Jupiter comparing avi length.

[Space Cowboy]

The subject of rotational blur and capture length has been discussed on another thread concerning Jupiter.

Here are 2 images from Aug 10th at 5-05am the first is a 1 min avi resulting in 1000 frames stacked. 2nd image is 3min avi of which 3000 stacked.

The focal length used was 6.6 metres (f26.4). DFK cam using 30fps.

Seeing was very stable.

Images have been enlarged. Not much in it but for me the 3 min capture has more vibrant detail.

[cgarry]

That is an interesting experiment Stuart. Have you played with WinJUPOS with the same data?

Cheers,

Chris

[Space Cowboy]

Not had a serious go with Winjupos Chris. I was under the impression it was more for RGB imaging?

[si@nite]

Very intresting Stuart! The bottom shots also pulled out more detail, eg small white spots are more pronounced and clearer detail in the bands.

In my estimation its certainly worth experimenting with longer runs, i myself have had better results with longer avis on jupe.

[StuW]

I agree that the 3 min AVI has more detail on Jupiter, but the shorter AVI seems to show more detail on the moon... strange

[cgarry]

I have only had a brief play with it some time ago, but found that it did not handle the colour AVI I tried to use. I have not tried it since.

I would have thought that de-rotating would work for OSC imaging too, allowing longer captures to be used with more good frames to play with. PIPP could always split a colour video in to separate R. G and B videos if that works better. I shall have a play with WinJUPOS myself if I ever manage to point a scope at Jupiter and get some data!

Cheers,

Chris

[Clayton]

Not had a serious go with Winjupos Chris. I was under the impression it was more for RGB imaging?

Not so Stuart

winJupos works equally well with colour data! (in fact it's even easier to use with colour data)

I agree that the 3 min avi shows more detail, and that's enough for me

I'm not sure why the practical time limit exceeds the theoretical ? Perhaps Mr. Nyquist has a part to play in this as well?? (perhaps the fact that detail is maximised when it covers 2 pixels allows a slight smearing to improve the visible detail rather than degrade it ??)

[neil phillips]

What my eye sees is that the extra frames although smoothing the image much more are also blurring finer details. I can only assume more bad frames are getting to the mix on the higher stack. What also needs to be considered Is, the higher stack will take much more sharpening possibly balancing out some of the problem of the more less ideal frames being tripled up. This is one case maybe where the rule of identical processing on experiments might be slightly misleading. After all i capture more frames with the intention of being able to sharpen more. and hence reveal more details.Winjupos is the way to go i think. Though i am a little concerned about what this is actually showing.

because will the extra sharpening really balance out the problems of many many more frames ( many of which are sub standard and will creep into the mix even using ninnox ? ) being added into the the mix.

It leaves me with the idea of shooting long captures using derotation on winjupos. But also doing maybe 50 sec rgb captures at 11 meters. to compare this problem that i think stuarts very iinteresting comparison seems to be showing.

Does anyone else see this effect ? or agree i might be right about this. Also something that just popped into my head is Stuart is using dob tracking. because i can not tell how well hes setup is tracking. is it possible the effect here is more pronouned because of worse tracking compared to say a well polar aligned NEQ6. I cant answer that without seeng the full 3 min avi done by Stuart. But its a interesting question. as is the idea that bad frames are blurring the fine detail in hes longer capture. To be honest i thought i noticed this before on some of your captures of jupiter Stuart. But strangely not on Mars. which i would have thought. would reveal this problem even more. Being smaller details and higher powers ? mmmmmm

BTW I think Stuart is right winjupos will not work on colour cams because the RGBs are being captured ALL AT THE SAME TIME. And for de rotation to work. Winjupos needs to calculate the different time spans between the different channels. Correct me if i am wrong somebody. Is that assumption correct ? Rob how will winjupos calculate a time span difference between say red and green. When there is no time difference to calculate ? De rotation works on the idea of rotation from the start of red to the end of blue. in colour cams the smearing is happeing in all 3 channels at the same time. This can not be calculated out surely ? as there is nothing to calculate out. Am i missing something ?

