PDA

View Full Version here: : HUGE problems with Registax


raymo
11-02-2014, 04:31 PM
The problem has to be me. I must be doing something wrong. I see all
these lovely planeteary pics, and I can get nothing like these results.
I am putting well focused videos into Registax, and following the instructions given in various tutorials and astro imaging books step by step, and yet the final result is only as good as, or marginally better than the chosen reference frame. I have tried every combination of every possible adjustment known to man, and all I end up with is tantalising hints of surface features[other than the main rings on Jupiter]. Just can't get even a reasonably sharp image. Unsharp masking in PS doesn't help
much either. HELP:help::help::help:
P.S. I even stopped my scope down to 3 different diameters to simulate different exposures when trying my DSLR's video
function. Incidentally, I have had slightly better results than with the webcam.
P.P.S. Regardless of under, correct, or over exposure, I get almost no trace of colour on Jupiter, and almost none appears
by manipulating the various colour controls in Registax or PS.
raymo

sheeny
11-02-2014, 04:57 PM
Which version of registax are you using? Procedures may be a little different between versions.;)

I'm currently using R6, so without knowing any specifics of what version you have or what your processing workflow is, I'll detail mine in case it shows something you've missed.

1. Load the video.
2. Step through frame by frame to find a sharp one, then create multiple alignment alignment points on that frame.
3. Align.
4. Limit the frames to the quality level you want, or the best # you want. I am only just starting to try limiting to the best #.
5.If you limited by quality, then on the stack page open the stackgraph and slide the slider on the top graph to the left until the number of frames to be stacked (down the bottom of the page) is about 100 (you can choose what you like).
6. Stack them, and save the file as a BMP or TIF with a suffix like Sun1 or Jupiter1, etc.
7. On the wavelet page play around with the wavelet settings till you find a combination that gives you a nice sharp image (without artifacts). Save that wavelet setting if you haven't already, so you can quickly get there next time.
8. Press Do All, and save the file as a BMP or TIF. I add a different suffix like say Sun1w so its different to the unsharpened file.

The reason you limit to quality, then later use the stackgraph is that with multiple alignment points, you have more good quality parts of images to choose from. If you limit on the frames to start with say to the best 100, you may be throwing out some parts of other images that may be better, just because most of the frame is worse. Likewise, if you don't limit the frames you stack, you're just stacking rubbish and diluting the good frames.;)

BTW, in statistics about 30 is generally considered the lowest reasonable sample size for stats like RMS and SD to reasonably reflect an infinite population, so I use that as a guide to a minimum desirable stack size. Its better to stack more, but there's no point in stacking rubbish (So I have been known to stack as little as 2 frames if that's all I've got that's any good!)

I hope this helps.

Al.

raymo
11-02-2014, 05:16 PM
Thanks sheeny; I pretty much do as you describe, with a couple of tiny
differences that I will try. I never get any sharp images after using
wavelets. I also don't understand why I get almost no colour.
raymo
P.S. I have R5 on one comp, and R6 on the other.

raymo
11-02-2014, 05:58 PM
I haven't used Reg for quite a while, and alignment was set to box type
aligning, and now I can't remember how to set it back to alignment points. I tried the multi option, but as I set the points each point created a box so I couldn't see where I was putting the points. Also, I couldn't open the stack graph because it wasn't highlighted.
raymo

sheeny
11-02-2014, 06:12 PM
Sounds like you are using R5 ATM. I don't have R5 installed any more:rolleyes::sadeyes:. Sorry I can't help ATM with the alignment point issue...

However, I have done some screen dumps of the 4 wavelet setups I currently use. I've just used the one image from my ED80, but I use different wavelets depending on which scope and the seeing. I hope this might give you some clues.

The wavelets apply at different scales: 1:1 uses a 1 pixel radius, 2:1 a 2 pixel radius, etc. So if you can estimate how many pixels a point is being spread over, that will give you a clue to which wavelet slider to use. I generally use a blend of sliders, but not all.

There's a big difference between linear and Gaussian as well. I don't use didactic... its just way over the top it seems (maybe I just haven't learnt to tame it;)).

1. This where I start for C11 images.
2. This is what I use for my ED80.
3. I use this if the image can't stand much sharpening before producing artifacts.
4. I use this one when I have the 2.5x Powermate in the train with my ED80.

Al.

sheeny
11-02-2014, 06:19 PM
Registax has a colour check button at the start. Its on the alignment page in R6 (top left). Its not checked in mine cause I shoot all monochrome, but I can't remember if it defaults to colour or not. Tick that and see how your colour problems go.;)

Al.

raymo
11-02-2014, 06:20 PM
Yes I am using 5 ATM. I can only ever use the top 2 sliders; all the others just produce chaos, even the slightest
touch. The colour box is already ticked.
raymo

sheeny
11-02-2014, 06:32 PM
Linear or didactic wavelets? Sounds like Didactic...

