Jump to content

Alpha 20 Dev Diary


Recommended Posts

iv been digging from in art station and found the person who dose the 7dyd zombies

and i found a few cool things some are not done but. oh well!

I also found a shark but.... i don't think thats real LMAO
 

Spoiler

Noah Trader hipoly.
Trader Noah(Up)
trader Bob(Down

Old Trader hipoly.
thick boi
Fat Zombie lopoly.
haz mat

Nurse

 


7 days to Die Zombies
Hazmat Zombie hipoly.

 

Edited by Adam the Waster (see edit history)
  • Like 2
Link to post
Share on other sites
On 5/14/2021 at 9:13 AM, faatal said:

It is not a percent. Destroy area mode is when they can't reach you, but it can also happen when they fall. The blocks chosen are not totally random. It looks for supports and it is randomly between the player and zed, but supports are often not near the player, so it looks in a 10x10x10 area. That is already fairly slow trying to make a quick decision on what to destroy.

 

Zombies types have random weighting for how much they care about block health. It creates some path variation, not radically different paths like some people want. That would require some very different rules.

Door are already weighted so zombies prefer them over walls with equal health.

 

The game does not know if a block is player placed. Each block only uses 8 bytes of data. With the amount of blocks in the world, it quickly adds up. I have thought about adding a bit to tell of any player placed/changed the block, but I don't even think any of those 64 bits in the 8 bytes are free. Adding a another byte means it gets slower for code to pass around because it is now bigger than the magic size of 64 bits that our processors like.

Is it possible to have some zombies not care at all about getting to player height and are in destroy area mode by default? It just seems like when they have that path up to you (usually by design) the conga line starts and everything else is ignored making it super easy to game the AI and to get them to go exactly where you want. Predicable zombies are just not very realistic😬

  • Like 1
Link to post
Share on other sites
52 minutes ago, Orville Tootenbacher said:

Is it possible to have some zombies not care at all about getting to player height and are in destroy area mode by default? It just seems like when they have that path up to you (usually by design) the conga line starts and everything else is ignored making it super easy to game the AI and to get them to go exactly where you want. Predicable zombies are just not very realistic😬

That is not as simple as it sounds. Is that a hill or a staircase or a bunch of blocks placed like a stair case or a combination of all of it. Should the AI beat the side of the hill because you are 5m up it? No. It would not make sense. So you have to looks at all those blocks, which is slow and try to figure out what the 3d cube of blocks around and between the zombie and you mean. Tons of edge cases for nonsensical behavior.

10 hours ago, AtomicUs5000 said:

Could someone summarize what the "Wandering Sleepers" are (as listed in the A20/21 roadmap)?
... or direct me to where it was talked about, please.
 
I must have missed discussion about it, and I tried searching but couldn't find mention of it.
 

They are sleepers that move about in a small area. The will cause more overhead, but hopefully that can be balanced by optimizations that allow them to say, follow a rarely generated path with world collisions disabled.

4 hours ago, Adam the Waster said:

iv been digging from in art station and found the person who dose the 7dyd zombies

and i found a few cool things some are not done but. oh well!

I also found a shark but.... i don't think thats real LMAO

We have several artists that do zombies.

 

There is a shark model and animations in the project, but not being used.

  • Like 2
  • Thanks 1
Link to post
Share on other sites
30 minutes ago, faatal said:

They are sleepers that move about in a small area. The will cause more overhead, but hopefully that can be balanced by optimizations that allow them to say, follow a rarely generated path with world collisions disabled.


Thanks.
Will they still be tied to the POI sleeper volumes?
What gap is this idea filling? Is it for when you might look inside a POI window just to visually stir things up, or is it more about adding some difficulty for players trying to be stealthy, or something else?

Edited by AtomicUs5000 (see edit history)
Link to post
Share on other sites
37 minutes ago, faatal said:

 

 

There is a shark model and animations in the project, but not being used.

Yeah i had to find a few people who did it and wanted to share but..... 



Why is their a Shark model! the base game is set in Arizona. AKA a place with no ocean! a Gator would make sense but why is the shark not being uses?

Spoiler

jaws theme!
asia-hawkins-lowpoly-02.jpg?1488581460
 

 

Edited by Adam the Waster (see edit history)
  • Haha 2
  • Afraid 1
Link to post
Share on other sites
35 minutes ago, faatal said:

That is not as simple as it sounds. Is that a hill or a staircase or a bunch of blocks placed like a stair case or a combination of all of it. Should the AI beat the side of the hill because you are 5m up it? No. It would not make sense. So you have to looks at all those blocks, which is slow and try to figure out what the 3d cube of blocks around and between the zombie and you mean. Tons of edge cases for nonsensical behavior.