[Space Cowboy]

(perhaps the fact that detail is maximised when it covers 2 pixels allows a slight smearing to improve the visible detail rather than degrade it ??)

I reckon you've hit the nail on the head Rob! At normal size ( i will do more repros at different sizes) the detail appears more vibrant on the longer avi but when the images are enlarged to nearly 10m fl the central banding detail does look slightly elongated.

Very good points you make there Neil. Yes the tracking was drifting somewhat during the 3 min capture as it does when tracking in the east (quirk of this dob). Field rotation could be a factor too with the alt/az tracking.

Mist descending meant the 2nd half of the 3 min avi was slightly darker which would hand an advantage to the first 1 min segment.

For sure I'm going to experiment more with these different avi lengths and post up the results.

Btw the moon on 3 min avi is poorer because I forget to set an alingment point on it.

[Bubble and Squeak]

This is very interesting to see. the 3 min avi certainly does have a few more features on it but I agree with Neil that fine details are present but more blurred on this one. Though this encroaches into the territory of 'personal taste' of course.

If you were shooting at that frame rate for those amounts of time then it would seem you've stacked way more frames than you've discarded. This I guess is another personal taste thing or an argument for another day but I've always tried to skim just the cream from my avis and only keep the very best frames.

This can sometimes result in me stacking 600-800 of a 2,800 frame avi. Guess we all learn in different ways and I would accept that my final images may look too noisy for some people's tastes whilst to me they look like I've not overblurred the smaller features on the planet by not stacking a large percentage of bad seeing frames into it. Something that you've possibly done by stacking 1000 of a 1800 frame avi in the 1 min version and 3000 of a 5400 frame avi in the 3 min version. Just some thoughts, not criticisms. They are both very beautiful images.

[Space Cowboy]

Gain was at 80% for this capture so personally anything less than 1000 frames stacked is too noisy. My processing method has changed over the last 6 months especially with the use of the brilliant AS!2 and I have used less sharpening with more attention to local contrast control so it's possible I could get away with fewer frames.

I'm certainly open to any points of view even if at first i don't agree.

[Pentatonicblues]

for my first Jup of this season, I experimented with 1 min and 3 min avis through RGB filters. Beacause Jupiter was relatively low at 5a.m. I had more useful sharp frames from the longer stack (usng a lower saved % of frames) but RGB alignment was poor, leading to colour fringes. Hence I am experimenting with winjupos, which does improve things greatly. I agree with the comments above, for a combination of reasons, the longer capture 'appears' better aesthetically but is probably showing less fine detail than the shorter one. Great experiment to see!

Stewart

[cgarry]

BTW I think Stuart is right winjupos will not work on colour cams because the RGBs are being captured ALL AT THE SAME TIME. And for de rotation to work. Winjupos needs to calculate the different time spans between the different channels. Correct me if i am wrong somebody. Is that assumption correct ? Rob how will winjupos calculate a time span difference between say red and green. When there is no time difference to calculate ? De rotation works on the idea of rotation from the start of red to the end of blue. in colour cams the smearing is happeing in all 3 channels at the same time. This can not be calculated out surely ? as there is nothing to calculate out. Am i missing something ?

My understanding from reading about what WinJUPOS does (rather than any real experience with it), is that it de-rotates every frame compared to a reference frame. So if a reference frame were picked from the middle of a long colour AVI, then WinJupos could de-rotate all other frames to match this frame and any blurring due to rotation would be elimininated. So I think it should be just as useful for colour camera imagers as mono camera imagers. Though since I have not yet successfully used WinJUPOS I could be wrong!

