Hi There,
Well after trying the Star Align tool to make a base layer mosaic to use as a registration layer for the panes, I have had a lot of problems. Thanks to Georg for some info and sending links to previous threads on this. I want to add in my results so far.
I'm using 2 sets of data. I started with the larger Orion project of 15 frames but after some issues I decided to go and practice on the 4 pane mosaic from Lagoon to B72. As these were 4 frames in the same row, I thought it better to play with these, and there is only one side to the stitching, so should be less complicated. They are 4 11meg files from a Tak106N.
What I found was that I could get 3 frames to star align on default settings, but the 4th would not find a valid intersection. I tried all variations, panes 1, 2, 3 then 4. 4, 3, 2 then 1. 2, 3, 4 then 1 etc.. It was only after I started using the previews that I could mosaic all 4 panes.
Here is a screen shot showing the preview in use but on a 5 pane Orion attempt. On the left is the reference 4 panes already stitched. I drew a preview of the overlap with the 5 th frame to be added. I drew a preview over the same overlap area on the 5th pane. Selecting the Preview from the already stitched 4 pane for the reference in Star Align, and with default settings, Frame Adaption ticked, Mask ticked, and Mosaic Union selected, I dragged the blue triangle to the Preview tab on the 5th frame. See the 1st attached pic.
Now when I tried the same techique with 15 frames in Orion and I started to get washed out frames, as Paulo had described in previous posts. He said to uncheck the frame adaption. When I did that I got the result in the 2nd attached pic. One washed out frame, one frame looking very clipped. The data is fine in this pane. It was the 8th frame I tried to add to the mosaic layer. If I started with that pane and built up from it, it looked good. Again the more frames I added the more washed out some frames became. I ended up doing lots of restarts, and somehow managed to get a final result of attached pic 3. Its cropped, and stretched, but its just an example. Some of the frames were not DBE'ed very well. I started all these tests to see if the Star Align worked for me. About half of the frames will need more accurate DBE and I will do that this wekeend I hope. So hopefully on a full redo of the base layer and with proper DBE in all frames I will be able to get a better result.
Ok so then I moved onto the Star Align for pane registration. Using the 4 pane Lagoon to B72 base layer I tried to register the 4 panes to the base layer. Again this failed. RANSAC iterations ran into the 20's. So I ended up aborting the process. Further reading and more threads from Georg and Paulo led me to try previews again. This time I drew a preview in the 1st pane. Say the Lagoon. I then drew a preview in the base layer correspoding to the same overlap area.
http://pixinsight.com/doc/tools/StarAlignment/StarAlignment.html#usage_011 Again I selceted the base layer preview 1 as the reference picture, and dragged the blue triangle, again to the preview tab on the Lagoon pic. I used the Mosaic Register option, and Projective as the interpolation option. All 4 panes registered. Success! I redid the files trying the 2D option, (as in the release notes showing the preview it suggested to use this option. Personally I think the Projective pic looks better. It seems to be less washed out.) Both pics see attached pic 4, show good round stars.
So then I was able to try out the Gradient Merge Mosaic (GMM) tool. Again issues. I did 2 versions. One with the 2D interpolation version, and the other with the Projective interpolation version, both with the overlay selected. See attached pic 5 for the result. The stitching looks good, but the stars have ringing on them, and they are elongated vertically. I don't know why this happened. From pic 4 you can see that the files are both good. All 4 panes are good in fact. The registration did work. So I tried again with the average option. This gave much better result, but there is some double overlap areas in one of the panes. See pic 6 for that result. I have run out of room for more pic attachments so I'll put them in a new post.
In short am slowly getting to the opint where i can use both tools to make my mosaic, but I'm still stuck. Making the base layer was trial and error, and I'm not even sure how I managed to ge tthem all to fit. I just kept going putting it together and didn't make any notes on it. The GMM issue might just be a matter of playing with the sliders. In the meantime I need to figure out the mask opacity to check the registration of the files.
If anyone has any idea on how to help I would love to hear your ideas. Hope the pic help others who might be having similar issues.
Tom
