345 replies to this topic

[Yuri_18]

Hi all,

I have done some tests with our materials and I can tell that as it seems to me AS2 works very well. Emil, thanks for the fine software.

[Freddy WILLEMS]

Hi Emil,
Just tried two Avi's Red and Green, and they are ok with NO drizzle !
Jupiter stays centered all the way, perfect..
And what ever alignment box I use works great, but them with no Drizzle.. So that problem is solved.
Now what I miss here Emil is the re-sample like 1.5 or 2 X like in Registax, that's why I used 1.5 drizzle to make the end result slightly bigger in scale then the original capture.
Is there a way the implement that in the software.
It's really fast Emil and everybody likes it and will use it eventually !! And much sharper..

[MvZ]

Are you sure it is caused by drizzle problems Freddy?? I know for a fact that an AP size of 50 on that kind of data is too low to work optimally, so the results should definately improved when you use bigger and better placed APs (like I described in my previous post).

You can easily test this by placing those 120 size APs on Jupiter, and then stack without the drizzling option, immediately followed by a stack where you turned the drizzling option ON. If you could show both stacks, that would be grrrreat.

Btw, I'm not adding a resampling method. That would be something for AS!3.

[Mert]

Hoi Emil,

First to start off with, I'm much impressed with the
results AS!2 offers without knowing anything.
Thank you very much for such a wonderfull tool you've
developed!!
After some reading and trying, I see that the stacking
is very nice.
One question I have is about the resulting image
I see there are 2 images being saved, one is the stack
without processing??? and the second with some sort of
wavelet scheme applied??

I compared 2 images I did, the first was processed with a
lot of steps, the second with AS!2 Specklm and CS2.
The funny thing I observe is that the AS!2 processed
version seems to have a very pixelized apearance.

Which is the "unprocessed" version saved? (is it
unprocessed only stacked??)

[PiotrM]

Here is my Mars comparison. The apps had a tricky thing to do as C11 had a temperature lag and from time to time the image to blurred slightly from tube currents. It seems that Both AS versions made good stacks:

http://i44.tinypic.com/2el5jsx.jpg

Bigger % stack gives less sharp image (AS deconv files)... so at 50% it stacks 50% of frames starting from the best ones?

[sfugardi]

Emil, thank you very much for the enormous amount of work you put into the latest AS! program. I have just started using it within the last 24hrs. The user panel is very nice. One interesting thing I noticed about the selection of AP sizes. If you pick a size that is too small, the program does not work, so I've tweak the size as low as possible until its starts. For my latest Mars I have it set to 65. Your program will take image processing to the next level for sure. Thanks again!

Regards,
Steve

[MvZ]

Hi Mert,

The (slightly) sharpened one has the "conv_" in the name, so the other one is the raw stack. The sharpened version is only for previewing (I often use it to scan through the results to find a good set of RGB data ), you shouldn't use it for processing. If you don't need the sharpened version, just turn off the 'sharpened images" option, and it will only save a raw stack.

Loverly images Piotr, looks like excellent seeing! The image sharpening routine built into AS!2 is very simple and does not take into account how many frames were stacked. More frames stacked pretty much always means that you also have to sharpen the image more to get a similar sharpness. This does not mean you shouldn't stack more frames though, it only means that you should adapt your processing to the amount of frames that you stacked. So again, the image sharpening in AS!2 it is only meant to provide a quick sharpened preview of the stack (I often just use the windows photo viewer to browse through all the "conv_" results to get an idea of what the data is like).

Steve. It's best to start off with APs that are a bit too large (like the size of Mars), and work your way down from there. Carefuly compare the results to see what works best for your images, after a while you'll get an idea of which settings you should use. The raw AP size number on it's own means nothing, it greatly depends on the image scale you image it, the noise level of the images, the seeing, etc. Simply put: keep experimenting with your data. You'll soon find that for Mars images recorded at F/34 an AP size of xx works well most of the time, but in case of the blue channel it might be best to use sligtly larger AP-sizes.

I often run all the red recordings in one go through AS!2, usually I can get away with the same settings for the green recordings, but for blue I often switch to slightly larger APs.

[PiotrM]

But just to be sure - if I set 50% for the stack to make - it will stack 50% of the frames starting from the best ones?

[MvZ]

Yes, of course.

When operating in MAP mode and when you use the gradient quality estimator, each frame uses the best 50% of the frames for that location (so the quality is then judged local to the AP, not on a global per frame level). When using the edge detector in MAP mode, each AP will use the exact same frames (because the edge quality detector will only provide a single estimation for the entire image).

