15 replies to this topic

[dcaponeii]

In my post regarding Rhea and shadow as a cautionary tale regarding the importance of collimation, I mentioned that the earlier data on Tethys and shadow was in progress.  I've just got them finished and indeed both Tethys and shadow are clearly visible in the GIF I've prepared from the data using the Wavesharp GIF builder and the ROI tool.  Processing the same as both the Rhea and shadow post and the post from two days ago on my first success at imaging Tethys and shadow.  Note:  I'm now pretty convince that it was simply the slew to Mars and back this morning that required a minor tweak of the collimation after having done so ahead of collecting these data.

 

[marvyyk]

So cool!! Thanks for sharing your capture! 

[rgsalinger]

Very nice indeed! No barlow being used?

[Borodog]

Did you drizzle that? Very weird artifacts.

[dcaponeii]

Did you drizzle that? Very weird artifacts.

Yes 1.5x drizzle plus some extra sharpening after rotating to north up. Both contribute to the artifacts. Without treating the moons separately it was the only way I could think of to make sure they were visible.

[dcaponeii]

Very nice indeed! No barlow being used?

Yes. A 2x Barlow element in front of the ADC.

[Borodog]

Yes 1.5x drizzle plus some extra sharpening after rotating to north up. Both contribute to the artifacts. Without treating the moons separately it was the only way I could think of to make sure they were visible.

I should have specified that it was a great capture and that the artifacts do not detract from it. Just very interesting black & white cross-hatched artifact; not quite like what I've seen before. I am thinking this must be what the drizzle grid artifact looks like with luminance sharpening.

[dcaponeii]

I should have specified that it was a great capture and that the artifacts do not detract from it. Just very interesting black & white cross-hatched artifact; not quite like what I've seen before. I am thinking this must be what the drizzle grid artifact looks like with luminance sharpening.

I have not noticed the grid.  You must have zoomed in a long way.  How much of that is coming from the file size reduction to fit the GIF onto CN.  For one of these I linked to an APNG in the gallery.  I'm wondering if you see it there as well?

 

EDIT:  When I look at that GIF about with my browser set at 200% I still don't see any grid pattern.
 

Edited by dcaponeii, 24 June 2024 - 10:00 AM.

[Borodog]

I have not noticed the grid.  You must have zoomed in a long way.  How much of that is coming from the file size reduction to fit the GIF onto CN.  For one of these I linked to an APNG in the gallery.  I'm wondering if you see it there as well?

 

EDIT:  When I look at that GIF about with my browser set at 200% I still don't see any grid pattern.
 

I didn't zoom in at all. I'm looking at it at 100% on my laptop's external monitor, which is 27" at 1920x1080, and the hatched grid is quite obvious.

 

I don't see a link to an apng. Must be in another thread?

 

Edit: Found it in the other thread. https://www.cloudyni...699_1329082.png

 

Yes, hatched grid is quite clear.

Edited by Borodog, 24 June 2024 - 10:05 AM.

[KiwiRay]

I didn't zoom in at all. I'm looking at it at 100% on my laptop's external monitor, which is 27" at 1920x1080, and the hatched grid is quite obvious.

 

I don't see a link to an apng. Must be in another thread?

 

Edit: Found it in the other thread. https://www.cloudyni...699_1329082.png

 

Yes, hatched grid is quite clear.

Isn't that just pixelation, only visible because the original image is scaled up? Or in the case of the animation above, compression artifacts (or both)? In both animations, I see it most clearly around the rings and planet's edge, but the grid squares are in exactly the same spots in each image - the grid doesn't move with the imperfectly aligned images. How would drizzling lead to that?

Edited by KiwiRay, 24 June 2024 - 10:39 AM.

[JMP]

Sweet!!

[Borodog]

Isn't that just pixelation, only visible because the original image is scaled up? Or in the case of the animation above, compression artifacts (or both)? In both animations, I see it most clearly around the rings and planet's edge, but the grid squares are in exactly the same spots in each image - the grid doesn't move with the imperfectly aligned images. How would drizzling lead to that?

