The photo-detection method especially gives an oportunity, where the OSM data are very poor regarding the vegetation.
What I am working on resp. what I am developing is a simple console app for use on MS Windows without a GUI (similar to "ObjectGen").
It works on the base of downloads done with AeroScenery at zoom level 16 with a resolution of 2.389 meters/pixels (actually I only support this level because at this resolution the photo-detection works optimal).
Even though there is no GUI, the handling is hold very simple. You just would have to copy all *.aero and *.png-files of your AeroScenery "16-stiched"-download-folder into the "...\in\"-folder of my app and execute the app. *.tsc- and *.toc-files are generatet and saved in the "..\out\"-folder. There is a possibility of batch processing of all stiched images within a folder and also to change in the config-file e.g. the density and trees-mix.
Also with photodetection you will get trees in many unwanted places - including lakes. Therefore it is recommended to either mask the images using the ereaser (with alpha channel) or simply colorize the unwanted areas in the color blue, red or white before converting, it's your choise.
Even I never tried it out, the placing of trees with your "forest orthophotos" (see your illustration) should work in principle at zoom level 13, this just only without distinction of trees and small trees resp. bushes (... and you shoud use a darker green for this cause light green is considered to be fields/grass).
To try it out you would have to send me your "forest orthophoto".png together with the GEO-coordinates, preferably within the corresponding *.aero-file at zoom level 16 from AeroScenery via direct contact.