Jump to content

Narrowband

Abell 2218 - sheer folly!


mikey2000

Recommended Posts

I'll post this just for fun and i hope it makes sense.  It's a foolish endeavour and feel free to skip to the next topic in the forum... 

 

I saw a rather wonderful pic of Abell 2218, showing clear evidence of gravitational lensing.  Who wouldn't want to recreate something so spectacular???? 

 

Here's a series of pics showing how well I managed with my 200PDS and ASI1600MM-Cool.   Before you get too excited, I image at about 0.78" per pixel (and guiding approx 0.5" rms that night) and Hubble is about 0.05" per pixel with presumably perfect seeing and guiding.

 

Here we go, with thanks to Hubble.com:

 

The Hubble shot (full resolution)551879686_Abell2218fromHubble.thumb.jpg.b8fca354a2d813cc3aaa190325a567e2.jpg

 

The full shot from me, showing upcoming crop areas:  The inner square is approximately what the hubble took.

 

1277779547_Abell2218WideImage.thumb.jpg.3588fb481ee7d64d357a7e3b7aebf101.jpg

 

 

Lets zoom to the first square, showing the hubble pic overlayed in the right place.

 

1712367688_Abell2218RegionWithHubbleOverlay.thumb.jpg.e5a288630a33f28edbb540c749c7b40b.jpg

 

 

Here's the same again but with out the hubble overlay:

1197214222_Abell2218regionwithnooverlay.thumb.jpg.3934b879d101fc4ca69c567dbddfacf1.jpg

 

 

And finally a the tiny crop of my pic, next to the hubble pic at a similar scale:

719901672_Abell2218FinalZoom.jpg.28788e078631620ff55a29037d2600b2.jpg528910766_Abell2218Scaled.jpg.a37fcec3d0da28c463a35f86606e0e28.jpg

 

So, no lenses but I think I found a few faint fuzzies LOL!

 

I don't see any lensing ?

(or colour - I took Lum only)

 

The thing is, lots of these hubble pics don't really tell you the image scale.  It seems the Abell2218 image must have been a hubble 1:1 crop....

 

I think I'll look for some larger targets....

  • Like 25
Link to comment
Share on other sites

First of all, I applaud the effort!

Second, I don't think you were far off with that image, which I find great as I was looking into similar endeavor not long a go. At 0.78"/px you are clearly oversampling for given conditions (what is your star FWHM in that image?), but it looks like effect could be recorded at 1"/px resolution if conditions were such that given resolution was optimum (which would mean about 1.6" FWHM).

As it stands, I believe lensing is smeared due to blur present in the image - and that just reduces SNR of effect dramatically.

  • Like 2
Link to comment
Share on other sites

Agree with @vlaiv - congrats for even giving this a go and thanks for an interesting, informative, and well-constructed post.  I'd be delighted to have emulated the Hubble image as well as you have!

 

Link to comment
Share on other sites

Thanks for the kind words!  If nothing else, I have learned the limitations of a ground based 200mm scope at the edge of a suburban orange zone ?

 

I might add that the subs were taken about an hour after full moon rise ?  Maybe I'll try again one a clear night (but maybe I'd rather focus my efforts on a more 'likely' target ho ho)

 

Vlaiv - about the FWHM question ....  I'm not sure.  Do I have to pick certain size/brightness stars an measure this value or are you interested in the whole-frame average? (I have pixinsight if you can suggest a tool to provide the info you are after.)

Link to comment
Share on other sites

25 minutes ago, mikey2000 said:

Vlaiv - about the FWHM question ....  I'm not sure.  Do I have to pick certain size/brightness stars an measure this value or are you interested in the whole-frame average? (I have pixinsight if you can suggest a tool to provide the info you are after.)

Ideally one would want average across the FOV. FWHM should not change much from star to star. It is pretty consistent measure. There are always slight variations due to noise and because blue part of spectrum gets scattered more in atmosphere and seeing influences it more, but values should be close. However, due to optical design and field curvature there is larger variation between central and outer regions (this is the way one can measure field curvature of particular scope).

I guess best way for determining resolution would be to take average of stars in a crop around target area - much like larger rectangular area in your first image outlining region of interest. I don't use PI myself, but I'm certain I can find the right tool online, give me a sec ...

Here it is, under DynamicPSF process, reference page:

https://pixinsight.com/doc/tools/DynamicPSF/DynamicPSF.html

Take gaussian profile of region of frame and look under FWHM column. There is probably option to get average value or something. If it's in pixels, we would need to convert it to arc seconds by use of imaging resolution.

Link to comment
Share on other sites

Dynamic PSF is a tool that needs a lot of hard work.  After reading your post above, I think I worked out how to get the value you want (I used the PI script "FWHMEccentricity")  It reports 4.41 pixels for the larger crop area in the original frame.  My image scale of 0.78"/pixel therefore gives FWHM of 3.44 arcsecs.

