I have developed an automated pipeline (in python) where the user can select any address in Massachusetts and a midi file to generate an image using terragen.
The location is geocoded and the resulting satellite imagery, landcover, elevation, road network, and building outlines are pulled from USGS and massGIS. This data is combined, georeferenced, reprojected to UTM, and resampled as needed to create the image masks for the scene. An unsupervised classification and cross-correlation is done to increase the resolution of the landcover.
The landcover categories implemented are:
water,
field,
deciduous forest,
evergreen forest,
mixed forest,
scrub,
barren,
urban
The foreground mountain is made from artificially generated elevation data (it is not part of Massachusetts). The midi file is read up to create some of the characteristics of the foreground including the color of the path and the flowers. The flowers themselves are placed along the path using a piano roll rendering of the notes in the midi file on either side.
I was hoping to get suggestions on things to change, especially with the atmosphere and lighting, but the scene took 30 hours (!) to render and since a canvas print of this is going to be a Christmas gift, I don't have time to do another high quality render before then.
I will post more scenes generated with different locations and midi files soon.
Enjoy.
~David