Registration error: Unable to find an initial linear transformation

A

acreutz

Active member
I have several very similar sets of images (to create a mosaic). A reasonably wide range (about 2deg x 3deeg). Using Blink, there is no obvious differences between the various sets
But for one of the three sets, registration fails, with all but two image (of 31) reporting the error:
"Unable to find an initial linear transformation".
I have never seen this before.
What is it? and how do I compensate?
Thanx
 
Through trial and error I have concluded that increasing the noise parameter (in WBPP) overcomes this issue. Still don't know what it is and why this particular set of frames needed a higher noise setting, but problem is "solved". Any additional information would be appreciated.
 
Image Registration. The frames are take with an ASI2600MC using a Dual Band filter. Scope is at AT 65EQD, binning 2x2, and exposure 10min. These are quite deep exposures and I suspect that there is a marginal difference between very faint stars and noise. Since routinely changing the noise parameter from "auto" to 2, everything looks good; although I do sometimes lose some frames that look visually OK.

On a different note: where has RANSAC gone? I gather that increasing SENSITIVITY in Star Alignment is doing the same thing?
 
acreutz said:
changing the noise parameter from "auto" to 2
Click to expand...
The only "noise" parameter I can find is "Noise reduction" which does not have an "auto" option (initial value = 0).
acreutz said:
where has RANSAC gone? I gather that increasing SENSITIVITY in Star Alignment is doing the same thing?
Click to expand...
The RANSAC parameters that used to be exposed are now fixed (here and in StarAlignment), and are not related to "Sensitivity".
I usually edit the ImageSolver script to permanently increase the parameters to the maximum values (5000 iterations; tolerance 8).
 
You must log in or register to reply here.
Back
Top