 

(Just out of interest, I checked a few of my other images and they come up with similar numbers)

Link to comment
Share on other sites

1 minute ago, mikey2000 said:

Dynamic PSF is a tool that needs a lot of hard work.  After reading your post above, I think I worked out how to get the value you want (I used the PI script "FWHMEccentricity")  It reports 4.41 pixels for the larger crop area in the original frame.  My image scale of 0.78"/pixel therefore gives FWHM of 3.44 arcsecs.

 

(Just out of interest, I checked a few of my other images and they come up with similar numbers)

I figured it's going to be such a largish number based on blur level of 1:1 image. 3.44" FWHM blur image should be sampled at about 2.15"/px. You could bin x2 your data to improve SNR without loss of detail. Maybe even x3 with very small loss in detail. That will improve SNR significantly and keep resolved detail about the same.

Link to comment
Share on other sites

Interesting thought!  I've not been in this game long enough to worry about x2, x3 etc and loss of detail.  I've alway just shot full resolution with my ASI1600MM-pro-cool as I read that the binning is in software.

 

If I use the camera in full rez mode then later downsample in post processing, would I still get the same benefit?  and if so, when is the best time in the process to downsample (eg, before integration, after stretching etc etc?

Link to comment
Share on other sites

17 minutes ago, mikey2000 said:

Interesting thought!  I've not been in this game long enough to worry about x2, x3 etc and loss of detail.  I've alway just shot full resolution with my ASI1600MM-pro-cool as I read that the binning is in software.

 

If I use the camera in full rez mode then later downsample in post processing, would I still get the same benefit?  and if so, when is the best time in the process to downsample (eg, before integration, after stretching etc etc?

With cmos sensors it's better to shoot in full res without binning in drivers. You can bin later in software to same effect (or even better).

Down sampling using one of resampling methods is going to be "midway" in terms of SNR gain and pixel blur. "Regular binning" produces same effect as larger pixels, but pixel size is related to phenomena called pixel blur which degrades recorded details a bit in addition to seeing, scope aperture and guide performance combination. On the other hand "Regular binning" has best SNR increase.

Down sampling using other resampling method like Bicubic, or Splines or Lanczos is somewhere in between - it introduces very small additional "blur" which is related to pixel blur (less than straight forward binning) - but SNR gains are less compared to regular binning. Each filtering method will have different "stats".

By far best method for doing this is splitting subs before stacking, however I don't think that PI can do this - that is something not implemented in PI. It's a bit like working with OSC sensors and using color from every other pixel (like red and blue pixels). In case of mono, depending on "bin factor" image is split into multiple images - each one containing pixels with odd or even indices (in case of bin x2) or indices such that reminder when dividing with bin factor be 0, 1, ... up to bin factor-1.

Then you stack resulting images - you get much more data to stack that way, but you keep pixel size the same - no averaging of pixels. Result is no pixel blur increase, while number of subs is such that SNR improvement is equal to regular binning.

But for time being, you can use PI integer resample tool - it allows for binning if you down sample by integer factor and select average method. Try bin x2 this way on linear stack - right after stacking to see if you can pull out additional detail with increased SNR.

Link to comment
Share on other sites

OK, I spent a somewhat significant amount of time reprocessing the data today.  Now I've got back from seeing Avengers Endgame, I've got time to post the results. 

 

Spoiler alert:  No gravitational lenses evident in any version.  (DOn't worry about Endgame spoilers - I'm saying nothing ?   )

 

Here are 4 extreme close ups.  One is from a x2 drizzle integration, one is a x1 'normal' integration.  Then I used the PI IntegerRescale (Average mode) to downsample, all processed and NR'd to the best of my ability and put next to the original hubble image (now in mono)

 

Amusingly, the Hubble pic had to be *shrunk* to match even the 4x zoom into the 2x drizzled integration.   

 

I can see that the Downsampled version has better SNR but the drizzle version manages to resolve some of the faint fuzzy's shapes a bit better

 

2019-04-26.thumb.png.f97197568934d09332265b9015348a7c.png

  • Like 1
Link to comment
Share on other sites

At these extreme magnifications, this is all somewhat academic.

 

over the last few nights, I managed to gather a bit of data on M51 (that old chestnut!).  It's still quite small in the frame but it's 'realistically small' as opposed to 'impossibly small' (like the Grav lens features of that hubble Abell image).   I'll get around to is sooner or later but I'll be sure to compare the various drizzle, standard and downsample methods to compare detail/snr etc

Link to comment
Share on other sites

I love these threads when imagers try and reproduce some of the classic HST images, some of the results are quite surprising as with this one. It make me feel really lucky to be around at a time when amateur equipment has developed to a point when this experiment can not only be attempted but produce credible results.?

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.