345 replies to this topic

[Kokatha man]

Interesting comments Emil: strangely, I'm finding Ninox isn't handling my Mars capture so far at all well and I've put that down to the peculiar type of seeing myself and another AA'er here in South Australia have noticed a lot so far this apparition.....MAP just isn't producing acceptable results in R6 using Gradient2 and I've had to resort to C of G alignments - I'm still Ninoxing but loading all the bmp's into R6 because regardless of MAP or C of G Registax is determining quite different quality selections, and am really only using Ninox to speed the R6 process up by virtue of the cropping & centring functions.

Contrary to one of your comments above, I haven't had any problems using Ninox with the larger targets in recent time, such as Saturn at present and Jupiter a month or so ago.....it's little Mars that is troubling Ninox...and the seeing between Mars and Saturn shoots has been pretty much the same so that doesn't seem to be a differentiating factor for those 2 at least...

But I'm about to give AS a go in a few minutes, and the points you've already made plus those in your last post about SAP & MAP, and particularly AP's in general are good to know beforehand - will report back when I think I have something to comment further about..!

[MvZ]

In my comments I didn't actually take into account how well the quality estimator of Ninox works.

I have found that for larger targets a gradient quality estimator works well (and Ninox uses a gradient quality estimator, althought it's not quite the same as is used in AS!2). For smaller and brighter targets a gradient quality estimator might not work that well, image distortions can sometimes be seen as 'extra detail', I have seen this happen in particular on Venus and Mars recordings. But that is where the Edge quality estimator could be used, for bright targets edge sharpness is much more robust to those seeing distortions as it only looks at the sharpest transitions from each direction, and you can't get much sharper transitions than going form the dark background to a well-defined planet.

[Kokatha man]

Hmmmmm...Well, image distorting is a major element of the weird seeing effects we've been experiencing here, and AS edge alignment might prove beneficial thereon (I haven't done those trials yet - sleep (or lack of it!) got the better of me last night...)

But don't forget that Ninox does offer several other command line switches/function variations such as altering the number of downsampled images it takes it's particular form of Gradient2 quality estimation from and "-dbf=planet" which is for discarding badly-formed frames (both of which have no impact with the particular problem I've encountered of late alas! )

Also there's the morphing switches/commands.....hmmm, maybe they are actually "anti-morphing" controls and these are my next stage of inclusion in the Ninox-regulated processing comparisons which should be stacked up against AutoStakkert!'s performance in various modes - time to get to work but I have to confess that pre-processing as well as processing with my current regimens makes for a super-workload (why I haven't tried the morph switches in Ninox to date - Registaxing for the morphing sample, then Ninoxing, then final Registaxing!) so I hope I find some genuine satisfaction/advantages in your program Emil: a pity wavelet sharpening isn't part of your software, but if it really does run very fast then that's a real bonus.....as you commented earlier, one can go on a merry-go-round with all this trialling and analysing - but then again you programmers most probably think that's a wondrous pastime..!

[Sunspot]

Emil,

I'm getting a much better handle on the personality of the program. One part that continues to frustrate me is the lack of file management. I would love to give the final stacked image my own naming convention and place them where they work best for me. Not a complaint, but a suggestion. Otherwise...onward and upward.

Paul

[Asimov]

As Paul said for me as well however I'm already used to that & so I plan ahead a bit & place the original capture in it's own folder rather than have 20 odd captures in one folder. Works for me anyway.

Great program!!

[swalker]

Just an aside, consider donating a little something to Emil if you like AutoStakkert! 2; It is a labor of love by Emil, but a little reward for a job well done goes a long way in encouraging additional features and tweaks.

[DesertRat]

I'll second Seans motion. The amount of work involved in such an undertaking should be rewarded by the imaging community.

Glenn

[Sunspot]

Ditto! Then I won't feel so bad about making suggestions...

Of course, that works both ways...I'd figure on seeing an article about the software in S&T after the final release comes out... (hint).

Paul

[MvZ]

Thanks for the kind words. Of course donations are appreciated, but remember that you don't HAVE to donate. You can use the software or make suggestions either way

Whether or not I do something with those suggestions depends on 1) my mood, 2) if I'm convinced it makes sense, 3) how time consuming it will be to implement.

