, as it will not show as artifacts in the image - you will simply have image at lower resolution, but with wider FOV.

I get plenty of blocky stars at 2.79"/pixel (80mm f5). I reckon with good tracking and seeing, then square stars will begin to show.

Drizzle is your friend here

[vlaiv]

4 minutes ago, Pompey Monkey said:

I get plenty of blocky stars at 2.79"/pixel (80mm f5). I reckon with good tracking and seeing, then square stars will begin to show.

Drizzle is your friend here

You actually don't need to do drizzle - just resample that image x4 (my guess is that 4:1 in title means zoom in x4 over native 1:1 display) with some good resampling interpolation instead of nearest neighbor and you will get smoothness on the same level as drizzle

[Pompey Monkey]

Well, the HST uses drizzle rather than interpolating.  So that's good enough for me 

 

[vlaiv]

2 minutes ago, Pompey Monkey said:

Well, the HST uses drizzle rather than interpolating.  So that's good enough for me 

 

Well that is because HST is outside of atmosphere and only blur they get on their images is due to system PSF, so they can actually predict (calculate) loss of resolution and if it makes sense to do drizzle at all. Remember HST has 2.4m mirror and probably using scientific CCDs with 9-12um pixel size (don't know exactly, but that would be ball park, well from wiki data, FL of 57.6m and 0.04"/pixel that is about right). In order to fully utilize 12um pixels without the influence of atmosphere you need ~ F/50-F/60 scope, while HST is F/24 scope, so pixels are not small enough. This is why HST developed drizzle algorithm.

Now as we deal in long exposure AP from earth, and we have atmosphere and seeing to contend with, very rarely do we need high resolutions that some people use, and for most part, due to combined aperture, seeing and guiding PSF, you will not see "blockiness" of stars, unless either you use inadequate resampling - like nearest neighbor or if you grossly oversample - and even then it is question if it will be noticeable when image is presented in 1:1 zoom. Just take a look at widefield shots done with short FL lenses - stars look like dots, not like blocky pixel thingies.

Just look at this quick "simulation" - background image is sampled at 1"/pixel, left foreground is binned to 6"/pixel and then displayed at the same pixel scale as original (nearest neighbor interpolation), while foreground right is binned 6"/pixel, resampled back to original resolution of 1"/pixel with simple cubic spline. It shows obvious loss of detail and distortion of the stars (as would any image enlarged x6 original sampling resolution), but no that kind of blockiness that we are talking about here.

If I did experiment with 3"/pixel - I doubt that anyone could tell the difference. One is original image at 1"/pixel, other one is binned to x3, so at resolution of 3"/pixel, but scaled up x3 with cubic spline. Can you tell which is which?