A few of my favourite lunar images, and some questions about aligning monochrome images

[tombardier]

Hi.  I just joined SGL this week.  I've managed to introduce myself in the solar forums so far, but thought it was time to introduce myself here too.  These are a few of my favourite lunar images.  The just-past-full-moon here was captured in the latter part of October, on my first night using a new ASI1600MM!  I enjoyed being able to capture the full disc, which I hadn't been able to do with my ASI290MM!  It was captured in LRGB using FireCapture, aligned with PlanetarySystemLRGBAligner (there needs to be some sort of cosmic alignment for it to actually work), and then all other processing done in PixInsight.

The rest of these were all captured on one night in March.  I've never had quite as nice seeing as this!  They were all captured with the same equipment as above, except for it using an ASI290MM instead of the ASI1600MM, and I also used an 850nm IR-pass filter for luminance.  The crater close-ups were captured in IR with a 5x powermate, and the colour data was stolen from the prime focus images.  In terms of processing, I hadn't found the PlanetarySystemLRGBAligner at this point, and so everything was really painful to do in PixInsight.  FFTRegistration gave me very poor results, and didn't result in images with the same geometry at all, which was awkward.  I ended up doing deconvolution and sharpening, and then manual DynamicAlignment on all 20 channels.  The alignment still wasn't perfect, and I ended up performing a convolution on the colour channels to avoid nasty colour fringing on craters.

Now PlanetarySystemLRGBAligner worked out very nicely for my full moon image above.  My attempts before discovering it all had nasty red and green fringing on the craters.  The thing in general is that PlanetarySystemLRGBAligner isn't very well maintained, and it's not very reliable.  Does anyone have any alternative way of aligning mono lunar images?

 

[Sunshine]

These are some great images! I really like the variations in colours which seem to demonstrate different feature characteristics, well done!

[callisto]

The 1st image is stunning 

So are the others btw 

 

Mark

[Kon]

Excellent work, nice sharpness and the colours are bringing a new dimension to it!

[neil phillips]

The mosaic isnt showing any major misalignments. But thats something that often happens with lower focal length work. The opposite is true of higher focal length work. Misalignments are harder to sort out. I have no experience in aligning rgb work on lunar. As i have found one shot colour cameras do just as well. And with 3x  less work. There maybe some one who has a solution. But its not something i have ever done unfortunately cant advise. Apologies. I did notice at reduced size. the higher focal length shots. Look quite sharp, as is often the case. But at full size they do appear quite soft. Appearing like further sharpening might be beneficial. So i attempted that, ( you saved in png ) which helped.

First thing i noticed trying to sharpen a bit more in registax is Lots of lines appeared along the edges of the craters. I have seen this effect before often when drizzle is used. Or under sampled images resized a lot. I am not certain to be honest. But the effect is there when any attempt to sharpen using wavelets in registax is done. I certainly think the images could do with a little sharpening, full size. But look fine as is, at reduced size.

I did do some sharpening examples to show you what i mean. Not sure if the effect will be obviouse here. If not copy both images and put side by side. Blow them up. And you will see what i mean. First image is with sharpening showing the effect i mentioned of either under sampling and resizing. Or possibly instead some kind of drizzle effect. That the softer sharpening is hiding. Thats ok. But its not doing the final appearance any good. if you want high focal length work to be reasonably in sharp relief.  The second is with a frequency domain applied. To remove ( actually subtracts ) the noise.  Then i could re sharpen again. Producing a somewhat sharper view to the full size appearance, Which i think has helped. You may not agree. If so just ignore. But thats what i see. Another way to compare is open the 4th shot down high focal length shot. you posted. To full size. Then do the same with the bottom shot i posted. you can see that further sharpening is helping a lot with detail. The seeing does look quite good. And the data too well done

Just re read the whole post. The lines that are appearing after trying to sharpen more. Are caused by resizing the prime focus data. and blending it into the 5x data i think

 

Edited December 4, 2021 by neil phillips

[tombardier]

Wow, @neil phillips, thank you for your reply!  You've done a wonderful job with that sharpening! Thanks for all the effort you've put in to your replay, and playing with my image!