Cheers,

Chris

[Kokatha man]

.....just a quick response: Rob is correct in that WinJupos works very well on colour images (that option is available on the tabs when measuring any image loaded into W/Jupos...it is in fact much simpler to do although I didn't include doing them in my tute... )

If I can get on top of the huge "work" workload I'm flogging myself like a dead horse over atm I will post some colour WinJupos compilations from the AS2 repros' I'm compiling in my sneak moments to stay sane as I slave away with "real" work....!

As for splitting the channels from an OSC you can do that in the "Save image" options of Registax ("save as seperate fits files.") Although I regularly did this when I owned an OSC and deconvoluted each channel seperately before "RGB combine" (in AstraImage) I have to honestly say that in reality any gains were minimal.....every now and then folks come up with some of these types of ploys (eg, various debayering methods*) but after a while any added processing seems to often be discarded due to the infinitely small incremental improvements they might make.....and nothing beats good seeing and collimation!!!

One aspect that is being overlooked is that a major reason for the longer avi's is not just to stack more frames but also to garner more frames in the first place so that you can be more aggressive in the selection of your quality cut-off.....if the seeing in any pair of imaging sequences is reasonably constant then obviously there's a fair possibility that the best 1000 will be superior when taken from the longer avi.....

Personally, Registax 6's noise reduction applications in wavelets affords the ability to work a smaller stacksize much better than before (although I should add that the function was also there in R5!) and consequently I'm more interested in getting sharper definition in my more "standard sized" stacks rather than going for large numbers per se as I can "squelch" noise pretty effectively.....that old maxim of adding poorer frames doesn't hold much water to me, and as for WinJupos I've made a couple of comments a few minutes ago in Neil's thread about mixing good and bad images using that software, and if I have a couple of sets or so of high definition imagery then in those cases I can be more aggressive with wavelets of deconvolution etc in the larger (but importantly high definition) stack WinJupos affords me.....

* I have to give a plug for "AutoStakkert" because I've just been reprocessing some (very) old OSC avi's of Jupiter and it does improve the outcomes over (say) ninoxed R6 processings, like myself and many others claim.....and specifically apropos my debayering comments here, for these OSC avi's it may well be that AS2's mode of debayering might also be an influential aspect...(I believe it uses one of the more "effective" debayering modes.)

[neil phillips]

.....just a quick response: Rob is correct in that WinJupos works very well on colour images (that option is available on the tabs when measuring any image loaded into W/Jupos...it is in fact much simpler to do although I didn't include doing them in my tute... )

Fair enough you live and learn. Its not something i looked into, like Stuart i didnt think it worked. I stand corrected apologies Rob

[Space Cowboy]

One aspect that is being overlooked is that a major reason for the longer avi's is not just to stack more frames but also to garner more frames in the first place so that you can be more aggressive in the selection of your quality cut-off.....if the seeing in any pair of imaging sequences is reasonably constant then obviously there's a fair possibility that the best 1000 will be superior when taken from the longer avi.....

I totally agree Darryl . Both these images had the same percentage (55%) of frames stacked so the only reason the 3 min image could have poorer frames included is if the seeing fell away after the first minute.

I will endeavour to give winjupas another crack!

[Kokatha man]

I totally agree Darryl . Both these images had the same percentage (55%) of frames stacked so the only reason the 3 min image could have poorer frames included is if the seeing fell away after the first minute.

I will endeavour to give winjupas another crack!

.....just a couple more things Stuart, whilst they both had "55%" of frames stacked I would be surprised if that % was much of a guide: I often used to look through the frames that Reggie would "Limit" at (you can do this by bringing up the framelist that usually disappears during the quality estimation stage if anyone doesn't know that) and sometimes you simply wonder how such & such frame could possibly be better than so & so frame.....what I mean is that even if the seeing had been rock steady/consistent between the 2 avi's (or avi sets if it's a mono cam) it won't necessarily mean there is any real correlation between the stacks re quality progression etc.....but what I'd really suggest is that you try comparing the top 1000 frames from the 3 minute avi with the stack of 1000 from the shorter one..?

[Space Cowboy]

Yep that would be a good comparison 1000 from each. This is what makes AS!2 so darn briliant because of its speed you can experiment with different stack sizes and settings so much more.