They are sleepers that move about in a small area. The will cause more overhead, but hopefully that can be balanced by optimizations that allow them to say, follow a rarely generated path with world collisions disabled.

We have several artists that do zombies.

 

There is a shark model and animations in the project, but not being used.

Interesting [spanish accent]. Btw, what are you working on at the moment, I mean aside from optimizations?

 

  • Like 1
Link to post
Share on other sites
1 hour ago, AtomicUs5000 said:

Will they still be tied to the POI sleeper volumes?
What gap is this idea filling? Is it for when you might look inside a POI window just to visually stir things up, or is it more about adding some difficulty for players trying to be stealthy, or something else?

Yes since spawns in POIs are controlled by sleeper volumes.

 

So enemies can wander around. Like bandits would move about.

1 hour ago, Adam the Waster said:

Why is their a Shark model! the base game is set in Arizona. AKA a place with no ocean! a Gator would make sense but why is the shark not being uses?

Because someone added it to the project many years ago. I can find all kinds of models and anims on the internet and slap them in, but artists generally like to do high quality specific case assets for their games.

1 hour ago, Blake_ said:

Interesting [spanish accent]. Btw, what are you working on at the moment, I mean aside from optimizations?

 

Besides some bugs, trying to get client side block stability to be calculated to work with changes to block placement visuals.

  • Like 2
  • Thanks 1
Link to post
Share on other sites

Don't see the need for wandering sleepers, tbh. They will only impact on performance while providing very little in return. IMO, anyway.

 

If you don't have collision on, does that mean you can go into a house, build a set of new walls and have the zombies walking through them? Disabling the collision is sensible in terms of reducing the amount of processing they need to do but, on the flip side, to me it seems to be introducing more possible problems in regard to not much game play return (again, IMO).

 

Obvs, I have absolutely no insight into how the logic is working so perhaps it's an easier and less impactful addition than I am imagining it might be.

Link to post
Share on other sites
14 hours ago, ricp said:

Don't see the need for wandering sleepers, tbh. They will only impact on performance while providing very little in return. IMO, anyway.

 

If you don't have collision on, does that mean you can go into a house, build a set of new walls and have the zombies walking through them? Disabling the collision is sensible in terms of reducing the amount of processing they need to do but, on the flip side, to me it seems to be introducing more possible problems in regard to not much game play return (again, IMO).

 

Obvs, I have absolutely no insight into how the logic is working so perhaps it's an easier and less impactful addition than I am imagining it might be.

I did mention bandits.

 

There are actually multiple types of collision. The collision that would be disabled is the character controller collision checks. If something entered/changed the volume then collision mode would change. We do know how to make this stuff work and if it didn't, I'd have 5 testers telling me in 1.2 seconds. ;)

 

Ironically, they would possibly be cheaper that the current sleepers, which do have their character controllers on while sitting there.

5 hours ago, Jost Amman said:

Sorry to change subject so abruptly, but I've got a question for @madmole or @faatal: will the "Gasser" be in A20.

Since the texture, model and rigging are already there, I was wondering if you can add the farting special ability in time for A20 release... :painkiller:

Not on my radar. I probably have 20 or 40 things to do before that, so likely time will run out.

  • Sad 1
Link to post
Share on other sites
21 minutes ago, faatal said:

I did mention bandits.

 

There are actually multiple types of collision. The collision that would be disabled is the character controller collision checks. If something entered/changed the volume then collision mode would change. We do know how to make this stuff work and if it didn't, I'd have 5 testers telling me in 1.2 seconds. ;)

 

Ironically, they would possibly be cheaper that the current sleepers, which do have their character controllers on while sitting there.

Not on my radar. I probably have 20 or 40 things to do before that, so likely time will run out.

speaking of Gasser (aka me when i eat taco bell)


Will there be any "New" types of special zombies in the near future? (not A20)

or will it be more about bandits in the later Alphas

Link to post
Share on other sites

Faatal

 

"That is not as simple as it sounds. Is that a hill or a staircase or a bunch of blocks placed like a stair case or a combination of all of it. Should the AI beat the side of the hill because you are 5m up it? No. It would not make sense. So you have to looks at all those blocks, which is slow and try to figure out what the 3d cube of blocks around and between the zombie and you mean. Tons of edge cases for nonsensical behavior. "

 

 

This is a question brought about by your answer to the above question.

 

Basically it's a multiple behavior scenario.

 

The simplified version, I copied all of the zombie profiles multiple times
changed the name by a numeric at the end of the name. Each one I varied the parameters
of awareness "sight hearing", approach speed, and the volume of their voice like suppressed
magnum vs un-suppressed magnum, then added them to the spawns. They all use same models but
I never knew the reaction they would have. This included sleepers.

