PhotometricColorCalibration - Anomalous Results with Plate Solved Images

[dcarr]

Hi

I am successfully using PhotometricColorCalibration (PCC), a great addition to PixInsight.  However, I have noticed an apparent issue that should be reported.

PCC seems to be outputting different results for a specific image of mine, depending on whether it has already been plate-solved (prior to executing PCC).

Although I have not tested across many images, this symptom appears (for me) to occur only:
- with my image of 47 Tuc, and
- only if it has been plate solved earlier with ImageSolver, and
- only if I do not select the ‘Force plate solving’ option in PhotometricColorCalibration interface.

So, there is no issue if I do not (pre) platesolve or if I select the ‘Force plate solving’ option in PCC.

I have attached a link to copies of my PCC processes, plus four images I used for testing, including:
- two identical copies of the image (original plus clone), except one has been plate solved
- the resulting images from PhotometricColorCalibration. The one without ‘Force plate solving’ is coloured very differently (I expect old red stars, but the colour appears a bit extreme).

https://drive.google.com/drive/folders/13UiqTjDEd3CrH3ReyaMyh1ZAzilC8go9

The images have been provided in 16-bit format because the 64-bit originals are enormous. Still large, but not as bad. The images were calibrated, combined using ChannelCombination and HDRComposition, and applied with DBE only.

I have looked at the resulting images after applying AnnotateImage and the annotations seem ok to me in all cases. (I was initially concerned that the zillion stars in and around the globular cluster may have confused things)

So, is this a bug, am I doing something wrong, or both?

Regards
Dean

[Herbert_W]

Hi,


I did some test and can confirm the bug(?).

The bug-combination is: prior solved with "ImageSolver" AND unchecked "Force plate solving" in PCC.

Best regards!
Herbert, Austria

[dcarr]

Hello again

Thanks Herbert.

An update. I have slightly amended the title of this post to better reflect the issue with PCC - it does not seem to be confined to a single image (47 Tuc).

I have had the opportunity to test PCC on another target (NGC253) with similarly anomalous results.

In short, if you solve the image first, you seem to need to select 'Force plate solving' in PCC, otherwise you may get strange results.

The attached picture provides a side-by-side comparison. The integrated RG&B channels were combined and applied with DBE. For testing, PCC was then applied to two copies of the image - one with 'Force plate solving' selected, the other not. After stretching, Curves Transformation was used to apply an exaggerated saturation for test/comparison, with no further processing. The image on the left (in the picture) resulted when 'Force plate solving' was selected in PCC; the image on the right was the result when this was not done (very blue).  I have also attached screenshots of the outputted PCC White Balance graphs - which are very different to each other.

For your consideration.

Regards
Dean