ONS-Rail

[Cat1981England]

Thanks Crusha. I'll post back here once we're done

[Cat1981England]

ONS-Rail-Beta-C1 download. If the map doesn't show up in instant action you'll need to download the CEONSS map pack.

List of changes:

- Placed a small building at ground level at the centre of map and lowered the node from the pipework to be inside of it.
- Placed two sewage tunnels which run from just outside each team base and come out inside this central building.
- Moved some of the boxes to build two walls either side of this central building to help divide the map.
- Removed all 64 anti-portals and replaced with 10 more efficient ones.
- Placed invisible collision hulls inside all anti-portals.
- Set custom cull distance for 90% of the maps static meshes.
- Set any possible remaining unseen textures to unlit.
- Removed any possible static meshes which are unimportant and don't impact of the maps visual quality such as the tree trunks around the edge of the map.
- Moved the 6 pipes above the map closer together to make 3 wider ones.
- Turned off collisions and replaced with simpler blocking volumes where possible.
- Hid pickup bases.
- Removed central road.
- Removed pickups such link ammo and mini health packs.
- Replaced all ladders apart from the ones at the four towers with jumpads.
- Moved target painters to the top of the water towers.
- Moved the two super shield pack and super health keg to the top of the towers and deleted classic sniper pickups.
- Greatly simplified bot pathing.
- Removed triggered doors system and some door openings at primaries.
- Replaced weapon lockers with the custom ones (you don't have to throw your weapons anymore).

There's no doubt a few other little thing that's I've forgotten about.

-------------

There's still a really odd 10/15fps drop for me around the central area which Pony pointed out on the first page which I can't find let alone fix. Anyway, I've played a few rounds on our testing server to make a comparison with Beta-15 and the fps has gone up a little, whether this will make enough of a difference with 32p online I don't know.

Using Fraps with a possible max of 91,
Beta-15 fps Avg: 89.1 Min: 55
Beta-C1 fps Avg: 90.5 Min: 76

I'm no expert in map editing and optimising (I actually have no idea what I'm talking about) but at a guess, the main culprits seem to be the wide open central area, the number and placement of the anti-portals and the extremely complex bot pathing. I've tried my very best to avoid making anything look too artificial which I know was a big concern for you Crusha. The central building is just a copy of another building on the map and looks fairly organic in it's location to me. The division of the central area should be a big help in improving the performance at ground level.

I've simplified the bot pathing with our current 9 bots max in mind. This includes removing the door trigger system at primaries in part because the bots get stuck at the doors but also because I don't know how often and what info is sent to each client as they play. I've also removed some of the doors to ehlp teams hold these nodes which should help in both come backs and finishing of the match before over time.

The central node, now that it's covered, should be a big help with come backs and spread the play about a bit. It's health has been set to 1000 and link multiplier 2x so it's built in seconds. If the defending team can hold onto one of their primaries, using the tunnels to take and build the central nodes and then attack the opposition's primaries, a single player (particularly if they're using the Kamakazi pickup) can turn the match around in under 30 seconds

Fingers crossed the performance is sufficient and the endless stalemates are reduced. If people wouldn't mind taking a look after Christmas and posting some feedback that would be great Hopfuuly that all makes sense, I've got some serious wrapping to do, merry Crimbo y'all.

[Zon3r]

Just tried it. played with 31 bots(+me, duh ) they are very good on this map, the performance is way better i think, tho in the middle there is a 60 fps difference if i look toward the center of the map(31 bots make alot of shoting), or when im facing the wall but i guess it's my oldtimer pc. about gameplay, the area is clean, i like it, IMO now it seems that the middle node is a chokepoint, cause it has a huge impact on the match. the match will be a battle for the middle node, or if one of the teams is just slightly stronger a battle around the losing teams primaries, Don't take me too seriously, it may be that i'm looking at thing too pessimistically, but let's see it online, nothing beats the test with real players

[Cat1981England]

Thanks Zon. I think you may be right about the centre node. What about this link setup then?

RailLS2.png (70.08 KiB) Viewed 12952 times

[[RMD]WACKO23]

you all do a great job on maps .

[Snuff]

second that...

[RottenToTheCore]

Thanks for the great job so far .

Hope the changes will work out in practice.

Rail always was one of my favourites, unfortunately no hardware setup ever could handle it :/.

[Cat1981England]

I'll put it on for a test run in a couple of hours when the server full.

[Maniac]

I have an idea about lag. Though my knowledge in 3D graphics is pretty shallow, i have a notion about raycasting.
So, I was wondering what is the difference between Rail and other maps. The closest by quantity of geometrics in FOV is ONS-Icarus: there are many small objects like metal strips and ruined buildings. But, from all the maps i know, only in Rails there are many small objects (metal strips, pipe carrier columns, water tower columns) which are semi-transparent for sight and are placed in such manner that they are always on line of sight between player and objects.
So, there can be problems both with tracing multiple of those at once (we need to hide a very little piece of distant object for each small obstacle) and with computing errors (e.g. a small object shadows a part of another small object which shadows a large object - so resulting numbers are close to computing margin or to predefined clearance).
So, what if all those small items are removed at all (just for testing) or replaced with solid non-transparent objects (e.g. wired holder to solid metal cube holder)? If it is not hard to do, i think it is worth trying on test server.

[Cat1981England]

Thanks Mani. I think you're right, it's certainly worth giving it a go to see if it can help.

It's only been played once with 20+ players, so we'll have to give it a couple more matches to get more feedback and then try and give it another (hopefully final) edit