Jupiter Dec 1st De-rotate comparrison UPDATE.

[Space Cowboy]

Ok, please bare with me. I've redone the 3 min avi de-rotation using the correct reference frame time. Here is the original thread :

http://stargazerslounge.com/topic/169805-jupiter-dec-1st-de-rotate-comparrison/

So here is the comparrison with the unrotated image :

De-rotated

No de-rotation :

I still can't see any difference so........

I split the 3 min unrotated avi into 1st, 2nd & 3rd minute segments then stacked each using 1000 frames per image. Below are the 3 shots :

1st minute:

2nd minute:

3rd Minute:

Clearly seeing deteriated in the 3rd minute.

Here is the 3 images animated showing the rotation during the 3 mins both in field and planet rotation:

I then stacked the 3 images into one using both Reg 6 and AS!2

Reg 6:

AS!2:

Just to compare here is the de-rotated image again :

All comments welcome

[scoobee]

I am no expert but loving your images and the clarity is amazing

[cgarry]

That is a lot of data for one post! But looking at the first 2 images, the non-derotated version looks significantly more detailed to me.

Cheers,

Chris

[Space Cowboy]

The de-rotated image is softer for sure and I used stronger wavelets too. It may be my eyesight but to me both Registax and AS!2 are correcting the small amount of rotation looking at the last 3 images compared to the animation. You would think there would be visible blurring.

[cgarry]

Yes, they will do that by realigning the align point features to the reference frame.

[neil phillips]

Hi Stuart, again i really am quite confused by this, Because it contradicts the examples done by Torsten on the winjupos help page ?

At first i thought it was As/2 versus registax effect.

But that doesnt appear to be the case. The only detail in the derotated version that actaully appears to be de roatated to any significant degree is the dark spot north polar region. As mentioned in my first reply ( its done exactly the same again ) look closely at the 4 dark spots upper left north. look closely at the bottom dark spot under the group of three dark spots. And in the non de rotated version the spot is somewhat stretched out.

in the rotated vid. it is much rounder ( though hard to tell because for some reason theres like a ton of blur going on ? )

Its interesting splitting the time. When i compare this spot on the de rotated version, to the 1 min AVIs, the shape is similar. Meaning not smeared out so that tallys.

The only problem is two things.

1 thats the only feature that seems to have benefitted greatly ? 2 the price you seem to have paid is vast amounts of over softening. Verging on what looks like a noise reduction filter of some kind ?

i wonder if such a filter is being used in the de rotation software. And i wonder why Torstens example completely contradicts this. Showing no such smoothing ?

Its all very strange. You mention using stronger, wavelets on the de rotated. version. But of course you shouldnt be doing that in such tests. all things should be equal always.

One can only wonder how soft the de rotated version is. Without using stronger wavlets. On the face of it i prefer the non de rotated version.

Any chance of seeing the same example but with identical processing. Meaning everything, referance frame. wavlets ect. I think i will run some tests too here. As i kind of was going to use vid de rotation. I assume mono and colour will behave the same here ? I wonder if dob tracking is having some kind of unusal effect here ? I wouldnt have thought so. But i am clutching at straws here. Trying to find reasons for the numerouse odd results this is showing.

thanks for taking the time run these experiments though. I hope feild rotation isnt throwing off the de roation software, because if it is. this will turn into a wild goose chase. Not your fault i know. As mentioned i am pretty much lost here. Trying to understand this

[Space Cowboy]

Sorry its a bit chaotic Neil I did not have much time to outlay things better. I used stronger wavelets on the de-rotated image to try and produce the same level of sharpening to make comparrison of detail easier. I'm on the ye olde farm computer for the next couple of days so don't have access to the softer de-rotated version.

As Chris says Registax and AS!2 has aligned the three 1 minute frames to the first frame which to me means the rotation has been corrected. I'm not suggesting that winjupos is a waste of time just that there is no real benefit at this focal length (6600mm) when using 3 minute avis. Personally I think you need to be using at least 5-6 mins before winjupos produces any real improvement.

There is one thing that I need to re-check and that is whether I need to sharpen the winjupos reference frame more? I did not want to introduce noise so used a low level sharpen......maybe that could influence the softer resulting de-rotated image?

The latest version of winjupos has an option for field rotation correction so my alt/az mount should not be an issue?

Whether my results are relevent to RGB imaging I don't know. Anyhow I hope my experiments are helping people understand whats going on rather than causing more confusion.

[stellawolf]

nice images well done

[Space Cowboy]

Revisiting this thread as Christophe Pellier from  the CN forum has made a similar study though he has presented the results much better than me :

http://www.planetary-astronomy-and-imaging.com/en/maximum-video-time-jupiter/

[neil phillips]

Revisiting this thread as Christophe Pellier from  the CN forum has made a similar study though he has presented the results much better than me :

http://www.planetary-astronomy-and-imaging.com/en/maximum-video-time-jupiter/

That's interesting Stuart. So what I think we are learning from this is 2min runs per channel would be fine to put straight into A/S2. No need to split the 2 min runs into one min runs and de rotate. But that 3 or 4 min Runs might work using video de rotation. Doesn't that contradict what you found ? Or have I read into this wrong

[Space Cowboy]

Yes it does show a slight difference in contrast between 3 mins none and derotated Neil. You can't see that on my images. Not sure what focal length he used.

[neil phillips]

Yes it does show a slight difference in contrast between 3 mins none and derotated Neil. You can't see that on my images. Not sure what focal length he used.