Al.

sheeny
11-02-2014, 06:36 PM
Here's a run down on my R5 procedure:

http://www.iceinspace.com.au/forum/showpost.php?p=817269&postcount=11

Al.

sheeny
11-02-2014, 06:37 PM
Tried unticking it?:shrug:

Al.

raymo
11-02-2014, 06:59 PM
Thanks for all your efforts Al. I'll try unticking it. I have attached an example image. And thanks for the run down.
raymo

noeyedeer
11-02-2014, 09:42 PM
I'd be happy with that raymo ... I could post a pic of a single frame unprocessed taken afocally .. and yours kills it ...

matt

see :) thou this was taken in November I havnt had any good skies since. I should add it was with my phone on video mode and I just take a pic from the video. it's lame but it's my best effort. does not show what the eyepiece shows thou

sheeny
11-02-2014, 09:55 PM
Doesn't look too bad to me either. There's a LOT that goes into getting a good image of Jupiter - superb optics, thermal control, excellent seeing, long focal length and a fast camera.

What scope and camera are you using? I used to image Jupiter at f/25 on my C8 and I'm pretty sure the top guys like bird, Damian Peach and Christopher Go imaged at about f/38 and in bigger scopes.

If you're shooting in colour, you're also behind the 8 ball, because the colour cameras are a lot less sensitive than a mono camera with filters... but hey you have to start somewhere.

For a OSC image, I don't think it's bad at all. There's probably a lot more to do in the capture side of things than in the processing in Registax.

:)

Al.

raymo
11-02-2014, 10:25 PM
This goes from bad to worse. I got rid of R5 and got R6, and then you sent me the info for R5, and I can't get R5 back again because the
download no longer exists. In R6 I have a problem because it won't
allow me to open some of my files. It says "access violation" and a load more jargon. I downloaded it again to no avail.
I use an Acuter solar system imager, a Logitech webcam, and my
1100D video mode. I use my SW80 and my 8" f/5 Newt on HEQ5.
Did you see the Jupiter image posted yesterday in beginners
astrophotography. He produced that with pretty much the same
setup as mine. My 1100D has the edge over my webcams so far.
Oh well, soldier on. Thanks Al and noeyedeer.
raymo

noeyedeer
11-02-2014, 10:41 PM
off topic raymo, did you sort out your avis from your webcam issue?

raymo
11-02-2014, 11:30 PM
You're going to wonder what kind of a wally you're dealing with here, but
since I had a thing called Transient Global Amnesia about 3 yrs ago, my
short term memory has gone through the floor. I used a file convertor,
but Reg 5 didn't recognise the codec. Someone suggested that I enable
a function in Reg 5 [I can't remember what it's called, but it was at the top right of the Reg 5 window.] I can't have a look in Reg 5 to see what it was called because I've dumped 5 and got 6. Anyway, it worked.
Thanks for asking, Matt.
raymo

noeyedeer
12-02-2014, 02:12 AM
lol. as long as ya got that isssue sorted , that's all that matters. :)

Camelopardalis
12-02-2014, 11:37 AM
I've been playing with AutoStakkert! 2 with some captures of a dodgy seeing night :sadeyes:

Curious to see if I can get the same results out of R6...but I don't get the wavelets...which should I be trying to move, how much, and why? :help:

raymo
12-02-2014, 12:28 PM
Hi Dunk, if you look back through this thread you will see Al has
posted a link to his run down on using Registax. .
raymo

Camelopardalis
12-02-2014, 01:07 PM
Ahhhh great, I'll experiment with those (sorry, I'd missed that), thanks!

Camelopardalis
12-02-2014, 07:05 PM
I must be really stupid :sadeyes: cos I can't figure out these wavelet things in R6... I get the bit about 1...6 being fine to coarse as I can see the way it behaves if I hit the preview button.

But should I be using automatic? Default? Gaussian? Use linked?

Then how much to denoise / sharpen for each one :confused2:

sheeny
12-02-2014, 07:20 PM
I posted some examples of my settings as screen dumps back in post 5. You'll have to experiment to see what works best with your setup, but a good place the start is the second screen dump - what I call "Dennis Wavelets" after the venerable Dennis Simmons Esq.

I don't use automatic. I don't use didactic. As you can see from my examples I sometimes use linear and sometimes Gaussian - it depends how much your image needs sharpening. Toggle between the 2 and you'll see the difference.

I haven't touched the denoise/sharpening controls inside each wavelet channel.

It's obviously very powerful, but like anything you learn it bit by bit.;) Patience Padiwan...

:)

Al.

Camelopardalis
12-02-2014, 08:11 PM
Thanks for the help Al, and sorry to raymo for hijacking :thanx:

I was doing something really stupid... I wasn't sliding the bars along :eyepop: I was trying to increment it using the up and down arrows :sadeyes:

With a bit of fiddling... wavelet 1 at around 60(!), wavelet 2 at around 40(!), here's what I settled on... I think this was just after Io had transit, which can be seen just off to the right. It's a bit small but I'm pretty stoked with that so far :D

Oh this is with my C8 and 2.5x Powermate and QHY5L-II.

Camelopardalis
12-02-2014, 08:48 PM
The bit I don't get is...the source avi looks awful :confused2:

sheeny
12-02-2014, 09:05 PM
Sounds pretty normal to me...:P

That's the beauty of stacking:thumbsup:

Al.

Camelopardalis
12-02-2014, 09:35 PM
Alrighty then :D got some more captures to chew through now :lol: