Jump to content

Developer Discussions: Alpha 17


Developer Discussions: Alpha 17  

1 member has voted

  1. 1. Developer Discussions: Alpha 17

    • Newly Updated
      1
    • Check out the newest reveals by Madmole
      0
    • Over 100 new perk books with set collecting and bonuses
      0


Recommended Posts

So, I played Navez for a few days and it is my 5th (I mean it) time playing it since A11.

 

Boys, I mean Pimps, you would better put some awesome soundtrack, much more awesome than Galaxy News Radio. We wanna have the best soundtrack for a game here. Thank you :)

Link to post
Share on other sites

I was wondering about turrets and why there is no NPC Turret versions yet, that will attack players.

 

Some Survivors (or bandits) could set up some of those as trap (like the mines) and players woud have to be aware of them or sneak around.

You would only need another turret block thats targetting player(s) and cannot be controlled and needs no ammo.

 

Would maybe tend to people using them to remove zombies as well, maybe kind of exploiting them (building the player base "behind" such a turret). Maybe avoidable if the ammo is limited only.

 

Well, just throwing thoughts into the pool of replies :-)

Edited by Sam_Neill (see edit history)
Link to post
Share on other sites
I was wondering about turrets and why there is no NPC Turret versions yet, that will attack players.

 

Some Survivors (or bandits) could set up some of those as trap (like the mines) and players woud have to be aware of them or sneak around.

You would only need another turret block thats targetting player(s) and cannot be controlled and needs no ammo.

 

Would maybe tend to people using them to remove zombies as well, maybe kind of exploiting them (building the player base "behind" such a turret).

 

Well, just throwing thoughts into the pool of replies :-)

 

Bandits were cut on a17.

Link to post
Share on other sites
Im not talking about bandits, but possible turret placement, left behind in POIs, which attack players and/or zombies (bandits only mentioned for a storyline/reason for their placement).

 

I love the idea but oh my god can you imagine the level of "I'm never playing this game again" type of posts we would see if a digging zombie is worth a thousand salty tears?

Link to post
Share on other sites
FAATAL, I know mod talk is kind of ot for this thread, but it's the only way to get your ear.

 

Kell managed to get all the old uma zombies working in a17, well over 100 community created zombies.

 

For performance reasons you do need to turn the uma texture res down, but performance is pretty much the same as vanilla with that setting.

 

There are some new zombie properties the old uma zombies don't have, and I was hoping maybe you could find the time to add the properties to the uma zombies?...

 

I have never looked at the UMA zombies, but my guess is that the main problems you mention are with prefabs. Zombie/animal prefabs have some changes like a large entity blocker GameObject was added. There may be layer differences. Anim controller states and scripts have changed quite a bit. The old stuff would have to be diffed and adapted, which I won't spend time on.

Link to post
Share on other sites
You got that in? Well done! (no sarcasm intended) :-)

 

I'm curious if it only allows for jumping or also allows for direct pathing over vertical and inverted surfaces?

 

Links are made in a horizontal direction, but with height variance, so it works on a slope, but once too steep it assumes you cannot walk it and does not link. A vertical surface is not pathable, unless it is a drop of a reasonable distance, then it just links down. It does not path along ceilings. The more nodes, the slower it gets and more memory used.

Link to post
Share on other sites
Thats true it is an design decision. And I think a pretty important one to figure out.

 

I bring it up, because I am not a designer. I program. Sure I design stuff as I program, but I don't make the final call. Some of my stuff is left as is, when the team likes it, but other things are changed if deemed too easy/hard/weird/confusing/whatever. Designers/management/owners make the final call in your typical game team.

 

The way I see it is you could:

 

1. See every attempt to not fight zombies on horde night "cheesing", And try to fix it.

2. Allow bases to be unbreachable, but not balancing the cost so that there would be no way to repair the damage by next BM (This would mostly apply to SP)

3. Making the top tire materials unbreachable or easily repaired.

4. Balancing ammunition cost for turrets so they can defend a BM and be filled up to the next one.

 

If 1&2 are true. I think we will start seeing more BM combat bunkers, away from the real base. Because there would be no purpose of building a base with bad ass protection since it would be to costly to repair in time for next bm.

The combat bunkers would basically function as a stationary temporary armor.

 

If 3&4 are true. It would give endgame or top tire material a real purpose.

On the other hand we might also see some BM combat bunkers with Lower tire materials for those who wants to go 1 on 1 with the zombies. Or maybe they will decide to put points in other perks than top tire materials?

 

Im not sure more settings are necessary. I mean If I want to have a harder play through Im not going to build a 20x concrete wall filled with turrets and then complain about zombies not breaching.

Ill strip down butt naked and face those bloody dead heads with my FISTS!!!

 

And a side note: "take a nap" is not what I think the ppl who are voicing these opinions are doing. Mining, building, cooking, repairing. Basically playing your game is what I bet they are doing.

"taking a nap" and logging out is what i think ppl WILL do if no place is coincided safe....Well AFK is :D

 

Designers will decide how they want that to behave in the default gameplay settings. I and other team members are thinking an AI block damage multiplier setting might be all that is needed to cater to those who don't like the default. It is on my todo list.

