Jump to content


Photo

AutoStakkert! 2

  • Please log in to reply
253 replies to this topic

#151 MvZ

MvZ

    Surveyor 1

  • *****
  • Posts: 1579
  • Joined: 03 Apr 2007
  • Loc: The Netherlands

Posted 31 August 2012 - 03:02 PM

Could you send me a huffyuv sample?

Steve, could you send me an mjpeg sample video?

AS!2 will work with uncompressed videos, so if you can select that during converting it should work fine (if you are going to use another lossy codec to convert your MOV, then you will lose details again, so that is not recommended). The other option would be to let virtualdub recode one of the other formats (again using uncompressed formats. Make sure to select either rgb24 for color recordings, or y8 for grayscale recording).

AS!2 will only work with uncompressed formats (and the exception to this rule in the new version will be MJPEG. If huffyuv turns out to be easy to implement, I will also do that, but to be honest I think it will turn out to be too time consuming)

#152 zAmbonii

zAmbonii

    Vanguard

  • -----
  • Posts: 2420
  • Joined: 19 Jan 2008
  • Loc: Ypsilanti, MI

Posted 31 August 2012 - 08:05 PM

How long of a clip would you want? I'm out of town at the moment so it may be a couple of days.

One of the reasons why I have hesitated switching to using AS was the lack of HuffYUV support. R6 handles the files well, but it has gotten to the point where I was getting tired of seams in my R6 images. Really was a cr*pshoot in getting decent images. I honestly have had no problems at all with AS with MAP alignment.

Almost forgot, I could use the Lagarith codec. not sure if it would be easier to get to work. I just have been using HuffYUV because I had never heard of Lagarith loseless codec until recently.

#153 MvZ

MvZ

    Surveyor 1

  • *****
  • Posts: 1579
  • Joined: 03 Apr 2007
  • Loc: The Netherlands

Posted 03 September 2012 - 07:04 AM

Sorry for the delay.

The clip does not have to be long at all, 10 frames would be enough (the contents doesn't matter)

#154 lcd1080

lcd1080

    Viking 1

  • -----
  • Posts: 999
  • Joined: 03 Jun 2010
  • Loc: Maryland, USA

Posted 03 September 2012 - 09:28 AM

Emil would it be possible to add a feature that permits viewing of .SER files as a full motion video? I know that one can move the slider in Registax and manually view the individual frames of the .SER file in motion but one can't review the video at a constant number of frames per second (at least I've never seen that feature). I'd like to do that because one could get a better idea of how good of a capture the .SER file is before deciding whether to process it in AS!2.

Thanks,
Pete

#155 shawnhar

shawnhar

    Fly Me to the Moon

  • *****
  • Posts: 5475
  • Joined: 25 Jun 2010
  • Loc: Knoxville, TN

Posted 10 September 2012 - 12:59 PM

The other option would be to let virtualdub recode one of the other formats (again using uncompressed formats. Make sure to select either rgb24 for color recordings, or y8 for grayscale recording).

Can someone please help me figure out what I am doing wrong? I am trying to convert a vid made with EOS Mov Rec to an uncompressed avi using Virtualdub64 so I can try AutoStakkert but all I get are horrible onion rings when I apply wavelets, the video does this in AS2 and Registax.
I have set the default color depth to RGB 24 in both the decompression and output, tried auto detect, that didn't work either. Compression set to Uncompressed.
I can stack the original avi in Registax and it has no onion rings, just seams, that's why I wanted to try this program.
Am I missing something? Just "Save as AVI" right?

#156 MvZ

MvZ

    Surveyor 1

  • *****
  • Posts: 1579
  • Joined: 03 Apr 2007
  • Loc: The Netherlands

Posted 11 September 2012 - 05:55 AM

First of all, I often have trouble with VirtualDub64, so I just stick with the 32-bit version (1.9.11).

