Reference Frame Selection - AutoStakkert2
Posted 29 April 2013 - 12:56 PM
Like Registax, I scan the video frames for a good quality reference frame for alignment points. But, in Registax 6, you can look at your reference frame in comparison with the frames that show up in the stack, having a better quality. (The first one in the order.) It's an extra step. But, I take the best quality frame from the stack, and re-run the video for a new stack and align.
With AS, I'm moving the slider, looking and scanning frames. Does AS do something similar to R6 that makes it easy to find best frame? If not, I suppose I can use R6 to find it and then run the AVI thru AS2. Any short cuts?
Posted 29 April 2013 - 02:10 PM
Posted 29 April 2013 - 02:49 PM
The positioning of the APs seems to matter a bit more in AutoStakkert than it ever did in Registax, but perhaps that's just due to me not really getting a good grasp on Registax before I switched to (the much faster and more intuitive, at least for me) AutoStakkert.
Posted 29 April 2013 - 10:21 PM
Ya, I have analyzed before APs are applied too, Grant. But, I've tried placing APs and analyzing after. I'm not sure which procedure is correct or better. I am confused by the "analyze" button and it's purpose. Darryl suggested a particular layout of the APs. He uses different sized boxes and multi-APs that spread over the planet in a consistent pattern. And, it totally works. And, I like it. It's repeatable rather than random.
Hope you guys get some great imaging opportunities, soon. Real soon!
Posted 30 April 2013 - 04:42 AM
If you do either it's surprising how little is changed re the % quality of the poorest image in whatever stacksize number you select, but of course there might be somewhat different stack permutations - but you'd have to quizz Emil re this...
Of course if you're presented with a particularly "dud" image when you drag & drop the avi into AS!2 it makes sense to do an initial "Analyse" to get something decent to use for your MAPs selections imho...Emil consistently makes a point about keeping in from the edges and lately I've found some MAPs selections can be double-edged swords: they can assist with the aforesaid Cassini as I demo'd in a recent animation here on CN, and I have also placed them around the NPZ's hex outline - but sometimes this causes nicks or splits in the NPZ and edge artifacts on Ring "A" which, by removing said MAPs boxes fixes the problem...
Posted 30 April 2013 - 07:25 AM
I wish I had more time to experiment on GOOD data, as opposed to the mush I've been handed so far this apparition...
Posted 01 May 2013 - 10:14 AM
"For Saturn, use manual alignment point placement. Diagonal lines on the rings are not good places for alignment points. Place the alignment point on the rings to include a perpendicular feature like the edge of the planet, or the black space at the tips of the rings."
Posted 01 May 2013 - 11:10 AM
Posted 02 May 2013 - 07:57 AM
No difference at all. You could select the worst frame and place the APs there, and it will give the exact same result (if the APs are placed in the same location).
The reference frame in AS!2 is actually a stack made from the best frames. The bigger the stack, the better the general shape of the object is (just one frame is always a bit distorted, but averaging many frames gives you a good shape of the target and reduce noise both of which are important). A bit blurry reference stack is usually no problem.
Ya, I have analyzed before APs are applied too, Grant. But, I've tried placing APs and analyzing after. I'm not sure which procedure is correct or better. I am confused by the "analyze" button and it's purpose.
For planetary images it doesn't really have a purpose but calculate the quality of the frames and give you an indication about this quality (it generates the quality graph, and orders the frames from high to low). If you switch quality settings, you can analyse again to see if that changes anything (usually the quality graphs are similar between different scales, but especially the smallest gradient 2 can pick up on noise instead of actual details easily).
The order of the frames after analyzing does NOT depend on APs that you might have set, it just shows the overall quality of the frame (as if the entire frame had a huge AP placed onto it!). Each AP actually uses it's own subset of frames anyways, so I decided to just show the global frame quality here (which still gives a decent indication of how many frames you might want to stack).
But for placing APs for planetary recordings it doesn't matter if you do that before or after analyse. If I know I want to stack 50% of the frames with gradient 4 and I'm not interested in checking the ordered frames, I just press stack right away and it takes care of the analysis phase 'behind the scenes'.
For Surface recordings you can only place the APs AFTER analyzing. This is because it will also calculate the maximum area of the recording where the APs can be placed without a problem (it depends on the motion in the recording, if there is a lot of motion, only a small FOV is available to place APs on!)
About the APs on Saturn: try to avoid too small APs placed on the rings EXCEPT on stable locations (right at the edge of the Cassini division for example, or where the rings intersect the planet).
Posted 02 May 2013 - 09:34 AM
Posted 02 May 2013 - 10:48 AM
Posted 02 May 2013 - 11:05 AM
The AP size might be good around the strongest locations (intersection of the rings with planet, and right at the edge of the rings where the Cassini division can be best seen), but the planet definitely needs a few large APs.
For Saturn I use something like this: http://www.astrokraa...info/sat_ap.jpg
But just try out different settings, and CAREFULLY compare the results by applying exact same processing settings. Choose APs that are obviously too small or really large, and you get a sense of the effects are.
Posted 02 May 2013 - 11:16 AM
Your AP selection pattern is similar to what I used to do, Paul. I changed to a different pattern after corresponding with Darryl. But, a slight variation of my own. (I have a feeling we are opening flood gate for screenshots of APs )
Posted 02 May 2013 - 11:59 AM
PS: I finally stop pre-castrating the videos before AS2
Posted 02 May 2013 - 01:29 PM
Posted 02 May 2013 - 01:42 PM
It basically influences the scale of the quality estimator. It ranges from 2 (high quality, low noise, fine details) to 6 (very noisy, coarse scale, details barely visible). 4 usually works just fine. I would use 2 for low magnification and very high quality solar or lunar images, 6 for very noisy images of Saturn in blue light at high magnification.
Posted 02 May 2013 - 06:55 PM
Posted 05 May 2013 - 02:39 PM
I just did a little test to compare between alignment point selections.
I have recently been doing something similar to what Steve had above, using about 30 alignment points, but at 50px size. I just did a stack of the same image using a total of 10 APs. They were similar to what Emil has in his link. The APs at the end of the rings were 100px and the rest were 70px.
Below is the result. The processing was the same on the two (4000 frames stacked out of 9000) and the same wavelets. The wavelets are a bit harsh and I dont like the way the rings look, but the settings show some of the cloud structures a little better.
It may be a little tough to see in this animated .gif, but the run with only 10APs, it looks like the cloud structres are a slight bit more well defined and brighter (look at the bright clouds at meridian about 2/3 the way up)