Mosaic: still can't get it to work

[Nocturnal]

Hi,

I have two images that I want to mosaic. Both contain the Flame Nebula so I know they have overlap. The only issue I can see right now is that one is calibrated, the other is simply an average stack (subs are still collecting). I did the following:

- load both images in PI
- start staralignment process
- select one of the views as reference frame
- select mosaic mode
- drag an instance on the other image

No solution found. I fiddled with some parameters and no change.

I will wait till I have some more subs, create a properly calibrated stack and try again. If someone sees what I'm doing wrong in the meantime I'd like to hear it.

[Nocturnal]

Well, good news. I think. After doing a quick stack in DSS PI was able to find a match. I guess the lack of calibration caused trouble. I had hoped to use an uncalibrated (but debayered) image to verify my aim in the sky was true. It looks like I need to do a full stack next time.

What I did notice is a rather undesirable seam between the two mosaic'd images of about 1 black pixel. Is that to be expected?

[Juan Conejero]

I guess the lack of calibration caused trouble.

Indeed. Excessive noise and/or hot pixels may fool the star matching algorithm. You can increase the Noise scales parameter to remove very small and bright features before the star detection phase.

More interesting to improve star matching when building difficult mosaics, is to enable the "Use scale differences" option (Star Matching section). In this way the star matching algorithm will be much more robust, and since the number of outliers (false putative star pair matches) will be much smaller, RANSAC will be able to find a valid set of star pairs under much more difficult conditions.

Note that this option cannot be used when there are significant scale differences between the images (the scale tolerance parameter allows you to set the required limit, if necessary). However, when you're registering images taken with the same focal length and sensor there are no scale differences, and enabling this option may be a very significant improvement.

What I did notice is a rather undesirable seam between the two mosaic'd images of about 1 black pixel. Is that to be expected?

Absolutely not. Are you sure that 1-pixel column or row isn't in one of the images before building the mosaic? If not, then I'll be asking you to upload your images

[Nocturnal]

Hi Juan,

ah, the stack was 'raw', not yet cropped to take out the few pixels at the bottom where there were stacking artifacts. Forgot about that, sorry! Thanks for the quick analysis

Will try again with the stack I'm about to create.

[Nocturnal]

That was it Juan. Slightly embarrassed I must admit The new mosaic doesn't have that line anymore. In fact it's quite amazing that PI was able to stitch them together as there was a slight difference in optical train between the two images. I changed the spacing between FR and camera to flatten the field a bit better.