Hey Lester, Bird has an answer. I have tried this and it seems to work really well.
I found that you can choose your alignment points and then get the FFT filter up. If you then click on alignment grid numbers where it lists all the alignment points. I found that the FFT went to 30...i then changed it to 9 and then i never got any seams at all.
i have tried it a few times and no seams. i even forgot to stack with feather turned on .. no seam
I even stacked different amounts in the stacking
--birds advice------------
As for working around the "features" in registax, here's a tip - after you've set your align points go to the "general options" -> "fft graph"
dialog and change the fft size from whatever registax has chosen (which is almost always completely wrong) to something sensible like 9 or 10.
The fft size sets some "blurring" to be done on the images before it tries to align them (but only during the align phase, not when stacking). This blurring removes noise and the idea is that the algorithm can do a better job if this value is set accurately. But, for reasons known only to Cor, it picks lousy default values. Larger values mean *less* blurring and so the align box is distracted by noise. I find the default value it picks is something like 20 or 30, way to large. Try
9 and see what happens.
watch the align box during the alignment phase, if it wanders around too much then the value is either too small (too much blurring) or too large (too much noise, cant see the details).
Unfortunately not there and' road with Registax, on the Multi Point.
If you have a very tottering resumption, the MultiPOint you strength's restituiscie a wrong image as yours.
I have tried all the possible ways, on the planets, but at the end I/you/they have reached the conclusion that and' better Iris, for the Planets.
And for the Moon and' excellent the Multi POINT, if you highly have a clear resumption, and very opposed, without starts from a part to the other, otherwise also on
I have been using the largest alignment boxes and the second to largest and staggering them all over the planet until it is covered. This image had only 4 largest boxes, but I did one with 11 of the smaller boxes and that wasn't any better.
Dave, thanks for your input and I will be trying all that, sounds promising.
Have you tried letting Registax pick the alignment points?
I usually try that first, before going manual. With planets it seems that the auto pick likes to work around the limb. It isn't necessary to cover the whole planet image - just the key features.
I also usually use an alignment box size of 128, unless the auto pick ends up with a huge number of points - then I'll clear the alignment points and try a 256 box size, which usually brings it down to 2-4 boxes.
Interesting info on the FFT! I'll try that next time I have a problem with edges appearing in an image (I don't get it very often but I do occasionally, and has reverted to SAP in those cases).
Registax is very sensitive to these settings, and they interact in ways that might not seem obvious...
I mostly use 128 pixel alignment boxes if the seeing is good - the important thing is that each box must be placed around a feature that's (1) easy to see , and (2) fits inside the box. This will depend on your image scale of course, but the grs is an example of a feature that's very useful for alignment, and also the very dark storms in the northern belts are good.
You have to avoid areas where there is little or no detail, or where the detail is only in one dimension (eg a horizontal divide between 2 belts).
It takes a bit of practice to find the best places, and again it all depends on your image data.
I try and set 4 or 5 alignment boxes that are completely inside jupiters disk. You can choose to have your alignment box include black space around jupiter if you like but if the seeing is not so good then the alignment may be thrown off by the movement of the planets edge.
Then, as DP mentions above, when you've set all your alignment points go into "general options" -> "fft size" and change the size from the automatically chosen value to something like 9 or 10. If this doesn't seem to work for you (cause your data is probably different to mine!) try other values, both larger and smaller, until you find the right one that produces the least seams.
Smaller values for this size will make the alignment run less prone to noise, but if you make it too small then it will lose the features as well. Smaller values = more blurring.
Thanks Anthony for your reply, very helpful as always.
I always put the centre of the box, (shown with the zoom option) on a feature with contrast to the surrounds. The boxes seem to lock onto the features okay, so I think I just have to play around with the FFT setting now.