- open the video
- under video select full processing mode
- under video select compression, and then select (Uncompressed RGB/YCbCr)
- under video select video color depth. Set the decompression format to Autoselect, and set the output format to either Luminance only Y8 for black and white camera's, or 24 bit RGB (888) for color recordings.
- under file select save as avi, and give it an appropriate name (add Y8 or RGB to the filename for example).

You should be able to open these files in AutoStakkert!2, regardless of any codecs you have installed on your system.

Please let me know how this goes. If it doesn't work, send me a sample that fails to open (you can actually select and cut out a range of frames using virtualdub using the slider and the dark half-arrow icons at the bottom of the screen. Set a start and end point, and the selection turns blue. If you now save to avi, only the selection is saved. If this file doesn't work in AS!2, you can send it to me, as it will be much smaller than the full file).

#157 shawnhar

shawnhar

    Fly Me to the Moon

  • *****
  • Posts: 5475
  • Joined: 25 Jun 2010
  • Loc: Knoxville, TN

Posted 11 September 2012 - 08:55 AM

Thank Emil!
I followed your instructions exactly and had the same result. The new avi's from Vdub will open, but s soon as the wavelets are applied ringing appears, as if the file is 8 bit color.
I deleted the 64 bit Vdub version and downloaded the 32bit, the result is MUCH better, but there are still onion layers after stacking and applying wavelets.
I am sure it is a problem with my data or Vdub, and not AS!2.
I just cannot seem to git rid of the rings, is there another program I could try to convert to uncompressed avi?

Attached Files



#158 MvZ

MvZ

    Surveyor 1

  • *****
  • Posts: 1579
  • Joined: 03 Apr 2007
  • Loc: The Netherlands

Posted 11 September 2012 - 09:47 AM

Could you perhaps send me the original avi file so I can have a look?

Pete, couldn't you just use the analyze function first and quickly browse through the quality sorted frames using the frame slider afterwards ? It doesn't take too long to do the frame analysis right?

I could also add some kind of play button next to the frame slider, wouldn't be too difficult.

#159 MvZ

MvZ

    Surveyor 1

  • *****
  • Posts: 1579
  • Joined: 03 Apr 2007
  • Loc: The Netherlands

Posted 22 November 2012 - 10:36 AM

question from Tapio

Emil,
[...]would like to know what method is used if you select the "sharpened stack option" ?
Wavelets ?


Nope, just some custom unsharp masking (and mixing in of the original stack).

#160 Andy Taylor

Andy Taylor

    Twisted, but in a Good Way

  • *****
  • Posts: 1873
  • Joined: 24 Sep 2008
  • Loc: Epsom - UK

Posted 22 November 2012 - 02:00 PM

The "onion ring" artifacts are usually caused by underexposure.

When imaging Jupiter there is a brightness gradient at the limb. This causes wavelets to produce the onion rings.

Try bracketing with different gain levels. Back off the gain slightly when you just see the brightest areas loose detail.

Also, when shooting drop your gamma to minimum.

#161 MvZ

MvZ

    Surveyor 1

  • *****
  • Posts: 1579
  • Joined: 03 Apr 2007
  • Loc: The Netherlands

Posted 22 November 2012 - 03:16 PM

> Also, when shooting drop your gamma to minimum.
If by minimum you mean neutral, then yes, but many camera's can go in either direction of the neutral setting.

#162 Andy Taylor

Andy Taylor

    Twisted, but in a Good Way

  • *****
  • Posts: 1873
  • Joined: 24 Sep 2008
  • Loc: Epsom - UK

Posted 22 November 2012 - 04:09 PM

> Also, when shooting drop your gamma to minimum.
If by minimum you mean neutral, then yes, but many camera's can go in either direction of the neutral setting.


That's what I mean - depends on camera/driver though.

The "onion ring" effect is caused at time of capture - nothing to do with ASII.

Great software by the way. :bow:

Any chance of putting it all on one panel? There seems to be a lot of "white space" on the settings window. This could be compacted to give a full window for us with lesser screens.

I keep having to move/re-size the AP window...

