While importing large resolution mask image to surface material the program stalls/hangs or crashes at the near end of the import process.
imported image resolution 65025 x 65025 px (same as the tiled mask would cover at 512x512 tile)
Ram use peaks at 43GB at the time of crash (out of 64GB)
Terrain is also large and using the 16384 x 16384 heightmap at 4m grid size (65536 meters)
Import large resolution png mask into world project. Specs below.
world specs:
Blocks and Tiles
Tiles:
128 x 128
Blocks per tile:
4 x 4
Blocks total:
512 x 512
Height Map:
Planar resolution:
4 m
Block:
33 x 33 vertices (128 x 128 m2)
Tile:
129 x 129 vertices (512 x 512 m2)
Total:
16385 x 16385 vertices (65536 x 65536 m2)
Height resolution:
5.000 cm
Height range:
-491.500 ... 2785.250 m
Satellite Texture:
Tile resolution:
128 x 128 px
Tile texture border:
3 px
Tile overlap:
1 px
Total:
15489 x 15489 px
Resolution:
4.231 m
Surface Map:
Tile resolution:
512 x 512 px
Block overlap:
1 px
Total:
65025 x 65025 px
Resolution:
1.008 m
Normal Map:
Tile resolution:
256 x 256 px
Tile overlap:
1 px
Total:
32641 x 32641 px
Resolution:
2.008 m
heres few more attempts with as minimal world setup as i could make.
world - genterrainentity - world settings set - import heightmap
At ~8K import heigthmap got in, trying to resize that into 16k -> crash
rest trials of importing mask at higher res also crash.
Crash now not as sever as before (black screens, complete hangups, gpu driver crashes with recovery) changes to before : updated gpu driver from latest stable to latest beta, manually set lot more extra virtual memory (running out should not be an issue except if the program cant handle more than 32gb)
Also this was suggested in Arma Discord:
Could be the version of libtiff they are using is outdated, iirc BigTiff >4GB format was only fully/formally supported since the 4.0.* version of libtiff
ref: http://www.simplesystems.org/libtiff/releases/v4.0.0.html
Perhaps something that can be looked into
As a suggestion, import of large textures via tiled smaller resolution images could be less straining process.