I thought I could see a difference when I tried. Wondering if its worth doing vid de rotate on the 3 min runs I was taking recently. I may test this again. you got me all video de rotating again. And its been sometime trying to remember the procedure. If I remember what im doing is taking about 2000 frames from the middle of the avi  using virtual dub. stacking on AS/2 doing mild wavelets. measuring using a de rotation of of RGB frames  then putting that into the video de rotate box ( page )  what was the Gig limit  do you remember Stuart. think I remember it was 2 Gigs ?

[Space Cowboy]

Yeah put the stacked mid avi image into image measurement and then open derotation of video. Been ages since I used the video derotation too, might give it another try with my 3 min avis lol

[Kokatha man]

.....as far as I'm concerned there is no point in derotating any avi's of 3 to 3.5 minutes duration (meaning that timespan for osc captures or "in toto" for r,g&b captures.)

Naturally the focal length/imaging scale is important in all this: ie, a larger apparent size for any planet means that the pixel shift will be greater due to the constant (ie, the rotational rate) but for my present Jovian imaging I'm running at around 5.6 metres f/l and I capture for 70 seconds per channel for a total capture timespan of about 220 seconds max (allowing for focus shifts & filter changes...)

I digested Christophe' post on CN and looked at his examples...personally I have every confidence in AS!2's ability to cope/compensate with the very small rotational shifts over these timespans and had made my own example with a hi-res red channel that supported these assertions (I will post it sometime, along with other images but we're getting ready to leave Katherine the morning after tomorrow and I have a large backlog atm...

I don't think your images reveal any major discrepencies from my own experience Stuart - sometimes you "think" one aspect of one type pf image is a tad better but then you look somewhere else and find a contradiction such that it becomes rather like splitting hairs: sure I do it for my own images (split hairs, that is) but you have to temper that with the fact that others wouldn't notice a rat's **** of difference.....and there's allways anomolies in various parts of each planetary image anyway..!

For my money WinJupos is a great tool but folks should realise it has drawbacks as well as virtues.....up here in NT I've made use of it to create derotated images from "classic" captures that span up to 25 minutes with considerable success, especially because of the constant battle with poor transparency here: however, W/Jupos images do soften the outcomes to a degree regardless, which can improve the overall image appearance and sometimes not - I suspect that a far more sophisticated program would be required to really "nail" optimal outcomes for derotation over extended timespans...but for a very workable/useable bit of software it certainly comes up "trumps" in many situations!

[Space Cowboy]

Totally agree Darryl! Tried another de-rotation avi last night on a 3 min capture and its made no difference from what I can see. I guess if one is shooting 10m+ fl then it will show improvements.

[neil phillips]

Was going to start up a new thread. But seeing as this is under discussion here. Thought I might as well post a link. I was able to do de rotation videos on my pc. this new laptop is the first time ive tried, and I keep getting this very weird what looks like negative video. Anyone seen this problem before?

 I think I remember there being a 2 gig limit on de rotate vids. So ive tried using castrator and pip to crop before running into winjupos to get the file size down.

And this is the result. Ive run it through Virtual dub, and used a compressed codec to get the file size down for SGL. This was a 5x powermate red with my 300p SW

Any help much appreciated

2013-10-30-0517_2-video0047 13-10-30 05-15-31_Y8castr-DeRot test 2 3.zip

[theo]

Hi Neil,

Have you tried convering the original AVI file to .SER format?

You can do this in PIPP and then run it through Winjupos without any limit on the file size. AS2 accepts it once derotated and process as normal from there.

I aways ran into problems with Winjupos and Castrator so tend not to use Castrator now because of this.

Pete

[neil phillips]

Hi Neil,

Have you tried convering the original AVI file to .SER format?

You can do this in PIPP and then run it through Winjupos without any limit on the file size. AS2 accepts it once derotated and process as normal from there.

I aways ran into problems with Winjupos and Castrator so tend not to use Castrator now because of this.

Pete

I will try that cheers for the sugesstion Pete.

[Space Cowboy]

video wont run on my machine Neil. Pipp is the better option for sure as Pete says.

[neil phillips]

Hi Guys SER file worked but getting bad edge artefact is that normal  ?

[neil phillips]

Heres The lovely Edge. 300p, Red on the  30th Oct.  3 min avi  De rotated. 1.5 drizzle wavlet only no other processing

[theo]

Hi Guys SER file worked but getting bad edge artefact is that normal ?

Hi Neil

I have seen artifacts on occasion after stacking and then applying wavelets to the derotated SER file. The workaround to prevent this that works for me is to get PIPP to do the frame quality selection on the full derotated SER and then output as an AVI again. This normally sorts it out in AS2. Bit long winded I know but does the trick where I have had this issue.

Pete

[theo]

Having looked at the image I'm now thinking that may be due to too high an LD value in Winjupos.

Are you using this Neil?

Pete

[theo]

Having looked at the image I'm now thinking that may be due to too high an LD value in Winjupos.

Are you using this Neil?

Pete

[neil phillips]

Hi Neil

I have seen artifacts on occasion after stacking and then applying wavelets to the derotated SER file. The workaround to prevent this that works for me is to get PIPP to do the frame quality selection on the full derotated SER and then output as an AVI again. This normally sorts it out in AS2. Bit long winded I know but does the trick where I have had this issue.

Pete

Interesting Pete. Is there any other settings you use on Pipp. like reject frames with over exposed pixels. ill try  your suggestion now. But thought I just tried that already with no change. I will double check.