What else to do when there's 6 days straight of clouds? Reprocess.
After speaking with Bird about the artifacts I was getting around the edges of Saturn where it meets with the rings, he suggested it could be one of three things:
1. Not enough light being captured
2. Registax alignment not locking onto the features correctly
3. Too aggressive in the processing
For #1, i'll try capturing a brighter image next time.
For #2, I changed my alignment box area (used size 128) and aligned it on the globe with a bit of the rings. Previously I aligned it on the edge of the rings to pick up the sharpest CD.
For #3, I've now reprocessed the same avi in the following way:
- Registax, to the wavelets tab, RGB align (no wavelets)
- PixInSight LE, wavelet processing and noise reduction
- Registax, histogram and gamma adjust
- Photoshop, colour balance and levels
I've gone much softer on the wavelet processing to try and maintain the "globe" shape, and a sure sign of overprocessing is when the Cassini Division gets wider and wider. So this time, i've gone soft to keep a smaller tigher cassini division.
So on the left is the new image, on teh right is my previous effort.
It's a tough call - I like the CD and shape of the globe better in the new image, but I like the banding right up to the SPC in the old image. The crepe ring is probably a bit more natural looking in the new image too.
So somehow I need to take the best out of both images There's a bit of "fuzziness" to the new image as well which I'm not too fond of.
Anyway, for your consumption and feedback.. here they are.
Hey Mike - how large is your raw data? I've been swamped by work and poor weather the last few weeks, but it might be fun to play with someone else's data :-)
That's ok - can you put it on the website and email me the URL? An archive with the compressed BMP's would be the best way to go.
I've had a play in Astra Image with the image you sent me - sometimes I can spend hours trying different settings and processing options, but so far I've tried:
- split colour planes
- unsharp mask, max strength, power 2.0 on each plane
- deconvolve, ME 5 x 1.5
- recombine
- unsharp mask, same as before, on the colour image.
The result isn't too bad... it would be interesting to start with your raw BMP's.
Hi DP - we're sort of in-between planets at the moment, and I'm busy with work stuff as well so I've been a bit quiet lately.
I was hoping to have my 13" mirror by now, but it's still not ready, so that's taken a bit of the excitement out of things - when I finally get it then it'll be full throttle again.
Still a nice shot regardless Mike. I do like Birds rendition though. Not as heavy on the processing. I would say it is the sequence and some of the settings. Wish Adelaide was not in a seeing hole. Even my great seeing down the south coast did not produce results as sharp as yours. Might move to Barbados. Seeing is good there I hear.
step 1 - break each colour frame into 3 monochrome frames (red,green,blue).
step 2: process all the red frames through registax and astra image, make a final red master
steps 3 and 4, same as (2) for the other colours
step 5: recombine the red,green and blue frames in astra image.
Doing each colour seperately allows registax to get a better lock on the details, it's evident that the colours all all slightly out of alignment in the raw frames, this is normal for an object that's low in the sky.
Processing the frames in colour mode makes life tough for registax, as the images all look a little blurry, so I don't do that.
The other processing step in Astra Image was to use the "maximum" filter, it has the effect of narrowing the cassini division back to where it's supposed to be, and Maximum Entropy deconvolution.
I used the commandline netpbm tools to automate the splitting of colour planes, via a short shellscript on my Linux machine (sorta like a batch file if you're using Windows).
I used the commandline netpbm tools to automate the splitting of colour planes, via a short shellscript on my Linux machine (sorta like a batch file if you're using Windows).
regards, Bird
so this program breaks each bmp of mikes into separate red, green & blue. is there a windows version???
I've got netpbm already (from when I planned to do pre-processing a few months ago) - what command line parameters did you use? Better yet, can you email me the shell script? I'm sure I can script it in DOS batch, even though it's been a while since I've played with DOS. Failing that, i'll use one of the unix servers at work It might process it a lot faster than my laptop, too
After creating each master in registax, did you use wavelets, or just save the stacked image?
What parameters did you use for "maximum" and "ME deconvolution"?
Sorry for the questions - next step is using your processing steps to produce the same image myself.. then i'll know how to repeat it for the next images.
I'm sure this procedure is going to add an extra hour to each avi file.. thanks heaps Anthony!
There's a bit of "fuzziness" to the new image as well which I'm not too fond of.
