Registax5.1 & DBK21 or Colour cam Tutorial

[Kokatha man]

Sorry for the inordinate delay with this - I thought I had to put this into pdf form and left it on the backburner: so "behind the times" that it is from last year's (2009) Jupiter apparition.....but here it is....!

PREAMBLE: This “getting started” guide for Registax5 is essentially for the DBK/DFK style OSC cameras using the latest version of R5.1.

Other webcam users of those such as the Toucam series can follow the selfsame steps except they will not need to worry about the initial debayering steps…..and they will have to adjust the suggested amount of frames to retain after “Align” to allow for the fact that they will not be able to capture nearly as many as those who use the DBK21 etc…..

It is not intended as an “authoritive” guide in any sense…..rather a “quick-start-how-to” that runs through some processes/ideas quickly: but it does give a little time to a couple of aspects of Registax processing that I have not found to be widely discussed…..my advice is to look at as many different “how-to’s” as you can and also take note of various other tips on AA forums: there are numerous excellent guides available and the more you can experiment, the better your understanding and eventually your processing…..I consider myself still very much a “learner.”

What I do advise readers who might wish to use this as a guide do is EXPERIMENT with the various settings (I will suggest some options) because I believe the best way to ascertain what works best is to play with various setting (alignment styles, reference frame quality and %, using “create reference frame” versus non-usage, wavelet sliders & values - including various increment adjustments etc, etc.) Also to do some follow-up reading to get an idea – however simple - of what you’re doing with each application/process. (eg I have as simple an understanding of things as you can get! J)

It has evolved from my own experiences…..and owes very much to Cor Berrevoets of Registax fame: he helped me enormously as I attempted to wring the most out of my C11/DBK21 combo after encountering problems with R5 that were not there in R4 with these colour cameras and 2Gb+ avi’s…..

The comments about debayering in particular pertain especially to those cameras that use Y800 codec and fast framerate greyscale transmission of images from camera to capture device (ie, laptop.) These types of avi's need to be debayered – but in principle much of the info can be used by any cam user.

And principally this quick-guide is based on Jupiter processing….as an indication of variables (possibly to do with disk size and fine detail) I have personally found that the #1 wavelet slider comes into its own with Mars…..whereas I tend to stick to #2 & #3 sliders generally with Jupiter…..

You can of course pre-process any avi's using VirtualDub and Ninox: here I should thank Anthony Wesley for “tweaking” his Ninox program re debayering after I consulted him on it recently.

I’d use Vdub/Ninox for various reasons with some avi’s –exceptionally good ones, or, when I run one avi almost non-stop after another and (say) the last half of the first avi and the first half of the 2nd are good. I can join 2 or more good sections together in VirtualDub by selecting/splitting off the parts you want of each avi into bmps and then crop/centre/debayer these in Ninox - IF the total recording timespan (time at start of first chosen avi snip to time at end of last chosen avi snip) is not longer than that generally recognised as the imaging time limits for any particular planet.

VirtualDub & Ninox are also useful if you have an avi taken with a very poorly aligned mount where you are constantly using the handpiece to re-centre the image during capture, especially if the image drifts far to one side of the screen often…..Registax can sometimes baulk at this – particularly if you choose a reference frame somewhere in the framelist that is in a certain onscreen position: as you start “Align” and the box has to jump back to the first frame - which could be somewhere else entirely on the screen - mis-alignment can often occur, stuffing the process; particularly with certain sized alignment boxes you choose (bigger alignment boxes usually work better to counter this, if you don’t Vdub/Ninox…..)

First I load the captured avi into R5 by clicking on “Select” in the top left hand side of the screen and select my avi from the files in the browse window that appears.....the planet’s image in mono will then appear, along with the framelist.

[Kokatha man]

If you use the greyscale data transfer colour cam (in greyscale mode – which you really should be doing when using these cams, and capture using the Y800 codec) you then click "No" to the pop-up window that appears asking “Image seems to be B/W, process in B/W?”

