I don’t have problems with extents or resolution in terra. SNAP should be able to load the NetCDF4-CF file and confirm that extent and resolution made it into the file. NASA Panoply is also useful for quickly checking NetCDF4-CF files, and should tell you if there is a problem with the saved NetCDF4-CF file. Terra and the netcdf tools in R have been the subject of intensive development work. I’m using R-4.2.1 with:
Package: terra
Type: Package
Title: Spatial Data Analysis
Version: 1.6-7
Date: 2022-08-07
Pixels that were moderately bright/dark in the ROI are brighter/darker in the resampled image. This indicates that colour shift is due to differences in the range of values for each band. You can easily check that. The geophysical variables I use have a global “standard” range and false colour pallet. With terra I set values outside the standard range to NA and chose a colour pallet similar to the standard (I do need to figure out how to use the standard pallet with terra).
[Edit: added URL for Panoply]