Terrain 6, 40km with 3.8 million objects

[Dwarden]

nice to You back on the maps Snake , keep it up

[Li0n]

as well as cell size (which is 1/5 the cell size BIS maps use (5 times more detailed)

Chernarus uses 8 m cell size.
Maximum grid size is 8192.
Performance good due to test machine specs.

[SavageCDN]

Well yeah it's a decent test machine but really not that powerful.. I've had a better GPU for the last year or so... and CPU as well.

Thanks for releasing this SnakeMan.

[Clintcom]

This map run's really good Snake man , nice work. I would love to see a desert version.

[ChrisB]

This map run's really good Snake man , nice work. I would love to see a desert version.

He does have a few desert terrains, I have 3 or 4 pos more, most not altogether finished, but very usable and massive fps. The 51km desert terrain has one great city as well (city 32 or 34 can’t remember, will look it out, its up the north there). It’s a very nicely set out city, possibly one of the best for arma 2, plan wise. Great fps for city combat, he may put it here or a link to it, not sure but I may have got it on here a year or two back, myself, wasn't a member then, but been playing for around a decade, so have loads of terrains/island/maps, for ofp, vbs2 lite (only the two it came with), arma and arma2 plus the arma ones that have been re-done to run with arma2. , not sure this one I'm mentioning may have been for arma originally, can't remember, get to my age your lucky to remember how to switch the pc on...

[Bushlurker]

Hi SnakeMan!

Good to see you back indeed! I can start pointing people at some of the useful PMC Editing Wiki pages again now!

An interesting experiment indeed! and informative... There's never really been a definitive "maximum objects" test before to my knowledge... people usually start to get edgy and worried once they hit the million mark, so to see over 3 million in-game and functional is good news - specially for the larger terrain guys, where you can easily hit these sort of figures with a few huge forests!

Another interesting experiment might have been cramming the same number of objects into a more usual sized 20x20km terrain - to see if the increased object density, more of which would be in view at any one time, would have any significant negative effect... Then again, the sheer increase in ground cells in this 4096 terrain - 4 times as many as a 2048 terrain, is usually a significant factor in itself - so all-in-all I guess we could take this as a good illustration of the fact that 3,000,000+ objects isn't necessarily a performance disaster! - that's welcome news for terrain makers!

Well done! and thanks for taking the time to try this!


While I'm here I might as well correct a couple of wrong impressions...

the technology here being its randomly generated/procedural terrain with clutter.

Well, not quite...
It's a regular terrain in the main area, the procedural stuff is "OutsideTerrain" technology - outwith the main limits of the map...

Consider the amount of objects (3.8 million) and the satellite resolution, as well as cell size (which is 1/5 the cell size BIS maps use (5 times more detailed))

Again, not quite...
a 10 meter groundcell size is the "classic Arma 1" size - Sahrani, for example, or some of the earlier quality user-made Arma 2 terrains - Panthera, Fallujah, etc (these being 1024x1024 grids, of course - giving 10kmx10km terrains)...

Sahrani, for example, was a 2048x2048 cell heightmap, with a 10m cell size, giving a 20x20km terrain (BI's biggest Arma series terrain to date, by the way)...

This map is essentially therefore the size of FOUR Sahrani's in a 2x2 square... that's a LOT of groundcells, whatever size they are... that in itself has implications, as SnakeMan mentioned, so thats another part of the "experiment" alongside the silly number of trees...
(As most terrainmakers will know, there's a "bug" in our terraintools which causes some problems with the 4096 size), plus... all those groundcells in a 4096x4096! Especially when they're actually being used in a quite rough and varied terrain...

It's interesting to note that BI themselves have avoided the 4096x4096 terrain size in all their terrains to date...

(chernarus is 20x20km).

Chernarus uses 8 m cell size.
Maximum grid size is 8192.

Chernarus is actually a 2048x2048 grid terrain, but with 7.5 meter groundcells, for a slightly more "hi-res ground features" look - giving a terrain size of 15.36km x 15.36km
Takistan is smaller, but higher-res still - a 2048x2048 grid terrain with 6.3 meter cells producing a 12.9kmx12.9km terrain...

As a final point it's interesting to wonder what BI are planning for the forthcoming Arma 3 Limnos terrain.... reputedly 30x30km+, and I bet they're using at least as detailed a cell size as Takistan's 6.3m... I wonder how they've implemented a terrain of that size???

Anyhow... welcome back again SM! and thanks for the interesting experiment!


B

[BasileyOne]

basically high-resolution/scale islands/terrain in Arma2 are EXTREMELY memory-hungry, so imprving them considerably without 64-bit binaries whouldn't be option anymore(from Arma1 times if not earlier)

[Dwarden]

you do realize that 64bit means double actual memory used, basically need increase of memory size on clients too

so 32bit with LAA which already can do 4GB directly and as (read our older devblog) we can do N indirectly via systems file cache

[Dissaifer]

@Dwarden - I'm really impressed that BIS is still getting up to 4 gigs worth of memory on the 32 + the system cache (but you take an I/O hit). Just curious when the 64-bit versions of Arma are coming... 3? 4? Just messing with you, I'm sure it's in the pipe somewhere.

@Bushlurker - I would actually be more interested in the tightly compacted objects on the map as well.

[Robster]

Hello everyone!

Has this terrain solved that problem aforementioned by Bushlurker regarding null zones where objects can't be placed into 4096 grid format? If it did... kudos on the release and please tell us all how to deal with it