You're quite right about me using drizzle.  I used 3x drizzle on the IR-pass/luminance channels, and then went through an agonising process or deconvoluting and sharpening each of the non-drizzled colour channels to then manually align them (using PixInsight DynamicAlignment, so not exactly manual), before resampling the colour channels up to meet the (probably excessively large) luminance channel, and then convoluting them again to soften them up to avoid fringing artifacts on the craters!  The PlanetarySystemLRGBAligner should avoid that entirely, but it's highly idiosyncratic, and I've got other, seemingly decent data that it just cannot deal with at all.  It's also a pain because I have to load up my luminance channel, and then turn each of my colour channels in to a 16-bit RGB TIF before registering it to the luminance image, where the software decides to work with it as an 8-bit image, and then I have to load it up in PixInsight again extracting the colour channel I want, and changing the sampling format back to 32bit!  It's a real pain!

These 5 close-up panels haven't been taken at a high focal length, but at prime focus on my Skywatcher 254mm f/4.7 newtonian.  It's just that the ASI290MM has both a small sensor and small pixels.  The full moon shot is a single pane at the same focal length, but with the ASI1600MM!  The two crater close-ups at the very bottom were captured using the 5x powermate (and the 290MM).  The Bulisadus close-up has the colour data from the pane you've just been playing with.

Try as I might, I can't find the lines you're referring to in your first image.  Maybe I'm just bleary eyed right now.  I've been out with my binoculars tonight!  EDIT - Actually, I think I've found it.  I think I'd refer to that as "banding", like the image just below? It is rather unsightly, yes!  I think the TGVDenoise process in PixInsight smoothed it over in my image.  I did experiment a lot to find a setting that would help with this, but not remove any detail.  I'm not really sure what you mean about the frequency domain thing.  Is it a tool available in Registax? 

I wish I'd captured the rest of the gibbous moon on that particular night!  I don't think I've captured data as nice since.  In compariron, and I do really love my full-ish moon image, above, but it was just one of the first images I took with my ASI1600MM Pro, on the day I bought it, and the seeing was pretty average to be quite honest!  Can't wait to see what it's like on a good night!  In the mean time, I think I might try a re-process of this data again! 

This was my attempt at a mosaic of these images. 

Thanks again!

[neil phillips]

1 hour ago, tombardier said:

Wow, @neil phillips, thank you for your reply!  You've done a wonderful job with that sharpening! Thanks for all the effort you've put in to your replay, and playing with my image!

You're quite right about me using drizzle.  I used 3x drizzle on the IR-pass/luminance channels, and then went through an agonising process or deconvoluting and sharpening each of the non-drizzled colour channels to then manually align them (using PixInsight DynamicAlignment, so not exactly manual), before resampling the colour channels up to meet the (probably excessively large) luminance channel, and then convoluting them again to soften them up to avoid fringing artifacts on the craters!  The PlanetarySystemLRGBAligner should avoid that entirely, but it's highly idiosyncratic, and I've got other, seemingly decent data that it just cannot deal with at all.  It's also a pain because I have to load up my luminance channel, and then turn each of my colour channels in to a 16-bit RGB TIF before registering it to the luminance image, where the software decides to work with it as an 8-bit image, and then I have to load it up in PixInsight again extracting the colour channel I want, and changing the sampling format back to 32bit!  It's a real pain!

These 5 close-up panels haven't been taken at a high focal length, but at prime focus on my Skywatcher 254mm f/4.7 newtonian.  It's just that the ASI290MM has both a small sensor and small pixels.  The full moon shot is a single pane at the same focal length, but with the ASI1600MM!  The two crater close-ups at the very bottom were captured using the 5x powermate (and the 290MM).  The Bulisadus close-up has the colour data from the pane you've just been playing with.

Try as I might, I can't find the lines you're referring to in your first image.  Maybe I'm just bleary eyed right now.  I've been out with my binoculars tonight!  EDIT - Actually, I think I've found it.  I think I'd refer to that as "banding", like the image just below? It is rather unsightly, yes!  I think the TGVDenoise process in PixInsight smoothed it over in my image.  I did experiment a lot to find a setting that would help with this, but not remove any detail.  I'm not really sure what you mean about the frequency domain thing.  Is it a tool available in Registax? 