-- 2.0.0.46 is online. List of changes: ---

- Changed name of Coarse Alignment to Image Stabilization, which makes a bit more sense.
- Added option to manually set coarse alignment window location in surface mode. You should set it around a feature that stays in the FOV at all times. The default location works 95% of the time, but sometimes AS!2 surface alignment loses track because there is nothing interesting to see in the center of the screen.
- AS!2 is now aware of OS limitations on amount of memory available per process. There should be no more lockups caused by out of memory on 32-bit OS.
- Fixed caption of frame number at bottom left, it didn’t update properly in batch processing.
- Fixed caption of frame view in batch processing.
- Fixed unable to press Cancel button in batch processing.
- Added some extra lines at the back of the quality graph at 0, 25, 75 and 100% to increase readability of this graph.
- Tiffs are now saved uncompressed (instead of LZW). Should increase compatibility with processing software.
- Added support for 8-bit single channel bmp files.
- Fixed surface alignment bug causing a more or less random lock up during buffering or image alignment.
- Should be more stable now, especially for surface recordings containing poor frames.
- Automatically turn off bad frames with horizontal or vertical shifting artifacts.
- Speed increase for surface alignment (approximately 30-50% faster).
- Introduced two Surface options: ‘Expand’ will try to make the very biggest image stack possible, the edges will contain less frames (this was the default option). ’100%’ will crop the image such that each pixel will contain the same amount of data (the edges should be fine).
- Batch processing for surface recordings. When more than one surface recording was opened, when processing, for each frame all APs are replaced by a set of automatically placed APs (in a grid, just like when you press the grid button).

[Sunspot]

You HAVE been busy! Now it's time to try the newest beta and have some fun!

Paul

[Sunspot]

Well, your results may vary, but I'm quickly becoming a believer! I've attached my latest comparison the first image using the older beta. This is the same image I used in my "unbiased comparison" thread. The image on the right is using Emil's latest release, as well as the experience I've gained by using the program and from Emil himself. I'm not saying the processing is identical, the latest adds more experience. But, if this is any indication for the evolution of Autostakkert...I'm really impressed! Still hoping for more flexible file management, but can't kick about the results.

Two questions:

I had dynamic background checked, noise robust 3 and hand selected 13 APs with the 50 box. I'm not sure (and this may be mentioned elsewhere) what dynamic background does and if it is necessary or not.

I had normalized stack 75% checked and not sure what that (or sharpened images box) does. What's your take on that.

Anyway, pretty exciting advance!

Paul

Attached Thumbnails

• 

[Hillbrad]

Wow Paul that's a great improvement. Looks similar to my gains from Autosakkert 1 to 2. I will be trying out the new beta version this week as well. Thanks Emil for such a great tool.

[Freddy WILLEMS]

Great stuff Emil, I'm trying it now and sometime I have some kind of 'seems artifacts' is this because of bad seeing or stacking ?
I have this on Jupiter, 15 alignment point set inside the planet, not the edges..
What are the best settings to avoid stacking lines on Jupiter? I only put the alignment boxes on the NEB and SEB no edges.

Somebody else have this with Jupiter ?

[MvZ]

I didn't change anything in the alignment routines, so I'm guessing most of the increase in image quality that Paul is seeing (ha, seeing), comes from a difference in the placement and size of the alignment points (so make sure to play around with that a bit, after a while you'll get the hang of it).

For planets you should pretty much always keep the dynamic background setting checked. It is used to automaticaly determine the brightness of the black space, which is a critical step for keeping the planet steady in AS!2 (so the image stabilization is not thrown off by a noisy background with perhaps even a changing brightness when you are processing daytime recordings).

The normalize stack option uses an automaticaly calculated guess of the background, and stretches the image stack in such a way that the maximum brightness is always at 75%. I found this to be very helpful when you are imaging under varying transparancy or during dusk or dawn to keep all the R, G and B channels with a similar brightness both at the dark and bright extremes. I also often play around with different capturing settings, and when I use the normalize stack option each stack comes out with a similar brightness. This makes it much easier to make animations.

Freddy, can you show me an example? AS!2 is virtually seem free if you cover the entire planet with APs (place your APs on Jupiter like in the second post of this thread)

[Sunspot]

Emil, \

I agree that the improvement mostly comes from a better placement of APs. Part of learning the program. Sure is fun to work with good data.

Paul

[Hillbrad]

I second that..it took me a little practice but once you get the hang of the APs I haven't got any seems on Jupiter or Mars. Haven't tried Saturn yet but I'm sure it's the same.

[Kokatha man]

Well, in the spirit of sharing fellas, let's see a few screenshots from ya's to demo your latest (say) Mars MAPS positions..!

I've been handicapped by pc glitches (which now seem sorted) and actually had to retrieve avi's from my laptop using Pandora Recovery for the trials.....

But I have to say that once I got a very basic hang of the operations (and for this I had to resort to looking at your videos for the much earlier version of AS! Emil) I am pretty darn impressed..!