#163 MvZ

MvZ

    Surveyor 1

  • *****
  • Posts: 1579
  • Joined: 03 Apr 2007
  • Loc: The Netherlands

Posted 22 November 2012 - 04:23 PM

Come up with a good layout (paint/photoshop using screenshots of the current layout), and I'll think about it ;)

#164 MvZ

MvZ

    Surveyor 1

  • *****
  • Posts: 1579
  • Joined: 03 Apr 2007
  • Loc: The Netherlands

Posted 26 November 2012 - 08:17 AM

Hello all. I just opened up a Yahoo discussion group for AutoStakkert!2

So if you have interesting new ideas that might be worthwhile to implement, or if you run into problems, have no idea which alignment point size to use or the quality settings, what the effect of drizzling is, or any other setting you don’t know how to optimally use, you can post your questions in this yahoo group as well.

http://tech.groups.y...p/autostakkert/

By the way, this does not mean I will ignore emails I receive any more than I already do, and I will also keep checking out this forum for questions, but I hope this new yahoo group turns out to be another valuable way of providing support to you.

#165 Baron

Baron

    Messenger

  • -----
  • Posts: 461
  • Joined: 23 Nov 2012
  • Loc: Rome, Italy

Posted 26 November 2012 - 08:41 AM

New september release (planned for september 22)


news for this version?

#166 dobsoscope

dobsoscope

    Messenger

  • *****
  • Posts: 409
  • Joined: 24 May 2006

Posted 26 November 2012 - 09:17 AM

well done Emil.!

I do not have a planetary cam yet, but would it be possible for anyone to send me RGB videos for me to try out please using this program?

thank you

#167 Sunspot

Sunspot

    Fly Me to the Moon

  • *****
  • Posts: 5547
  • Joined: 15 Mar 2005
  • Loc: Surprise, AZ

Posted 26 November 2012 - 09:54 AM

Excellent! I just joined.

Paul

#168 Mert

Mert

    Skylab

  • *****
  • Posts: 4222
  • Joined: 31 Aug 2005
  • Loc: Spain, Pamplona

Posted 26 November 2012 - 06:20 PM

I'm with you Paul, also joined the group!

#169 runrob

runrob

    Lift Off

  • *****
  • Posts: 17
  • Joined: 02 Apr 2010

Posted 03 December 2012 - 08:03 PM

Hello Emil: A great program but all of a sudden I have had a glitch develop and have attached an imaging attempt of Jupiter. The program was running fine using WinXP on my laptop and then all of a sudden this problem developed in my results. Is this an operator error or something that has happend with the AutoStakkert 2 program? Bob R.

Attached Files



#170 MvZ

MvZ

    Surveyor 1

  • *****
  • Posts: 1579
  • Joined: 03 Apr 2007
  • Loc: The Netherlands

Posted 04 December 2012 - 06:13 AM

I'm hoping it is an operator error. I answered in the yahoo-group. For those interested I copied my answer here, feel free to add comments.

--------

As far as I know, the program does not simply become corrupt, so it must either be an operator error, an error in the avi recording, and/or a bug in the program that didn't show itself because the conditions weren't right for it.

My best guess would be to run AS!2 again on the Jupiter recording, but pay very close attention to the settings you used.

It is not often that seams like this occur, usually it comes down to just a couple of settings that influence the alignment of the different points. If AS!2 for whatever reason looses track of an AP, these things can happen. Another reason is, and this to me appears to be the most likely cause here, is that Jupiter ran off the screen and wasn't fully visible?

First, visually inspect the result of the image stabilization function. After selecting planet-mode for Jupiter, go to the frame view window and drag the slider at the top of the screen from left to right to see if the alignment is good.

Then crop the width and height down significantly to speed up the processing. Jupiter is relatively small compared to all the black space around it, there does not seem to be a reason to keep all this black space here. It might also help to reduce artifacts.

Set the quality estimator to 4, and do the same visual inspection after pressing the analyze button. Is there perhaps a deformed frame in front of the bunch?