Btw I use noise reduction in image analyser rather than Registax 6 especially colour noise reduction.

Its all good stuff! I can see my HD filling up with repros over the weekend lol.

[Space Cowboy]

Ok here goes, 4 images coming up. This time I've used my best Jupiter avi from Nov 2012.

All images are processed identically and they are split into a 1 min, 2 min and 3 min capture length each with 1000 stacked frames. The final image is 3000 stacked from the 3 min capture. There are some alignment marks showing as I did use over 200 alignment points for each image (don't usually use this many)though interestingly the 3000 stack does not show these artefacts.

1.5x drizzle has been used in as!2 so with a capture fl of 6.6 metres these equal a fl of 10 metres.

I can't see any rotational blur on the 3 min capture. This makes me think that the previous results were down to a deterioration in seeing during the longer capture rather than rotational blur.

1 min capture 1000 frames stacked :

2 min and 1000 stacked :

3 min and 1000 stacked :

3 min and 3000 stacked :

[neil phillips]

Yes i agree rotation isnt a problem here. More importantly Stuart it actually shows, that your tracking on this night, doesnt seem to have tripled up any smearing. Though i had to take your image over to reg 6, add some extra sharpening that is needed by the extra frames, to confirm this. Which is why i said earlier with a test like this, identical settings might be misleading.

At first glance it still seems like there is smearing on the 3000 frame stack compared to the 1000 frame examples. However this appears to be a illusion. created by the fact. that 3x the amount of frames, is actually smoothing the detail. Its there. but its smoothed. and the eye doesnt see it quite so easily. Once extra sharpnening is apllied,all is equal, the detail is as fine, in some cases more fully formed ( which is what we would expect ) and without the bad noise that the 1000 frame stacks seem to have. Interestingly i would say i tend to over sharpen slightly. you tend to under sharpen slightly. If we meet somewhere in the middle i think we both may be hitting the magic balance im not sure ? This was certainly a great night for you. a fantastic image thatr shows many things. one SW optics are really darn good. two colour cameras on bright objects like jupiter seem to do as well as there mono brothers. Three. dob tracking isnt just a poor mans stab at it. it actually can deliver. Your a genius mr cowboy. This is one instance when a cowboy ISNT a cowboy. if you get my drift. No pun intended. Your image on the bottom looked better to me with extra sharpening. But thats taste. not capturing. Everyone is different, and you may not agree.

[Space Cowboy]

Cheers Neil. Yeah the last one may need more sharpening but I just wanted to show what differences the extra capture time and frames made without changing the processing. The version of this nights image that appeared on sky at night looking back now was over sharpened but since using as!2 1.5x drizzle I've gone softer on my sharpening so to speak.

I will re-process those aug 10th shots, there was a fair bit of image twitching during that capture possibly caused by the cam's usb cable being layed across my arm (rushed capture to beat descending mist). Us Cowboys can be rough and ready at times lol

Looking at these last images re-enforces my belief that the stacking/aligning process compensates for a certain amount of rotation especially field rotation. When doing animations there is about 1 degree of rotation between captures.

Its a great shame the Auto 250 Dobs are no longer sold (SW now only sell goto 250 dobs) because at £650 this scope was in my opinion the best value planetary imaging scope money could buy.

[Space Cowboy]

Here is the aug 10th images again this time enlarged to a equivalent scale of 15 m fl.

1 min 1000 stack, 3 min 1000 st and 3 min 3000 st. i can't see any rotational blurring. There is very little difference in the images and I did give the 3000 stack extra sharpening.

1 min capture, 1000 frames stacked :

3 min capture, 1000 frames stacked :

3 min capture, 3000 frames stacked :

[gary1968]

Stunning images, gives me something to aspire to, thanks for sharing.

Gary

[Space Cowboy]

Thanks Gary!

[luis]

I like very much the 3 min and 1000 stacked .