[Kokatha man]

Note: there are 3 tabs with their text running vertically on the extreme LHS of Registax5 screen marked “Options” “Quality settings” & “Alignment options”

Go to "Options"…..choose "Use Debayer" under “Debayer” and check the correct GR, BG, RG or GB combo of the 4 radio buttons in the “Method” box directly below “Use Debayer” that gives the onscreen image its proper (captured) colour.....meaning the colour you set the image to when adjusting the red/blue sliders prior to capturing!

Also, almost right at the top and to the right there are 2 check boxes with radio buttons that need to be checked…..the “Extended Mode” to “Default” and the “Use extended mode for” checked to “Always”

Edited August 11, 2010 by Kokatha man

[Kokatha man]

First go to the actual “Framelist” window, click on “Frame 1” and by using the up and down arrow keys on your keyboard choose a good reference frame by looking at each frame that is selected as you move through the frames.Make sure when you are tabbing through these frames to select a good one that you do not inadvertently uncheck a frame box as it will cause Reggie to throw a wobbly and you’ll most probably be perplexed as to why….!

Sometimes you need to use your mouse to click on a frame in the list to “kick-start” the keyboard’s arrow key function for checking between frames….! Or you can grab the tab on the slider and slowly drag this from left to right watching the frames flick through…..and if you have unbelievable eyesight you can just depress the down key and watch the frames whirr past your very eyes….! J

There are various opinions – some say that it helps with alignment to NOT have Reggie trying to align using a frame that is overly-detailed.....the thinking is that Registax has less trouble obtaining a general grading of the quality of all the frames in “Align” (which is what it does therein) if you DON’T try and have it match all the rest with a “super-detailed reference frame”…..it grades on some sort of averaging using specific indicators/patterning in the images such as adjacent pixel-area contrasts and edge-definition as I (very simplistically) understand it…..supposedly working better with a “good” frame that is neither super-detailed, nor too much towards what would be termed a “poor” frame…..

Then there is the (perhaps) more common practise of finding a really well-detailed image amongst the framelist and going with that…..however, either approach requires you to find a frame that is at least reasonable in detail – so what you’re doing here is at least looking for a half-reasonable frame amongst the ones you go through, but don’t get overly-obsessive imho: experiment and make your own conclusions!

Edited August 11, 2010 by Kokatha man

[Kokatha man]

Next click on the “Alignment options” tab and choose “Default” for the “Alignment Method.” Below this is “Alignbox size” and here you can click on the size that either nicely encompasses the planet (eg, using Mars, Saturn) or the largest (256) that encompasses the majority of the Jovian disk’s area.

Also down further in this panel you find the “Align using Centre of gravity” checkbox which you can check (this negates having to choose an alignment box size or using the FFT display graph and setting a specific value, Cor recommends this mode) - you’ll most probably discover that the framelist will disappear when you do this, but in the check boxes above the black screen you’ll see a “Frame List” box – uncheck and recheck this and the framelist will re-appear! Cor recommends using this method for alignment because it calculates the optimum FFT value (see next para)

If you want to choose to choose your own alignment box size (and you can choose a small alignment box to only encompass some particular feature like (say) the Great Red Spot on Jupiter)…..then you will have to use “Align manually” and set the FFT graph value – for this a smallish red central element and no detached green elements to the graph other than that around the red core is a recommended ideal…..and of course you will need to check the FFT Graph box to get the display up and set your values!)

When you have decided on the frame you wish to use as your reference frame (in the screeshot "Frame 31") you click on the planet image to ensure the alignment box (whichever size/style you have just chosen) is centred over the planet's disk.

Below the “Align using Centre of gravity” checkbox is a dropdown window with various “Quality settings” – choose “Gradient2”

Just below this is another box marked “Lowest quality” and here you can set the % of frames you wish to keep – if you set it at 90% it means Reggie will align/quality estimate all the frames in the avi and select all those that are within 90% or better of the reference frame…..it does not mean you are getting the top 10% of all frames, merely what I just stated (which could be ALL frames in the avi!)

It isn’t actually necessary to make any % selection here …..you’ll do this after “Align” finishes with the slider bar tab that runs along the bottom section of the screen…..but you might as well make a setting of % that reflects what you think this avi may achieve in terms of the overall appearance of the video played in (say) Windows Media Player…..I’ll choose 90% for this example.

Once you’ve chosen your reference frame move the mouse over to the middle of the planet image and “click” – this will place the Alignment box over the planet…..if not using “Centre of Gravity” you will have to position this box manually…..

Edited August 11, 2010 by Kokatha man

[Kokatha man]

You then press "Align" up near the top LHS of screen and let the program run through this process…..at the finish of this “Align” process you will have gotten the program to select however many frames there are in the framelist within the acceptance % set in the “Lowest quality” box.

The post-alignment framelist may (as in the screenshot) appear still - or it may not – if not, just click that framelist checkbox up the top on and off and it will re-appear…..you will see how Reggie has rated all your frames from best (if you scroll to the top of list) to worst down the bottom. You can look at Reggie’s ratings and see whether you agree – and if you want to do your alignment again on the best possible frame you can…..choose what you think is the best by looking at the top half-dozen frames in the framelist now and re-do the whole alignment again using this frame as your initial reference frame!

But whatever you do there, you’ll then have to “Limit” the number of frames either to that which you’ve already set with the % box (ie, what is now showing) or to a lesser or greater number of frames by dragging that tab on the slider bar I mentioned above to the left or right…..to the left lessens the total frames you’ll stack but raises their overall % in regard to the reference frame, whilst to the right does the opposite…..around 1000+ frames is a good number if you can afford it to “drown out” noise when they are all stacked.

Remember – it is often better to use less frames but of a higher % than more frames of a lower % - but applying wavelets etc further on down the processing will create more grain/noise in stacks with lower frame numbers.

Once you decide on how many frames you want to limit at, hit the “Limit” button next to the “Align” button up near the top LHS.

In the example below you can see I’ve limited at 1000 frames with a quality of 97.3% - this was a good avi: that little yellow tab above the number “3” in “Frame (3028)” right down at the bottom centre in the screen shot above is the slider bar tab spoken of…..

Edited August 11, 2010 by Kokatha man

[Kokatha man]

Then you MAY choose to do a "create reference frame" from (say) 50 frames.....this will optimize 49 other frames against your original reference frame, stack them and present them to you as a sort of “master reference frame.”

Or you can skip this “Create reference frame” process and do the optimizing with the original reference frame you chose…..proponents of this say it makes no difference either way…..

If you hit the button below “Create a Reference frame” that says “Create” (next to which is a box where you set the number of frames you want to use for this “master reference frame”) it will do an optimise & stack of the relevant number of frames chosen and take you to a wavelet screen to allow you to apply a small amount of initial wavelet sharpening (I term this the “initial wavelets” application.)

[Kokatha man]

When you are taken to this screen you’ll see yellow text on the black screen instructing you to “enhance & continue” (meaning apply wavelet slider applications.)

I typically use only #2 & #3 sliders set for 11.0 & 12.2 (max!) after Reggie has ran through this “create reference frame” process and taken you to the screen with the wavelet sliders: you can use less slider application than the values I’ve suggested above – or vary the sliders used – just make sure you’re not heavy-handed here, because over-sharpening this “master” can make Reggie’s job more difficult in the next “Optimize” process: do NOT sharpen to where you start to see the image go grainy – this is a NO-NO and is easy to do with only a stack of 50 frames!

After deciding upon slider values press the “Do All” and then “Continue” buttons up the top…..

[Kokatha man]

After the preceding you will be taken to a screen with 2 images (the coloured “reference” image & the green “current” screen) and one of the buttons up the top lit in blue which says “Optimize” – directly below this and slightly to the right is another button saying “Optimize” also (grey coloured – yellow lit with mouse-over)…..hit this next.

An “Optimized restart window” will pop up asking “A new reference was created, continue?” to which you click “Yes” (this refers to the fact that this “new reference” will now be the “master reference” and it has been created from the original reference frame and the 49 others…..)

[Kokatha man]

The optimization process will then start - wait till this process is completely finished (the graph that is displayed may make several complete transits before the optimize process is completely done.)

[Kokatha man]

When the optimize stage has finished hit the button marked “Stack” which is next to the blue lit optimize button…..this will then turn blue, and directly below this and to the left a bit will be another “Stack” button which is underlined green now.

Press this 2nd “Stack” button and the stacking process will begin…..

You’ll notice the 2 graphs displayed at the end of “Optimize” and also at the end of “Stack” (if you check the upper LHS box saying “Show Stackgraph”) – and you’ll see a slider tab at the top LHS of the stackgraph: if your green graph line is a real “saw-tooth” up & down fluctuating type you can lower that slider tab and it has the effect of removing frames from the stack that are wildly-variant from the median (those individualised sawteeth)…..this can improve the final stack quality but you will be jettisoning frames by doing so – the frame quantity box down the bottom RHS will show you how many frames you have left if you employ this…..the graphs shown don’t require any frame culling.

[Kokatha man]

Then it is time to hit the “Wavelet” button along the top next to the first “Stack” button – the “Wavelet” button will turn blue…..taking you to what I call “final wavelets application” stage.

Here I first do an “RGB Align” – there are a set of 14 buttons underneath “Functions” on the upper RHS of screen.....”RGB Align” is one of them.

Press this button and a pop-up panel screen will appear: press “Estimate” at the bottom of this screen and in a few seconds or so the channels will undergo an alignment – seen by the change in values of X/Y positions in this pop-up screen (if you’ve done a Vdub/Ninox processing before using R5.1 you’ll find that the channels are nearly always perfectly aligned….. suggesting that Reggie has most probably been able to perform all his functions more efficiently/effectively…..)

I then apply final wavelets slider applications.....here I bump up #2 & #3 sliders values from those set during initial wavelet applications to 31.2 & 32.5 (max!) respectively.....you should see a big improvement in image detail/resolution by doing this.

Again, as in initial wavelet applications you do not increase the values to where the image goes grainy (you have more latitude here if you chose around 1000 frames to stack) and do not increase your applications to where detail may seem very enhanced but the image overall becomes “heavy and dark” wrt detail and tones…..

Once again experiment to get your image how you wish it – a “balance” is always a good objective…..try various sliders and settings to work out what they do and how much you want to use each of them…..and how many sliders you think achieve your objectives with the image. An old adage dictates that good avi’s require minimal processing and virtually no “forcing” of enhancements for maximum image quality.

Edited August 11, 2010 by Kokatha man

[Kokatha man]

Next, and especially if you don't have/use AstraImage, you can do quite marked additional enhancements with the "Wavelet Filter" adjustments by increasing the "Initial" and "Step" values at the bottom of this control panel when you enable them.....but if you intend using deconvolution in AstraImage you have to balance WF applications with subsequent deconvolution applications in A/Image, as WF tends to restrict the amount of deconvolution you can subsequently use in AstraImage…..this may be counter-productive ultimately…..

Wavelet Filter alterations essentially change the parameters of those wavelet sliders on the LHS of the screen we set to 31.2 & 32.5

This “Wavelet Filter” button is also under “Functions” in that block of buttons on the RHS of R5.1 screen…..press it and another pop-up panel screen will appear.

At the bottom of this are 2 value/number boxes, the first marked “Initial” (default setting 0.10) and the 2nd marked “Step” (default setting 0.00)

If you raise these values carefully you will again see very marked changes to the detail/resolution in the planet’s image…..be discrete in the amounts you use and remember what I said about over-applying normal wavelet slider values, as well as about the Wavelet Filter application’s affect on deconvolution in AstraImage later…..try increasing “Step” values to 0.10 or 0.20 and perhaps “Initial” to 0.14 or 0.16, watching for the same “graining” and/or image heaviness where colour detail becomes too darkly toned and “smudgy” that I warned about before…..but you can use these controls to bring the image almost to final tweaking phase in Photoshop or other software if you really wish to…..compare the image below to the one in the preceding post…..

[Kokatha man]

Press “Do All” underlined in green up the top and then choose "Save Image" (in red text next to “Do All”) after any/all final wavelets screen applications just described.

(I now DON'T go any further in Reggie)..... I save as “FIT (16BIT) SEPARATE” files in the save window’s options, ready to take the image into AstraImage.....with a final tweaking in Photoshop.....saving as separate fits files splits the image into 3 individual R,G & B channels and allows better individual channel enhancement in A/image (or P/shop for that matter) and provides better final colour-rendition imho…..

If you want to use the Wavelet Filter controls I’d recommend doing a “default” save of the image as separate fits files (ie, no WF applications) as well as a couple of varying WF applications…..you can save as many varieties of the image as you like and see how each of them responds to various deconvolution and sharpening etc adjustments in other programs…..just make sure that you keep a record of what each set of RGB file images has had done for later evaluation/reference!

After deconvolution applications to each channel in AstraImage you can recombine the separate saved individual channels back in AstraImage afterwards.....this seems to give a bit better dynamic range in the colours than not saving them as split channels.....but you might like to also save each variation as a single tif colour file (uncompressed) for a quick processing in (say) AstraImage to get a feel for things…..but whether it’s separate RGB fits files or a single colour tif file you work further with…..that’s all in another processing program….!

[Space Cowboy]

Great tutorial!

[AndyUK]

Wow... This is just what I needed too... . Thanks VERY much! If it's not a sticky yet, then hopefully it soon will be...!

[ampleamp]

perfect timing for me, thanks very much

[StarlightKnight]

K

Just come across this, tried it out and wanted to thank you

Its helped me already.

Thanks again

Don

[bus_ter]

I'm glad this got brought back to the top of the list, a great guide which is ideal for all the new Jupiter images appearing from the cheap webcams :-)

[Alan]

Hi KM and thanks for this excellent tutorial.

I note from the first box that the image on your AVI capture is much brighter and with higher contrast than mine and wonder whether you could advise on the gain, gamma and shutter speed combinations you use? My set up is similar to yours with C9.25 instead of C11 and DBK21 at F25-30 on Jupiter.

Thanks,

Alan

[Kokatha man]

I sent this reply to Alan last night in a PM (had only just seen it, sorry!)

I thought that I'd post it here for anyone else's benefit who might be interested.....this is using IC Capture.AS and of course gain values, fps etc vary with scopes/apertures re optimum values.....with the seeing playing a very big role!

The tute wasn't meant to include any capture methods not only because it's a seperate aspect.....but those abovementioned variables play a very large part in what settings you'd use in any single situation.....

I haven't used my DBK for quite some time but I set gamma at between 60 to 70 for capture (remember to lower the gamma when focussing, to increase contrast and help find precise focus - but reset it back to that range before capturing!!!) and mostly I captured at 30fps and 1/30second exposure.....if the seeing is good the red, green and blue histograms should be steady or just pulsating slightly - ie, expanding and contracting along the x-axis a small amount - this is what I describe as "licking" and you should set the gain so that the red histo ( the one which is usually the most extended) is "licking" the 95%+ value (ie, not quite reaching the end of its x-axis....ie, not quite 100%)

Red and blue sliders are adjusted so that the onscreen image's colour apearance is as close to how the planet appears (or you want it to appear - although it's a good guide to use other imagers piks to help determine your own preferred colour-rendition.....)

Blue slider is considerably higher than red for Jupiter - from memory the red was set at around low 20's and blue in the mid 40's but if you're particularly interested i could look this up.....although for different scopes and conditions this could vary somewhat and is most probably best determined as above.....eg, I used different values of r&b for Saturn and Mars than Jupiter.....

Cheers, Darryl.

Edited November 23, 2010 by Kokatha man

[paraman]

Thank you for this tutorial I tried it tonight on one of my previous avi's and compared it to my first effort, to say there was an improvement would be an understatement.

Roger