Lawd knows what I first did (took about 3/4 hour for AS! to process..! ) but after that I was quite astounded.....loading avi's into it and it whirring through in a minute or two (although I do use an 8-thread machine )

I'm quite impressed with what seems like a substantial jump in image quality output - and I'm still right at the elementary stage of getting to know the software.....I just placed 4 overlapping MAP boxes (size=80) centred around the limb peripheries and a smaller (40) one encompasssing the NPC but I'll retry the "Edge" method also and play with the boxes a bit more and see what transpires.....

A couple of things to comment upon - being a lazy son-of-a-gun who uses his handpiece to correct whilst capturing (I'm a travellin' imager most of the time ) I'm finding that there is that slight difficulty with RGB alignment that I'd also get if I didn't Ninox (crop & centre) in my prior processing regimens.....would Castrator used before AS! assist therein as Ninox used to....?

Also doubly-impressed 'cos even though I only used 5 alignment boxes and the Mars avi I used isn't fantastic.....with R6 I couldn't use MAPs without ending up with seams - and had to resort to CofG alignment with selfsame Ninox'd avi's..!

I took to assigning a folder/directory to each avi prior to AS! so that it deposited the stacks (with prefixes) where I knew/wanted them, so that wasn't any real problem.

Pretty sure I'm going to be a convert after only very brief elementary play-around.....lawdy, all that preparatory work with Vdub and Ninox and then Reggie chugging away.....I'd almost be prepared to kiss you Emil - if your chickens wouldn't get jealous..! :rofl5:

[Freddy WILLEMS]

Hoi Emil,
Tied again with the same alignment settings as in your example and after sharpening in Registax 6 I had this result.
Nobody else has this ?

Attached Thumbnails

• 

[MvZ]

> I'm finding that there is that slight difficulty with RGB alignment that I'd also get if I didn't Ninox (crop & centre) in my prior processing regimens.....would Castrator used before AS! assist therein as Ninox used to...

Not really, AS!2 uses the same centering method as Castrator does, and applying it once or ten times shouldn't make a difference. A COG measurement of the blue channel is often shifted towards the polar cap a bit (because compared to the rest of the planet, the polar cap is much more bright in blue than in red light). I think this, and other such differences in brightness regions, might cause a slight mis-alignment between the channels.

Freddy, those are seems alright. If you open that recording in AS!2, which Image Stabilization settings do you use (dynamic background?), and if you browse through the recording (using the slider at the top of the image window), does the planet stay perfectly centered throughout all frames?

How big are the APs that you use, and how are they placed?

Could you provide a screenshot of both of the AS!2 windows?

[Freddy WILLEMS]

I will emil,
Just for now, those where ninoxed frames, BMP's.
I did not have a good Avi' so I did not load the Avi, but the ninoxed frames like around 7000 into AS2.
I will get that screenshot.

[Freddy WILLEMS]

Hi Emil.
This is a recent processed Jupiter Avi,
No ninox, just straight the Avi into AS.

Attached Thumbnails

• 

[Freddy WILLEMS]

Wavelets into Registax 6
Let me know what I'm doing wrong..
I love the program it looks better then R6.. but only have those **** seems,
Also Processing my Mars images from this morning and have the same seems. Use different computers..

Attached Thumbnails

• 

[PiotrM]

Try maybe single point. Multipoint sometimes goes crazy (like in R6).

[Freddy WILLEMS]

Thanks Rik, will try, have the whole weekend..
Just want to add that I do not have any problems with my moon images processed in Surface mode.

[MvZ]

No, there is no need to switch to single AP just yet, we are not running away from the problem (although you are right that fever things can go wrong a single AP placed around the entire planet, you can't get a more stable AP than this, and the image quality is always at least decent)

You should do all these things Freddy, and the image should improve.
- Use the width and height sliders to crop the data to make sure you don't run into memory problems. AS!2 loves to buffer, but I see your pc does not have a lot of RAM available (450mb free is not a lot these days) so buffering to increase processing speed is out of the question. Set a width of around 320 and a height of 300.
- an AP size of 50 is WAY too low for this data, AS!2 simply won't be able to align the data in those tiny windows. Try again with an AP size of 120, and place overlapping APs ALL OVER THE PLANET (there shouldn't be any gaps there like you have in the center of the planet. Where there is overlap, there is backup for a potential troublesome AP - AS!2 will sort that out automaticaly)
- increase the gradient estimator to 4
- Don't use drizzle. Only experiment with drizzle when you know the data is undersampled and of high quality, otherwise just stay away from this setting for now.

Please report back with the results.

Edit: just to repeat one question to make sure the image stabilization works: "if you browse through the recording (using the slider at the top of the image window, just drag it along with the mouse), does the planet stay perfectly centered throughout all frames?"