QUOTE (mcaplinger @ May 6 2020, 05:21 PM)
Bjorn, could you describe your workflow for getting the brightness matching to work so well from image to image for these products? I've been using a simple lambertian removal and it doesn't work very well at all, at least for the full range of geometries over the orbit.
These image products are several months old and involved a significant amount of work in Photoshop to reduce color and brightness seams - too much work for my taste. As a result I have made changes that greatly reduce the manual cosmetic enhancements and post-processing needed in Photoshop. What I currently do is this:
(1) Added flat fielding in December 2019. This change is probably what resulted in the biggest improvement in brightness/color matching in the overlap areas between adjacent images (not framelets). Without flat fielding the images are a bit brighter near the left/right edges relative to the center than they should be. Flat fielding also greatly reduced the seams between adjacent framelets. The flat field I'm currently using can be found
here.
(2) Use a slightly modified Lambert function, i.e. cos(i)^0.95 instead of the usual cos(i). Earlier I was using a modified Lommel-Seeliger function but the modified Lambert function works better in conjunction with (3). Areas near the terminator require special handling though since there is faint illumination on the nightside near the terminator.
(3) Multiply the result of (2) with a new/experimental limb darkening function that is a function of the emission angle. This worked almost perfectly for PJ24 and PJ25 but results are slightly worse for PJ26. It also worked rather well for making a quick-and-dirty global map of Jupiter from the G1 marble movie images. I'm now in the process of modifying the function to include the phase angle (its 'weight' must change with phase). An additional complication is that the photometric properties of the polar regions are different from areas closer to the equator. I plan on making some of the photometric parameters vary as a function of latitude. Also I'm currently using identical parameters for R/G/B in the limb darkening function but this will change.
(4) Following the previous processing steps I have several cylindrical maps that I mosaic in Photoshop and decide which one(s) has priority in overlap areas by cutting away parts of some of the mosaics in the overlap areas. This is fairly trivial and could probably be automated.
(5) Following (4) there are usually some residual color/brightness seams in overlap areas that are close to the terminator and/or limb but these are usually not big and can be easily corrected in Photoshop by modifying the brightness and color balance of one or both images near the overlap area. This processing step used to take a lot more time before the changes described in steps 1 to 3.
It should be noted that I haven't tested this for the full range of geometries but I suspect it would work well once the phase angle is taken into account in (3).