Link to post
Share on other sites
Links are made in a horizontal direction, but with height variance, so it works on a slope, but once too steep it assumes you cannot walk it and does not link. A vertical surface is not pathable, unless it is a drop of a reasonable distance, then it just links down. It does not path along ceilings. The more nodes, the slower it gets and more memory used.

 

I know the pain of A* well, I've written it from scratch a few times (which is why I'm amazed you used it). It looks like I'll have to do it again to use the new job system in unity.

 

Have you considered declaring no nodes at all for the node map? Just having everything on a virtual grid with cells only set if they are to be used?

 

It should be a little easier on memory for the minimal trade-off of having to set the weight of the node during pathfinding. Given the underlying voxel engine I have little doubt there are fast helper functions to get a blocks properties quickly.

 

Updated: I'm pretty drunk so I'll add that if I can make A* pathfind over 500,000 km to an accuracy of 500m at runtime* you can do 3d pathfinding within a 0.5 km radius :-)

 

* the trick for that is to use layered smaller grids instead of a single large grid.

Edited by The Gronk (see edit history)
Link to post
Share on other sites
Im not talking about bandits, but possible turret placement, left behind in POIs, which attack players and/or zombies (bandits only mentioned for a storyline/reason for their placement).

 

Why would this happen before bandits? Did it ever come up as a thing from TFP?

 

You are conflating ideas from what is supposed to be with what you want.

Edited by stasis78 (see edit history)
Link to post
Share on other sites
Have you considered declaring no nodes at all for the node map? Just having everything on a virtual grid with cells only set if they are to be used?

 

It should be a little easier on memory for the minimal trade-off of having to set the weight of the node during pathfinding. Given the underlying voxel engine I have little doubt there are fast helper functions to get a blocks properties quickly.

 

It is a layered grid, with layers allocated as needed. Normal links are just a byte in a struct, not an allocation. Custom links are a ref, but are used sparingly. Memory use is fine as it is. Cost calculations are not cheap since 2 blocks are looked at. It would be at least 50x slower to calc the cost vs using a precomputed value. Memory use is a factor, but not the prime concern. Raycasts and block analysis are the biggest overhead.

Link to post
Share on other sites
Did you find the job system function for mass raycasts? Are/have you considered using it?

 

https://docs.unity3d.com/ScriptReference/RaycastCommand.html

 

Not of much use for the order that grid locations get updated. It is an iterative process that requires hitting mesh colliders to analyze blocks to then know where the next rays if any should be cast. RaycastAll would be nice, but it needs to be able to hit the same chunk mesh multiple times which it does not. It really does not matter at this point since the system is fast enough for the area it needs to scan.

 

I'm hoping Unity 2018.3 will give us some speed boost with physics (as proclaimed) with little work.

Link to post
Share on other sites
Not of much use for the order that grid locations get updated. It is an iterative process that requires hitting mesh colliders to analyze blocks to then know where the next rays if any should be cast. RaycastAll would be nice, but it needs to be able to hit the same chunk mesh multiple times which it does not. It really does not matter at this point since the system is fast enough for the area it needs to scan.

 

I'm hoping Unity 2018.3 will give us some speed boost with physics (as proclaimed) with little work.

 

The unity engine is essentially being ported to the job system, more accurately, we are allowed to use the system they've implemented internally. It's the best thing they've done in many years.

 

For long-range pathfinding would it not be better to use a separate layer of cube colliders? This would solve things like the "jail door" problem. The overhead may be a little much though.

 

Very short-range pathfinding should be removed from the A* system altogether, the chess-board movement is especially noticeable then. I shouldn't have to look at my compass and align myself to the cardinal directions to predict enemy movement within a handful of metres.

 

"It really does not matter at this point since the system is fast enough for the area it needs to scan."

No, it isn't. You already admitted that it's a bottleneck. No system is fast enough that it doesn't need improving.

Link to post
Share on other sites
I have never looked at the UMA zombies, but my guess is that the main problems you mention are with prefabs. Zombie/animal prefabs have some changes like a large entity blocker GameObject was added. There may be layer differences. Anim controller states and scripts have changed quite a bit. The old stuff would have to be diffed and adapted, which I won't spend time on.

 

Ah well.. nothing ventured nothing gained. I think modders are looking into adding more of the vanilla type zombies anyway, so maybe we'll see improved versions of the old gang. Thanks for the reply though!

Link to post
Share on other sites
"It really does not matter at this point since the system is fast enough for the area it needs to scan."

No, it isn't. You already admitted that it's a bottleneck. No system is fast enough that it doesn't need improving.

 

I said:

"Cost calculations are not cheap"

"Raycasts and block analysis are the biggest overhead."

 

Did I compare it to other systems or provide absolutes? No. I would say the same thing about every system I work on, since they all have some bottleneck. If the bottleneck is .001% of CPU, do I care? No.

 

Every game is full of systems that are fast enough, because they obviously shipped them and they work. Sure everything can be improved, but it is called diminishing returns.

Link to post
Share on other sites
Guest
This topic is now closed to further replies.
×
×
  • Create New...