I wish I'd captured the rest of the gibbous moon on that particular night!  I don't think I've captured data as nice since.  In compariron, and I do really love my full-ish moon image, above, but it was just one of the first images I took with my ASI1600MM Pro, on the day I bought it, and the seeing was pretty average to be quite honest!  Can't wait to see what it's like on a good night!  In the mean time, I think I might try a re-process of this data again! 

This was my attempt at a mosaic of these images. 

Thanks again!

Yes you have to blow it up to see the lines. It was only a problem. when trying to sharpen more. Going forward. If your serious about colour work. I would highly recommend keeping things as simple as possible just using a colour camera Thats really all is needed for colour work. And your mono 290 for hi res mono work. Frequency domain was a tool i use on meesoft image analyzer. Recommended this recently. But it may have been dropped from the program ? Personally i think the art of lunar imaging is getting a razor sharp image that also looks natural. Not as easy as what a lot of people think. Same for colour work. Its easy to go way too far. Subtle colour and sharpening looks so much better. Its a balance that has to be learned i feel. 

Your data mosaic has some excellent parts. Very good sharpness (up from this post) Other areas are not blending as well. But even with average seeing. A whole mosaic thats even in sharpness and light. Can be got. Just bide your time. A good tip. is to adjust the histogram back to 70% on each section keeping everything looking the same. ICE is good for building the mosaic. Monitors change everything. you may not be seeing what i am seeing. I dont know how good your monitor is. But different monitors are a huge problem when we try and talk about images posted up. Anyway here is a example of me working for months on perfecting this kind of thing. I can do better i am sure. it was windy on this night . But you should be able to see the evenness of sharpness and light across the whole globe. I can tell you know a lot about processing. But try not to over complicate things. The mosaic i took here was, processed in AS/3 one wavelet sharpening in registax built in ICE and some final tweaking of levels and sharpness. in image analyzer. But anything will do photoshop gimp ect. I certainly talk a lot with guys who are on the road to great things. As i find it interesting. Hope i am helping.

 

Edited December 5, 2021 by neil phillips

[tombardier]

Thanks again for your amazing reply!  I would honestly love not to complicate anything.  The convoluted procedure I took with deconvoluting and sharpening all of the colour channels was purely so I could get them registered to one another.  Any subsequent processing was minimal (just an adjustment of the colour saturation and a TGVDenoise).  Where I spend the most time was in trying to find a good PSF for deconvolution and also sharpening without wanting to overdo it.  I think I can probably do better with this data, so I think I'll try it again.  I think I'm going to be stuck with the 3x drizzled luminance channels though, as I don't have the original video any more.

I do tend to think sharpening/colour saturation can be a matter of taste.  I happen to think I have quite refined tastes, but I'm sure there'll be people who feel nauseous, seeing the amount of colour in my images. I must say your image is "to my taste", most definitely!

Thanks again.  I'll tag you, if you don't mind, with the results of my second round of processing this data!  I might just start with this one panel we've been discussing.

[neil phillips]

2 hours ago, tombardier said:

Thanks again for your amazing reply!  I would honestly love not to complicate anything.  The convoluted procedure I took with deconvoluting and sharpening all of the colour channels was purely so I could get them registered to one another.  Any subsequent processing was minimal (just an adjustment of the colour saturation and a TGVDenoise).  Where I spend the most time was in trying to find a good PSF for deconvolution and also sharpening without wanting to overdo it.  I think I can probably do better with this data, so I think I'll try it again.  I think I'm going to be stuck with the 3x drizzled luminance channels though, as I don't have the original video any more.

I do tend to think sharpening/colour saturation can be a matter of taste.  I happen to think I have quite refined tastes, but I'm sure there'll be people who feel nauseous, seeing the amount of colour in my images. I must say your image is "to my taste", most definitely!

Thanks again.  I'll tag you, if you don't mind, with the results of my second round of processing this data!  I might just start with this one panel we've been discussing.

Tag away. Its great to talk lunar with like minded people trying to get the best images. You should check out westcoastcannuck. He has a great touch using deconvolution. And made it a art form. Has spent time working on the right PSF. thats somthing i could learn from others. Like you or him. But i admit. I like things to be simple. And have stuck with my procedures. But i know i should really be experimenting. Theres always a better way, or at the very least. A different way. Would like to see just your greyscal luminance data on its own. But anyway great chat and nice work Tombardier