Step 1: Initial Alignment Load your captured movie or series of images
into Registax. I primarily use a Meade DSI Color, so I
use a series of .bmp images. Either through a Windows or
Registax bug, I can only open about 250 at a time using
the "open" ("select" in Registax)
dialog box, but you can just drag and drop the files from
Windows Explorer. 
Select an alignment box that is a little
larger than the planet, click the center of the planet,
and then click "Align". Registax will do the
initial alignment and rank the pictures by quality and
alignment error (the graph to the right). Amazing, how
mediocre the unprocessed frames are, isn't it?
The slider near the bottom of the window
determines how many frames are kept. For this tutorial I'll
keep a couple hundred. Up to about 2/3 of these might be
usable, though. It depends on how many you have and how
bad they are, but obviously, when the quality curve
starts to drop off a lot, they need to be rejected.
Click the "Limit" button.
Registax will cut unused frames and move to the
Optomization section.
|
Step 2: Optomization Click "Optomize". Registax will
correct alignment errors as best it can.
Click the "Stack" tab to move
to the stacking section.
For expedience, you can just click the
"Optomize and Stack" button, which will skip
you to step 4. That works fine when you can see from the
initial alignment that you won't have any major alignment
errors and won't need to reject any more low quality
images.
|
Step
3: Stacking
Click "show stackgraph". If you want, you
can cut off a few more frames using the sliders for
quality and registration difference. Sometimes there are
anomalous spikes (there is one in the initial alignment
in step 1) and if they aren't corrected, those frames
will need to be cut out because they won't align and
stack correctly. For this example, the graph looks pretty
good and no further adjustment is necessary.
Here I've also clicked the "expand image"
option. This uses the entire width of every frame added
together. If there is a lot of movement, that may mean
you only have a few frames covering part of the image and
that can degrade quality. Here, only part of the right
side is cut off in some frames and it isn't noticeable
when they are stacked.
Click the "Stack" button.
|
Step
4: Color Correction
This I just figured out. With the DSI-C, blue and
green sensitivity is weak so I boost them both by 1.5x,
the maximum Registax allows. It gives a more natural
color balance than in many of my earlier pictures. My
first came out too red, while later ones I balanced in
Photoshop and they just didn't look natural. Registax
only processes part of the image at once, which is fine
because it provides a "before and after" view.
|
Step
5: Histogram Stretching
This image is slightly dim, so by stretching the
histogram (moving the right-hand slider to the left and
clicking "stretch"), the image is brightened.
The histogram lists the brightness of each pixel. The
large spike at the left is for the black portions of the
image. If the spike isn't all the way to the left, the
image will look grey, so move the left-hand slider to the
right to blacken the background. The wider, flatter hump
just left of center represents the pixels in Saturn.
|
Step 6:
The Magic of Wavelet Processing
I don't fully understand what wavelet processing is,
but it appears to be a local contrast enhancement
algorithm. It detects and amplifies small differences in
contrast and it is amazingly effective. The top slider is
coarser detail and the bottom one finer detail. Usually
my wavelets look like a bell curve, with the top slider
being most sensitive to overprocesing (it'll look grainy
or pixelated if it is overdone). The more images you
stack, the more you can amplify the detail. Use trial-and-error
to find what looks best.
|
Step
7: Finished!
Click "Do All", then save the image.
That's it! There reallly isn't much to it. It only
takes about a minute or two of input, though if you are
stacking a thousand subframes, the initial alignment,
optomization, and stacking steps can take several minutes
for the computer to process. With larger images, the time
is multiplied.
|