引用自 Oswalt：

Do human snipers firing at or killing nearby bots kick them into high adrenaline but cause them to target the nearest spotted human player to them instead of the one firing further away?

I couldnt tell you how they prioritize targets but yes the adrenaline thing is a real feature. They may target you because a bot radio'd your position, they used their own tracking, or a bot nearby told them etc..

引用自 Oswalt：

This happened on rainforest today, sniped 2 bots next to an RPG bot (at the edge of the grassy grounds where rainforest meets beach). RPG bot spotted me way back at the starting edge of the map (water apc spawn) and let one loose at me before I shot the 3rd bullet. 191m rocket kill, he gets me respects =p. At this range enemies are almost dots on a 7x scope.

I tried to get up and run but only moved 3m before the rocket sent me to sniper hell xD

If you were back by the APCs, the edge of the map is right there so the RPG couldve missed u by a lot and still killed u because of the invisible sky back wall. Bots only use RPGs on targets of 2 or more people in a 5-15 meter radius. Also after that RPG is fired from the bot, its is essentially on a random dynamic flight path. Every 1/4 second the RPG is nudged in a random direction. Thats why sometimes RPGs go way off or get nudged and seem to curve right into your face like a laser directed it haha.

Thanks for the extensive feedback.

There is two ways to view this but first let me state that bots do not wall hack or track after losing LOS, they suppress last known positions but within a cone of fire. Now there is history of some bots shooting through static props and it looks really hacky at times but speaking with the NWI AI programmer over the years and spectating 100s of hours, I believe it is still almost entirely suppression fire.

There are two scenarios that this is most obvious in:

Scenario 1:
Player A pops out of cover, bot spots, bot fires up to ~5+ seconds of suppression fire at last known area (this is a cone of fire so it inevitably will hit the cover of the player and bullet penetration of an APC is the same as a tiny pickup truck.. its just thickness of the object). The APC for example in rainforest and convoy etc. have invisible "bullet blocking" brushes that act as near impenetrable armor and are up to 12+ units thick in some cases. Here is a screenshot example from editor[www.dropbox.com]

Scenario 2:
Bot pops out of cover, while suppressing, bot strafes back into cover but continues to fire in direction of player, still lined up regardless of prop in front of him. Youll notice this much more when spectating, especially when a bot is suppressing and then does a complete 180 and continues to suppress, hitting random teamates and walls, ground, etc.

I've increased the min/max time to suppress and I believe bots are slaves to finish/end their suppression loop when stated in the code due to this increase.


As for long distance shooting, bots have a long-distance cone of fire that shrinks with time. It normally takes around ~7-9 seconds for the time they spot you and start shooting. This resets if you go in cover for so many seconds. Obviously stance matters big time. Bot also hold breath when firing.

The reason why its still fair is we don't give bots scopes due to possible bugs with high population and constant respawning. Apparently when bots have scopes, they zoom in and one shot players after a shorter amount of time has passed, especially if player is stationary.

As stated before, it varies since there is so many maps old and new, a wide variety of scenarios, custom props, bot code we can't see to 100% verify. My advice is to spectate a long time on many maps, but don't go in solely looking for the evidence you seek, look at all the times an RPG/grenade/bullet misses or when a bot is just being plain dumb as rocks. You'll see that most of the time, the bots are terrible but have instances of glory but also sometimes bullshit moments too. Most of the time its the same suspects that rack the kills, shotguns, juggs, explosives.

引用自 Panda：

But But..... You have "binos," attached to something, that shoots bullets.

It would make more sense that your guns have shorter range scopes and bino's could be used for long-range spotting. I typically dont like using longer range scopes unless the map is just super open.

Thanks for you reply, its been a long-term topic discussed on here so I will lock this thread as I do have plans that have and will further address this. The most objective thing I can say is that the armor the bots wear, we get the same armor except for the juggernaut. As for reaction etc.. the AI code is complex and factors in things like suppression, fellow dead AI, sounds, objective, AI spotting, adrenaline that affects bot reaction time but makes aim shakier, close/far distance, etc.



Some discussions on this:
https://psteamcommunity.yuanyoumao.com/groups/SernixINS/discussions/0/133259956019857099/

https://psteamcommunity.yuanyoumao.com/groups/SernixINS/discussions/2/343786746001894443/

Some of my plans to address some of this but also expand our server(some of this already is implemented):
https://psteamcommunity.yuanyoumao.com/groups/SernixINS/discussions/0/1290691937724770832/

No point resetting etc.. but I get your point.. Stats was never my intention to track but I can see the appeal aside from being top leader board (like detailed stats such as accuracy etc.). Its just one more thing to maintain if it breaks so no thanks.. Gametracker tracks regardless if I want it to or not so we just added to MOTD page at one point.

引用自 Jpoon：

I was on the Axin server and W3 was testing a Zombie mode where the bots would charge at you with a knife. It was super fun. I think Sernix should try that out but make it more challenging by doing head shot only in order to kill the bots.

This is possible but thats time taken away from other features. I might look into slipping them in when I redo the bot models/skins.

引用自 budweiser：

Melee bots sounds like fun, I missed the test by the time steam downloaded axin's theatre :P

Can the game director detect blown cache objectives? Maybe like ambush events we can have the director report enemy supply shortage and spawn melee bots for the next 3mins.

Great idea! We could do something similar to this in future.. Like some sort of supplies around map and if found, bots will have shittier equipment like less primary weapons, less grenades/rockets, etc.. I don't think we would do something like knives only but we can have a few bots that do that too, maybe give them machetes and other weapons etc.

Some good points on scalability. Its especially hard to motivate a lower-pop server to play important classes and for the team to proceed without said classes. What I may do is scale certain things like lives based on how many medics or engineers slots are used but only in a way that does not motivate players to avoid those classes. Ill brain-storm this a bit more to weigh pros/cons.

Suppression against bots wasn't changed but in general bot aim and adrenaline effects have been tweaked (pros/cons for bot aim/reaction time etc. in different situations) which can make them seem like suppression doesn't have as much an effect in certain scenarios, especially closer range I'd imagine.

If a medic feels hes overworked and rages and vents on others.. you can report him, preferably with screenshots or recording. If people dont like reviving/healing people, dont play medic (I personally love it and never feel overworked, its my favorite role). If only those slots are open and that person doesnt want to play medic, ask another to switch. Medics dont have to revive players in risky situations either and if people berate the medics for not going in super risky areas, those people are up for review as well.

In terms of easing off their workload, this was addressed way back when with the ability for any player to revive any player, regardless of their wounds. Round saves and respawn lives of course ease this too but I agree scalability needs slight more improvements and right now I believe players simply just get more lives at lower pop (along with quicker respawn times).


"Also suggesting map gives 0 retries, that way players don't disconnect when round 2 of a mistakenly chosen map happens.... which is way too often."

Can you elaborate on this please? I dont understand.

The test server is Axin's year long work that branched off of a base build of Sernix. I setup the server for him and the community to work out bugs and give feedback for things we can implement in the official build. We will only bring pieces of his build over because there is simply too much going on. I rarely do big overhauls that change every aspect of Sernix because it will literally burn me out trying to maintain it, fix balance/bugs, and implement new things.

引用自 Nimbus：

More distinguishable uniforms/camo/gear?

Thats a lot of work and even then.. The positive is you know how well u do as a medic vs an MG!

If you guys know of a custom server that uses it, I wouldn't mind implementing. It would need to be hacked in because base game engine doesnt support it but Id imagine itd be easy since u can just create a weapon that doesnt have ammo but has a bino zoom. It would likely take a weapon slot but u can use it as a 3rd weapon that weighs virtually nothing.