A little trick:
A good way to find out how well a quality estimator works, is to stack only the top 5%, and play around with different quality estimator settings and carefuly analyze the stacked results. When stacking 50% of the frames it might be more difficult to tell the difference, but when only using the very best 5% of the frames any potential difference in image quality will be much easier to spot.

[Mert]

Thanks a lot Emil for your feedback, very useful!
This program for sure makes life much easier

[rumples riot]

Just tried this and this is my assessment. Fast processing, sharper detail and better contrast just with basic setting. I will present with two images one with Registax5 and one with AS. If this is anything to go by I will be using AS in the future.

It gets my thumbs up.

Attached Thumbnails

• 

[rumples riot]

now the one with Reg5

Attached Thumbnails

• 

[Freddy WILLEMS]

Hi Paul what size of alignment box did you use ?
Did you use drizzle ??

[rumples riot]

Did not use drizzle and the alignment box was 150.

[Sunspot]

Paul,

Just wait until you try it on Mars. I find the improvement better than when I use it on Jupiter. Of course I was using smaller AP's...might have to experiment with the larger ones. One thing I definitely see is the the images are less grainy than with Reg.

Paul

[rumples riot]

Hi Paul,

yes hoping to get some data on Mars this week, gully winds have been killing any chance of seeing though.

[GreatGigInTheSky]

I just want to report that AS2 works flawlessly under Wine in Linux, and to say thanks, Emil. This is great stuff.

[Freddy WILLEMS]

Hi Emil
It seems that it works great now.
Used the 150 alignment boxes (what you and Paul suggested) and no problem even drizzle with those 150 size boxes came out well !!

[MvZ]

Awesome

[Kokatha man]

.....now that you've drawn my attention more fully to the frames/scrollbar etc in your response to my current thread Emil, I've become a little confused by what I'm seeing/interpreting there - here's a screenshot where I selected 1200 frames from one of the poorer avi's (it was late in the night amd showed a different face to Mars)

As I read it from both the position of the scrollbar slider and the info in the planet image window, as well as the graph in the other (left hand) window, I'm only looking at a stack of 739 ot 9012 instead of the 1200 I wanted/thought I was specifying....?

What's happening here.....maybe I was blissfully ignorant in all the other processes but wasn't actually getting what I thought re frame stack numbers....?

Attached Thumbnails

• 

[Sunspot]

I was interpreting the "Output Images" setting checked to indicate that 1200 frames are being stacked. At least I was hoping that, so I'm interested in the answer as well.

Paul

[Kokatha man]

ps: my gut reaction is that I'm not interpreting the graph & figures correctly (something to do with where it's selecting all the optimum frames in the whole list..? ) because here's the R6 wavelets application window, and with my standard values I apply here this amount of denoise in slider #1 would not be enough for only 700-odd frames stacked.....

Attached Thumbnails

• 

[Kokatha man]

.....and for completeness re this question, here's the other display in AS!2 by clicking to display the other graph view display.....

Attached Thumbnails

• 

[MvZ]

The slider and selected frame have absolute no influence on the amount of frames stacked or anything else for that matter (e.g. you cannot select a reference frame manually as you can in Registax 6, as AS!2 makes its own reference frame). It is only to preview the frames. The actual amount of frames that will be stacked, is the number you entered in the box (or the percentage you entered in the box below that).

Try placing the APs more towards the center of Mars, and also add an additional AP or two right around the center of Mars (there is plenty of detail there to track on, so placing some additional APs near edges/corners of contrast areas makes sense). Also the extra APs around the poles seems a bit too small to add anything, so I would make those a bit bigger.

One more tip: crop the planet a bit more, it should make processing a bit faster (and it makes sure the program doesn't use that much RAM memory)

And another one for free: try using the edge quality estimator and compare the resulting stack to the gradient quality estimator stack. It often works nicely on Mars, and you might get a bit more sharpness.
Now I'm off to bed....

[Kokatha man]

.....thanks for takng this time Emil - and I appreciate we're on different sides of the planet re sleep etc..!

I was tending towards the response you gave, and yes, I thought about placing a couple in the middle and also increasing the size around the NPC and Hellas basin areas.....and another "yes" for edge quality estimation processing.....when I put the thread together for these 3 quite good nights of imaging I'll bundle the lot into this....!

Thanks once again for this software and your assistance - greatly apreciated....!