I would like to ask how difficult would be to add x and y offsets to the excellent chromatic aberration correction tool?
I regularly shoot through a Common Main Objective (CMO) stereo microscope which – by design – suffers from the following flaw: both viewing optical trains are offset from the optical center of the objective. The transverse chromatic aberration is not zero at the image centers like in the majority of common cases, but this aberration center lies outside the images. This way left side TCA is the opposite of the right side TCA. It would help a lot if I was able to give an offset value to the TCA correction routine besides having the red and blue sliders.
Auto CA correction will deal with this if you are working from raw files. As far as I can tell, it tiles the image, determines the best shift for each tile, and interpolates between them with no regard for radial symmetry.
Plus it operates before demosaicing, so it improves the quality of edges in the presence of aberration.
Thank you for the tip. Unfortunately it did not help much: the amount of TCA with the auto correction option was almost the same as without. It is interesting however, that by playing manually with the sliders a much improved reduction was obtained. It would even be better if the slider allowed a higher value to be set.
I still consider the real solution would be a user-settable offset value for both the x and y directions, with a slightly widened range for the blue and red sliders. I think users of shift lenses could also benefit from this improvement.
BTW what is the relation of the auto option with the manual sliders?
Thanks for the clarification. It is clear now what is the difference between the slider-driven and the automatic modes. However in this particular case an enhanced manual mode would benefit more than a purely automatic approach: contrasting edges are not guaranteed to be present in all pictures. On the other hand the numeric values determined for a given setup could be reused for all further images.
I have uploaded a sample Canon CR2 raw file showing the aberration to:
I was unable to upload it to filebin.net (at least it did not show up after the upload).
As you can see the extent of this aberration is very large compared to the usual cases. I suspect this is the cause why the TCA correction routine is unable to process it: it would need much larger correction values than the hardcoded limits. At least these limits could be extended somehow.