Mars - Derotation or not?

[CraigT82]

Following some advice and calculations on rotational smearing from @vlaiv,  I decided to do some testing on the processing of Mars images, specifically whether or not Winjupos derotation of long runs is worthwhile, here are some results.

 

From left to right  we have...

Column A = a single 2 minute capture, best 30% stacked 

Column B = 5x 2 minute runs stacked together giving 10 minutes total. No derotation.

Column C = 18x 2 minute runs derotated in winjupos.

 

From top to bottom we have...

Row 1 = Wavelets applied after stacking. No other processing

Row 2 = Wavelets applied after stacking, plus some additional unsharp mask in gimp

Row 3 = Wavelets applied after stacking, plus some additional LR deconvolution

 

Presented without further comment!

 

 

Edited September 16, 2020 by CraigT82

[Tommohawk]

I'm not seeing much differences between any of these! But not on home screen at the moment.

One query. When you join 5x2 min vids, do you then use same percentage of total frames for the stack? 

I meant to add though that having read about issues with WinJupos, and AS2 ability to deal with it to some degree, I've stopped doing derotation

Edited September 16, 2020 by Tommohawk
added comment

[CraigT82]

27 minutes ago, Tommohawk said:

I'm not seeing much differences between any of these! But not on home screen at the moment.

One query. When you join 5x2 min vids, do you then use same percentage of total frames for the stack? 

I meant to add though that having read about issues with WinJupos, and AS2 ability to deal with it to some degree, I've stopped doing derotation

I think that the differences are quite subtle is telling (although they are more obvious on a monitor)... I don't think I'll bother with long captures and derotations on Mars anymore, not least because I find trying to fit the alignment frame in winjupos actuallly quite stressful!

For the middle column, I stacked each 2 minute capture individually (best 30%) and then stacked the resulting stacked tif images together.  you could also join the videos together in PIPP and then stack the best percentagre of that long video, or even simpler I could have just captured for 10 minutes!

[ShrewView]

That's really interesting. They are all very good images, and i'm assuming you've tried to use as close to the same levels on each but if i had to pick one i'd lean towards the single 2 minute stack with wavelets and decon. To me the derotation is slightly better than the 10 min stack without it, but compared to the single two minute run adds nothing. In fact its not quite as well defined to my eyes.

Great stuff though and given i'm currently processing and trying to get the best from the data i collected last night/this morning its a great help.

Thanks

 

[vlaiv]

42 minutes ago, CraigT82 said:

For the middle column, I stacked each 2 minute capture individually (best 30%) and then stacked the resulting stacked tif images together.  you could also join the videos together in PIPP and then stack the best percentagre of that long video, or even simpler I could have just captured for 10 minutes!

Capturing single 10 minute video / joining 2 minute videos is going to give better result than stacking individual results.

What if all the best frames come from first 2 minutes? In joined video this can be exploited, but in stack of individual results - you must use frames from other videos (this is of course extreme example to show that it is better to have single video).

[AbsolutelyN]

WinJupos seems to be working for me. I stacked 5 5000 frame clips over about 20 mins. Then derotated them and sharpened. The result to me seems better than any one individual frame. Maybe all the detail is in a single frame but feels easier to tease out. 

 

Edited September 16, 2020 by AbsolutelyN

[CraigT82]

17 hours ago, AbsolutelyN said:

WinJupos seems to be working for me. I stacked 5 5000 frame clips over about 20 mins. Then derotated them and sharpened. The result to me seems better than any one individual frame. Maybe all the detail is in a single frame but feels easier to tease out. 

 

That's a really nice image,  Yes the end result will be more pleasing, and If it's easier to sharpen that may be because there is more frames in the stack, not necessarily because it's been derotated. Have you tried just stacking together around 10 mins of video without derotation to compare with the derotated image? NOTE I'm assuming 10 mins, don't know what your sampling rate is, might be worth calculating to see what the longest capture time for your sampling is?

[AbsolutelyN]

That's a good point, quite likely because there are more images stacked just as in standard deep sky stacking. How do you stack multiple videos? Do you stack the resulting stacked image of each video in PS or is it possible to stack multiple avi's in AutoStakkert to a single image? Sorry, new to these apps.  

[vlaiv]

Just now, AbsolutelyN said:

That's a good point, quite likely because there are more images stacked just as in standard deep sky stacking. How do you stack multiple videos? Do you stack the resulting stacked image of each video in PS or is it possible to stack multiple avi's in AutoStakkert to a single image? Sorry, new to these apps.  

Probably easiest thing to do is join videos with PIPP and then have only one video in AS!3 containing all the frames from all videos - and process it as you would single video.

[AbsolutelyN]

19 minutes ago, vlaiv said:

Probably easiest thing to do is join videos with PIPP and then have only one video in AS!3 containing all the frames from all videos - and process it as you would single video.

Thanks - never used PIPP before .... will see if I can figure it out. 

[AbsolutelyN]

Well PIPP is certainly a very handy little program, thanks for pointing me to that. I combined 5 5000 frame files spanning 10 mins of rotation into a single AVI and limited it to the best 50%. Then did similar processing as with the Win JUPOS image. The Win Jupos version had one more 5000 frame file added but that was captured about 20 min later on so assume it would have rotated too much to use.  

There isn't that much in them to me though I prefer the Win Jupos one. That could be those few more frames or that I just did a better job processing. Top pole is better with PIPP. 

This amazing image by Łukasz Sujka was taken at about the same time so a good comparison of features with way more detail. 
https://www.astrobin.com/full/et9ouu/0/?nc=user

Edited September 17, 2020 by AbsolutelyN