Drizzle vs Nondrizzle

[Rodd]

2 hours ago, wimvb said:

That's my understanding of the matter, yes.

OK--one less thing to worry about--not that drizzle is a mainstay of my workflow.  I am not undersampled enough I don't think to really take advantage of it.  Except the stars.  But, even at full resolution viewing on teh forum its not really zoomed enough to notice 

Rodd

[wimvb]

My experience is that oversampled, round stars are somewhat easier to process, and produce better masks than undersampled,  blocky stars. But processing larger images requires a more powerful computer / takes longer.

[Rodd]

5 hours ago, wimvb said:

My experience is that oversampled, round stars are somewhat easier to process, and produce better masks than undersampled,  blocky stars. But processing larger images requires a more powerful computer / takes longer.

Agreed.    Here is another very surprising result for me.  I have always heard that more data is needed with the asi 1600--and in general with any camera to a point--but especially the 12 bit 1600.  Well--check out these 2 images.  1 has 152 5min subs and the other has 79 5min subs.  What happened was I started with 179 and eliminated the worst of the worst--FWHM values out of line with the rest of the subs.  Then I went bonkers and eliminated many many more, trying to get the FWHM value down.  I succeeded in this--I brought the FWHM value down for the integrated image from 3.9 to 3.8--big deal, right?  (sounds high but my pixel scale is 2.46, so its a pixel and a half--not that bad really).  Anyway, for the life of me I could not see a difference--maybe a tiny bit smaller stars.  But then I began inspecting the images and the 79 sub one contains more contrast and finer details.  Zoomed in to 18:1 revealed no more noise then the 152 sub image---well maybe a teeny bit at 18:1, but really, hardly noticiable at that scale.  Have I been wasting my time collecting lots of data?  What about the 12 bit thing? I found a new species of gremlin....gremlin's of the mind!  The images are not labeled...to facilitate a blind inspection.  Both have equal histograms (as close as I could get them anyway in the horizontal axis).  No processing--just crop and dbe.

1

2

 

 

 

[Laurin Dave]

Hi Rodd

The second one looks better to me on extreme close up...  On the plus side if you'd only collected 79 subs you'd still probably have thrown away half and then final result wouldn't be as good, I guess the thing for us all to do is figure out what's causing the bad subs and fix that if it's within our control.  Both great images in any event.

 

Dave

[wimvb]

A really interesting comparison would be to integrate the 100 subs you excluded, and see how good/bad that image would be against the 79 best subs.

[Rodd]

43 minutes ago, wimvb said:

A really interesting comparison would be to integrate the 100 subs you excluded, and see how good/bad that image would be against the 79 best subs.

Well--most of them are in the 152 sub image.  The 18 on top of those had to be thrown out--in any case.  So its 152 vs 79 of the best.  I am wondering if the screen image doesn't suffer the dynamic range loss of a 12 bit sensor as badly as a printed image.  Quite frankly I am not seeing the need to collect a huge amount of subs (like 200-300 which is what I was advised).  By teh way--which image looks better to you.? I have an opposite opinion to Dave

Rodd

[Rodd]

1 hour ago, Laurin Dave said:

Hi Rodd

The second one looks better to me on extreme close up...  On the plus side if you'd only collected 79 subs you'd still probably have thrown away half and then final result wouldn't be as good, I guess the thing for us all to do is figure out what's causing the bad subs and fix that if it's within our control.  Both great images in any event.

 

Dave

I see it the opposite--to me, the first one has better contrast and sharper details.  But, I will say that not on the forum--the zoom is not enough.  To me they look the same on the forum--at last on the computer I am using at work.   On my computer at home, I took it to 18:1--which is way, way closer than full resolution mode here.  

Rodd

[peter shah]

I cant really see much of a difference....but both are tremendous renditions

[Rodd]

25 minutes ago, peter shah said:

I cant really see much of a difference....but both are tremendous renditions

Thanks Peter--a compliment to Takahashi and ZWO as no processing other than cropping and gradient removal was done on either.  I guess some targets are just easier than others!

