Looks spectacular
wlbragg wrote in Tue Jun 02, 2020 3:17 pm:I suppose someone could make a texture scheme for the moon and run terragear using moon elevations and we could potentially develop moon scenarios past and future. From orbit to landing.
Available data might be a bit low resolution for terragear and landing in the way Earth works(?). AIUI elevation data on other planets like Moon (or Mars) is from orbital laser altimetry from satellites? May be low res, especially if it's an old mission. The elevation data is also probably defined with respect to a different reference ellipsoid than Earth (the oval-ish sphere shape of a planet which bulges at the equator due to rotation), and the data might be in a special file format..so may need post processing before feeding into the terragear tool chain, or just converting data to .btg format which AIUI uses a geocentric coordinate system.
The current version is sort of the texture version, a heightmap rendering I think, with photo textures..adding land classification for use with regional texturing might be very hard unless there are already maps released
Adding terrain complexity procedurally on the GPU beyond the (limited?) resolution of laser altimetry data, and whatever source is currently being used, may become easier & faster in future years. AIUI the plan for next gen scenery [
1] involves targeting open GL 3.3 API core profile by dropping support for older API - and presumably extensions beyond that are ok(?). If this plan works out, it should(?) allow using ARB_tesselation_shader [
2] extension to access proper tesselation shaders. These are faster and less bounded when doing tesselation than general purpose geometry shaders. I'd guess the idea would be to tesselate and add more/rougher noise based on a measure of 'roughness' derived from the low resolution data - maybe by post processing it to create a map of roughness and any other useful properties to generate noise. For example it can get rougher as the 'local' gradient gets steeper, or maybe as the local height gets increases. Maybe such a scheme would work by sampling a noise heightmap and adding that on top of the low res elevation data. This higher res result would be used when finding positions for vertices generated by tesselation. The idea being the noise heightmap would add octaves of noise with smoothly varying amplitude in different places based on a roughness map that gives noise properties. Smooth areas would get low frequency, low amplitude noise. Rugged areas would get larger amplitudes across all noise frequencies. If a terrain classification raster was also available, it would be conceptually like WS 3.0:) But it's a far off possibility
Kind regards
FlightGear Forum
by eatdirt » Tue Jun 02, 2020 12:06 am 
