Mosaic StarAlignment Issue

[brew]

I am having a strange situation while trying to align panels of a 2x2 mosaic. I assume I am missing some parameter or something. I have done a couple of mosaics in the past so I am "somewhat" familiar with the procedure.

Hopefully I can attach some screen images to show the problem. Basically, when I align an individual panel with the scaffolding image, the intensity is knocked WAY down. The images are so dim they are practically unviewable.

I created the scaffolding image using StarAlignment with 2D Surface Spline, distortion Correction enabled, Register/Union - Mosaic mode, Ransac tolerance 6.0. Other settings are default. I am running PI 1.8.04.1195.

Example 1: RGB panel alignment.

In this case I am aligning an RGB image with the scaffold using the settings shown in StarAlignSettings.jpg: 2D Surface splines, Distortion correction, Register/Match mode. Basically the same as used for creating the scaffold, except using Register/Match mode. The RGB is one of the same panels used to construct the scaffold. The RGB image is linear.

I have attached two images showing a crop of the panel to be aligned (Before.jpg) and a crop of the aligned image (After.jpg). I have attempted to size the images similarly for comparison. Both have an unlinked STF applied.

The After image is much dimmer, especially in the red. Pixel values at a couple of corresponding locations are

Location               Before   After
A (bright Star)            0.97   0.92
B (Nebulosity knot)         0.0145   0.0020    <=====
C (Star right of center)   0.55   0.46

The stars are attenuated somewhat, but the red nebulosity is attenuated by a factor of 7!

Example 2: Lum panel alignment.

In this case I am aligning Luminance images using the same scaffold used in the previous example, using the same StarAlignment settings. The Lum panels have already had some stretching applied. Again, the original Luminance image (BeforeLum.jpg) is much brighter than the aligned image (AfterLum.jpg).

Here are some readings from corresponding points in the images:

Location               Before   After
A (bright Star)            0.99188   0.0011
B (Star Upper Left)         0.9697   0.0011
C (background)            0.3      0.0009

The image has lost almost all signal!

[brew]

And here is the StarAlignment settings.

[RickS]

Are you registering stretched data against a linear scaffold image?  If you do that with Frame Adaption turned on it will try to match the intensity of the scaffold.

Cheers,
Rick.

[brew]

The scaffold is created from linear data.

In the second case the panels are stretched. Did not realize Frame Adaptation would do that. Does this imply that building the scaffolding with stretched images will cause the resulting aligned images to be stretched?

In the first case, both the panels and the scaffold are using linear data.

[RickS]

Frame adaption won't do a non-linear stretch - it does a linear fit.

I can't think of any reason why you'd have problems if the scaffold and panels are all linear.  Might be worth turning off Frame adaption as an experiment though.

Cheers,
Rick.

[brew]

You are right on - removing the Frame Adaptation from the piece alignment worked like a charm.

Thank you!