If everything was fine up to now, start with placing just one relatively large AP (still in MAP mode) and see what the stacked result looks like. Any better? Try placing more APs (manually), for now just use an AP size at least half the size of Jupiter (width and height). You can use the left mouse button to place an AP, and the right to remove one. Using the mouse wheel will change the size of an AP that is going to be placed next.

If you followed these steps carefully, and it still doesn't look good, you should try and fine a way to upload the recording so we can have a look.

#171 tonyb

tonyb

    Viking 1

  • *****
  • Posts: 706
  • Joined: 27 Jul 2005
  • Loc: WA, USA

Posted 16 January 2013 - 09:44 AM

Are there any guidelines for how many # of frames to select or the % for stacking? I'm currently selecting # of frames above 50% in the quality estimator graph. Thx!

#172 MvZ

MvZ

    Surveyor 1

  • *****
  • Posts: 1579
  • Joined: 03 Apr 2007
  • Loc: The Netherlands

Posted 16 January 2013 - 10:22 AM

The best guideline for now is to not follow any strict guidelines, but try things out. AS!2 is fast, so you can make use of that by playing around with different stack-sizes and processing methods. Use what works for your images, under those specific conditions with that particular image scale and camera settings.

There are a couple of other things to keep in mind:
- Stacking 50% vs 60% of the frames does not make a lot of difference. Doubling the amount of frames that you stack does however, and that is when you see big differences in the resulting stack. So 5% versus 10% is a huge difference. You'll probably not even notice 25% vs 30%.
- You can't really mess up an image if you include a couple of less sharp frames. 5% of blurry frames versus 95% sharp frames will still give you a 95% sharp result.
- Processing needs to be changed depending on the amount of frames you stack (or better: on what each particular stack can handle)
- Bright and high contrast recordings (often lunar and sun recordings) usually require less frames to stack. Jupiter is full of details, but the details have pretty low contrast, and are thus easily lost in noise. So for Jupiter you normally need a lot of frames, unless you have loads of signal to work with.

The green quality graph MIGHT help a little bit, unless there is one really good or really poor frame which will bias the graph (so then the 50% line is not really useful). The grey quality line is probably a bit more indicative: if it is pretty much constant throughout the recording, never really going up or down for longer amounts of time and without a rising or slowing trend: you can probably stack a lot of frames.

But the most important thing is to try things out a lot, look at the frames, and carefully process and compare results. Play around with it. You'll probably come to some kind of agreement for each target you image. I pretty much always stack 60% of my Jupiter frames taken with the 10" telescope. With the 16" I also played around with less frames, like 30-40%. For solar recordings I only stack 150 frames (because I always worked with really low noise recordings and I know that this is about the lowest amount of frames I can stack when no matter what processing I throw at them, I get clean results).

Guidelines usually lead to sub-optimal results, especially when things change (different camera, different settings, transparency, etc etc...)

#173 tonyb

tonyb

    Viking 1

  • *****
  • Posts: 706
  • Joined: 27 Jul 2005
  • Loc: WA, USA

Posted 16 January 2013 - 11:21 AM

Many thanks - great information.
Another question is alignment point selection - if I understand the tutorial on the AutoStakkert site, it is best to select alignment points away from Jupiter's edge. the screenshot in the tutorial shows automatic aligment points with several points right on Jupiter's edge however.

#174 MvZ

MvZ

    Surveyor 1

  • *****
  • Posts: 1579
  • Joined: 03 Apr 2007
  • Loc: The Netherlands

Posted 16 January 2013 - 11:44 AM

That is correct. I think the image is a bit misleading. It's not like it will go wrong if you place them like that, but the chance that it goes wrong is much higher than if you manually place the APs away from the edges.

I should probably replace the image.

#175 tonyb

tonyb

    Viking 1

  • *****
  • Posts: 706
  • Joined: 27 Jul 2005
  • Loc: WA, USA

Posted 16 January 2013 - 11:56 AM

Many thanks and yes it might make sense to replace the image and show manually placed APs in that Jupiter example. Thanks again.






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics