I just can't believe the seeing at the moment. I don't remember a time where the jetstream has hung overhead so consistently and the seeing has been so consistently bad.
My last night of good seeing was the 30th March
Anyway, out again last night at about 10pm to push my luck. The seeing was average to start with, I took 2 avi's before the seeing turned to mush so I gave up in disgust. 10 minutes out there, and back inside again
Here's the results of the effort, avi#1 was the only usable avi. Haven't seen Red Jr in a while, so at least I captured him. Amazing how close they're getting now.
Usual processing routine -
Virtual dub to save avi as bmp's
ppmcentre to crop and rank
split into rgb
align and stack separate rgb channels in registax. Create reference frame with wavelets 3@10, 4@15.
Stack 100 frames of each channel, Wavelets applied 3@12ish, 4@25ish, 5@35ish, 6@48ish, save as tiff
Stack 200 frames of each channel, Wavelets applied 3@12ish, 4@25ish, 5@35ish, 6@48ish, save as tiff
Open the 100 r/g/b channel images in AstraImage, Convert to greyscale, ME deconvolution 3/1.3, Recombine (with R/B shift), Gamma adjust 0.7, Save as TIFF
Open the 200 r/g/b channel images in AstraImage, Convert to greyscale, ME deconvolution 3/1.3, Recombine (with R/B shift), Gamma adjust 0.7, Save as TIFF
Slight curves and unsharp mask of each image in photoshop, save for web.
So, the left image is the result of the 100 frame stack. The right image is the result of the 200 frame stack.
My conclusion? I'm still convinced that my "standard" of stacking the 100 best frames is producing a better result than stacking more frames, where you'll end up stacking bad quality frames with the good ones.
In my opinion, the image on the left is sharper. Obviously I have to be careful not to process too hard, else it ends up quite grainy due to only 100 frames being stacked. If the seeing is better than good, I'd expect you'd have MORE of the best quality frames to stack, so in that case maybe stacking 200 would work equally as well, and produce a smooth result too.
But in bad seeing (which is all I get these days), this seems to produce the best result in the conditions.
Matt, I use the "netpbm" unix tools (under a unix emulator on Windows), combined with a shell script that I wrote (with help from Bird as to which of the netpbm tools to use).
It creates a red/green/blue directory, takes each colour bmp, extracts the red channel and creates a bmp in the red directory. Same for green and blue.
DP uses "aviraw" to achieve the same result (a split of colour channels). I haven't used it so I can't go into detail about "how" to use it - but i'm sure DP will (he's great like that ).
Paul, post above in reply to Matt explains how I split into r/g/b.
Quote:
Originally Posted by ponders
Why do you convert each rgb channel to greyscale when taking them back into AstroImage?
Even though they are greyscale separate channels when sent through registax, they are processed as colour and when I save the TIFF from Registax, it's a colour file (even though it's just the red channel, for example).
I need to convert to greyscale in AstraImage because the LR/ME/VC deconvolution can only be performed on greyscale images.
AstraImage also performs a histostretch at the same time as it converts it to greyscale, so you'll notice the result is brighter.
DP, no extension tube. Not even ToUcam pulled out as far as it will go. Seeing not up to it. 4/10 it was, and got worse after 10 minutes.
load your avi into aviraw (doesn't matter if it is raw or not) save first as a single rgb and then separate r,g & b avis. you still have to convert these three new movies to bmps etc for ppm centre etc
Great Images and experiment Mike... and I know what you mean about the seeing.. look on the bright side, at least your near the edge of the stream and if it shifted north you'd escape its clustches - it's sitting pretty well dead centre on Brisbane so we're stuffed for a while yet methinks.
Quote:
Originally Posted by davidpretorius
load your avi into aviraw (doesn't matter if it is raw or not) save first as a single rgb and then separate r,g & b avis. you still have to convert these three new movies to bmps etc for ppm centre etc
just a note on this DP if planning to process the separate colour channel avis in registax you might need to pass the avis out of avi raw into VirtualDub and resave as avis before registax will process them... at least I have to do this
the main point in registax is to "optimise" each and every frame to a reference frame. as we know the seeing is never perfect, so each frame is different, either out of focus or distorted due to the atmosphere. so we somehow need to alter each frame to try and get as many good to great frames to stack together.
so after loading an avi or group of bmps in to registax, how do we know what to optimise to. we want to find the best image we have and try and use that to say, hey other frames, this is what you should look like.
this can be done three ways
1. manually going thru every frame and deciding on which is best, then choosing an area and pressing align!
2. choosing the first frame in the sequence, registax will then try and work out what it thinks is the "best" frame and move it to the top of the list, then by choosing the new "best frame", you run align again.
3. by using the "argument" -qestimator -renumber=last in ppmcentre, you do a similar thing. ppmcentre uses the gradient method to pick the sharpest image and re sorts the order.
once we now have the best image, we limit the number of frames to where we see fit.
next, there is a create reference button with a "50" next to it. what this will do is grab the 50 best frames and stack them. this then goes to the wavelet page where you can apply some wavelets.
What we are doing here is not just saying to the optimise stage," here is the best image, try and make the others look like this" we are actually using the best 50 or so frames stacked, applying some wavelets to get an even better image to try and optimise against.
This works really well on ppmcentre'd images, but i have found that it you load just the avi in, then moving more than say the 3rd slider (5.8) then errors occur. If they are ppmcentre's then 3rd can got to 10 and 4th to 15.
Not sure if this error is caused do to the image moving, ie not centred or not.
Probabley enough said at this stage without confusion setting in.
Remember, we want to try and optimise the images to look as good as is possible before stacking.
There is a beta version of registax and bird had once of his own where images were being multi point aligned ie compare two images and try and "morph" each image to look like the reference. Should be out in the coming months hopefully
I watched Mike do a run through with PPMCentre at Lostock but I can't for the life of me remember how to work it. I've tried to read Birds description but as a non techie I understood maybe only every second line.
Dave or Mike can either of you help out with a basic how to? Using simple Australianese mate, cobber, bloke
I worked out another way of centering the bmps last night rather than using ppmcentre. Unfortunately it doesn't rank them but ImagesPlus (for those who have it) does a damn fine job of centering all the bmps prior to dumping them into Registax. Just use the planetary alignment feature.
You will need a good deal of RAM though . It took a while to do 600 bmps on my little laptop (512 meg ram)
Last edited by [1ponders]; 08-05-2006 at 05:02 PM.
![[1ponders]'s Avatar](../vbiis/customavatars/avatar45_9.gif){kind=avatar}
).