Rodd

[wimvb]

Near the very top, the 152 sub image (nr 2) may be ever so slightly smoother. But definitely not by much. And I can't see any difference in the details.

[Rodd]

3 hours ago, wimvb said:

Near the very top, the 152 sub image (nr 2) may be ever so slightly smoother. But definitely not by much. And I can't see any difference in the details.

Like a fool I did not write down which is which--now I don't remember which is which!

Rodd

[wimvb]

The first image is named 79.jpg, the second 152.jpg. I took that as a clue.?

[Rodd]

23 minutes ago, wimvb said:

The first image is named 79.jpg, the second 152.jpg. I took that as a clue.?

2 things--1, how can you see what they are called?  I intentionally called them 1 and 2 to make it a blind test.  In order for me to know what they are, I have to edit the post and then I can see them--but it only shows me the name of the images that were uploaded--not the ones that were posted--so I don't know which I put in the #1 position..  Perhaps you downloaded them--that would make sense.  That brings us to the second point..2, no fair!!!!?

Rodd

[wimvb]

No download, just clicked them for maximum resolution. The name shows up in the heading.

6 hours ago, Rodd said:

no fair!!!!?

Ok, I guess I'm disqualified from the competition. ?

[happy-kat]

On my tablet I just touch and hold on an image and on the pop up control where I can open in new tab, copy, download etc. The full name is given of the image.

[James]

This is only anecdotal and my own experience but when I was imaging through a little Borg at 3.9" per pixel drizzling helped a little - as others have experienced in terms of roundness of stars for example.

Through a scope at 1.4" per pixel it makes no difference that I can determine (beyond taking extra processing time).

What really made a big difference for me was dithering in the first place. I only started to do it to try out drizzle but drizzle or not, dithering makes quite a difference to my images

James

 

 

[Rodd]

4 hours ago, wimvb said:

No download, just clicked them for maximum resolution. The name shows up in the heading.

Ok, I guess I'm disqualified from the competition. ?

No worries...it fizzled.

Rodd

[Rodd]

2 hours ago, happy-kat said:

On my tablet I just touch and hold on an image and on the pop up control where I can open in new tab, copy, download etc. The full name is given of the image.

Yeah-I guess when you download images you get the name.  I'll have to remember to name them appropriately for comparisons.

Rodd

[Rodd]

1 hour ago, James said:

This is only anecdotal and my own experience but when I was imaging through a little Borg at 3.9" per pixel drizzling helped a little - as others have experienced in terms of roundness of stars for example.

Through a scope at 1.4" per pixel it makes no difference that I can determine (beyond taking extra processing time).

What really made a big difference for me was dithering in the first place. I only started to do it to try out drizzle but drizzle or not, dithering makes quite a difference to my images

James

 

 

That's no surprise.  There was a time when I was starting out that I did not dither--before I had Maxim and was using CCDops for capture.  But its hard to know just how much not dithering hurt the image because the images were so bad.  Once I got Maxim I have always dithered.  I did forget to activate it once, though.  I don't recall seeing any major issues with the data (other than I did not have to crop!).  I guess it depends on the camera.

Rodd

[wimvb]

1 hour ago, Rodd said:

 I guess it depends on the camera.

It depends more on your guiding. If you have near perfect guiding with a star always ending up on the same pixels, then kappa sigma clipping has nothing to work on. The whole idea of using outlier rejection (aka clipping) is based on stars ending up on different pixels.

[Rodd]

5 hours ago, wimvb said:

It depends more on your guiding. If you have near perfect guiding with a star always ending up on the same pixels, then kappa sigma clipping has nothing to work on. The whole idea of using outlier rejection (aka clipping) is based on stars ending up on different pixels.

I was referring more to eliminating noise patterns through dithering--which I understand is a big benefit for certain cameras.  But I have never really noticed this too much--then again, I have always dithered.

Rodd