Drizzling in Autostakkert always leads to that. It just looked unusual because it's been luminance sharpened instead of RGB sharpened. This is why I have never liked drizzling OSC data. Andrew (Tulloch) convinced me that other methods of up-scaling can also produce gridding artifacts, so it isn't a problem that is particular to drizzle in Autostakkert (as I used to believe), but I still don't like it, and no-one has ever yet been able to produce a drizzled OSC lunar or planetary image that is clearly superior to a simple scaled up non-drizzled version. It takes much longer to stack with drizzle vs. a mouse click to scale up and get results that contain just as much detail and less noise.

 

I would believe that if you were very under-sampled and stacked a LOT of data there could be a conceivable benefit to drizzling OSC data, but none of use who shoot OSC planetary meet those criteria. We all shoot at critical sampling and the amount of data we have to stack is limited by the rotation of the planet. You could do it with the Moon, for example, and I have, although I did not convince myself that the result was superior to simply scaling up the non-drizzled image. But there is no way that drizzling a critically sampled image can be anything but a waste of time and a handy way to make artifacts.

[KiwiRay]

So the grid is induced by up-scaling, whether through drizzling or not. In this case, most of the upscaling occurred during creation of the gif (after the images were aligned), which is why the grid doesn't shift in unison with Saturn - is that what you're saying? So if Don had rotated each image prior to creating the gif, the grid orientation would still be the same, whereas if the grid had been induced by drizzling, it would rotate with the images.

 

I'm not arguing that drizzling produces better results than simply working on an upscaled image - examples I've tried agree with your assessment.

[Borodog]

So the grid is induced by up-scaling, whether through drizzling or not. In this case, most of the upscaling occurred during creation of the gif (after the images were aligned), which is why the grid doesn't shift in unison with Saturn - is that what you're saying? So if Don had rotated each image prior to creating the gif, the grid orientation would still be the same, whereas if the grid had been induced by drizzling, it would rotate with the images.

 

I'm not arguing that drizzling produces better results than simply working on an upscaled image - examples I've tried agree with your assessment.

Not all up-scaling algorithms produce gridding. Not having access to the drizzled but not yet further up-scaled images, I can't say for certain that the problem was caused by drizzling, but I would bet you beer that it was.

[dcaponeii]

Isn't that just pixelation, only visible because the original image is scaled up? Or in the case of the animation above, compression artifacts (or both)? In both animations, I see it most clearly around the rings and planet's edge, but the grid squares are in exactly the same spots in each image - the grid doesn't move with the imperfectly aligned images. How would drizzling lead to that?

 

 

Not all up-scaling algorithms produce gridding. Not having access to the drizzled but not yet further up-scaled images, I can't say for certain that the problem was caused by drizzling, but I would bet you beer that it was.

I watched it happen this afternoon on an image I was working with in Wavesharp.  It happens when you use a reduced ROI (cropping in around the area where the moon and moon shadow is located.  I goes away when you open up the ROI to the full capture image size.  Whether or not the artifact is caused by drizzling isn't clear but it only appears when I use the ROI = "R" setting in Wavesharp.

 

The funny thing is that I can't see it on any of my monitors (1080p external) or either of my laptop monitors in these posts.
 

Edited by dcaponeii, 24 June 2024 - 02:10 PM.

[Borodog]

My external laptop monitor is truly gigantic for 1920x1080. I think I said it is 27", but that can't be right. It's much larger than my workstation monitors and those are 27". I think it must be 32". In any event it is easy to see individual pixels. It's a good monitor to process on because it hides no crimes. My workstation monitors are smaller but higher resolution, 2560 x 1440, and the gridding is difficult though not impossible to spot on those.

 

Edit: Confirmed, it's 32". I actually got it for mixing (as in audio production), but it also makes a great monitor for AP processing, although the blacks are very dark compared to my laptop, which makes DSO backgrounds tough to ever get right.

Edited by Borodog, 24 June 2024 - 03:12 PM.