 

All blocks in the game can be categorized into groups or hierarchy. simplest "station created,
player placed, terrain, even prefab pois have a combination of those." Player placed can be
absorbed by station created.  

 

Instead of adding a weight or calculation. Can they be given a binary designation of I and O.
If that zombie is spawned, and follows the basic pathing to a specific x,y perimeter and z is
not equal. then any "I" block is destroyed, as they move more "I" blocks are within sphere to
be destroyed.

 

Call it the "Foreman Zombie" heat is generated at an expedited rate by this type of zombie's
activities, bringing next level zombies such as wights, cops, demolishers. These zombies favor
non line of sight, so it would draw a player from safety to get rid of them before the rest come.
The crew drawn by this zombie would be cloned models, but follow the same logic as the Foreman.
Destroy area.

 

Would this add a lot of overhead.

 

  • Confused 1
Link to post
Share on other sites
On 5/15/2021 at 12:30 PM, Adam the Waster said:

Yeah i had to find a few people who did it and wanted to share but..... 



Why is their a Shark model! the base game is set in Arizona. AKA a place with no ocean! a Gator would make sense but why is the shark not being uses?

  Reveal hidden contents

jaws theme!
asia-hawkins-lowpoly-02.jpg?1488581460
 

 

Fun fact: Ravenhearst added a shark back in... alpha 17, I think. Vicious beast. Unfortunately, it was perfectly fine with leaving the water to chase players, which got it the nick "land shark". Sadly, it didn't make the cut for the public release... But we sharknado aficionados remain hopeful!

18 hours ago, faatal said:

There are actually multiple types of collision. The collision that would be disabled is the character controller collision checks. If something entered/changed the volume then collision mode would change. We do know how to make this stuff work and if it didn't, I'd have 5 testers telling me in 1.2 seconds. ;)

 

These pesky testers, wrecking up perfectly good concepts with irrelevant considerations such as common sense and gameplay mechanics!

Link to post
Share on other sites

On the feral sense thing, I'm sad that it is just about how easily zombies detect you. When I first heard of it on dev stream I thought it was about how zombies approach the player. This sophisticated path finding being used has reduced the diversity of effective bases a lot, and it is a bit absurd -- not even google maps is that good!

 

The now-abandoned idea of player's making a "trail" would have been my favorite pathing solution, and my second favorite would be path to player through line of sight. I don't like how zombies will see you and then move away from you because they know how to path to you to avoid obstacles.

 

  • Like 2
Link to post
Share on other sites
1 hour ago, dcsobral said:

Fun fact: Ravenhearst added a shark back in... alpha 17, I think. Vicious beast. Unfortunately, it was perfectly fine with leaving the water to chase players, which got it the nick "land shark". Sadly, it didn't make the cut for the public release... But we sharknado aficionados remain hopeful!

These pesky testers, wrecking up perfectly good concepts with irrelevant considerations such as common sense and gameplay mechanics!

i don't really play with mods so i can't say 

but that would be funny AF

Link to post
Share on other sites
2 hours ago, 4sheetzngeegles said:

Faatal

 

"That is not as simple as it sounds. Is that a hill or a staircase or a bunch of blocks placed like a stair case or a combination of all of it. Should the AI beat the side of the hill because you are 5m up it? No. It would not make sense. So you have to looks at all those blocks, which is slow and try to figure out what the 3d cube of blocks around and between the zombie and you mean. Tons of edge cases for nonsensical behavior. "

 

 

This is a question brought about by your answer to the above question.

 

Basically it's a multiple behavior scenario.

 

The simplified version, I copied all of the zombie profiles multiple times
changed the name by a numeric at the end of the name. Each one I varied the parameters
of awareness "sight hearing", approach speed, and the volume of their voice like suppressed
magnum vs un-suppressed magnum, then added them to the spawns. They all use same models but
I never knew the reaction they would have. This included sleepers.

 

All blocks in the game can be categorized into groups or hierarchy. simplest "station created,
player placed, terrain, even prefab pois have a combination of those." Player placed can be
absorbed by station created.  

 

Instead of adding a weight or calculation. Can they be given a binary designation of I and O.
If that zombie is spawned, and follows the basic pathing to a specific x,y perimeter and z is
not equal. then any "I" block is destroyed, as they move more "I" blocks are within sphere to
be destroyed.

 

Call it the "Foreman Zombie" heat is generated at an expedited rate by this type of zombie's
activities, bringing next level zombies such as wights, cops, demolishers. These zombies favor
non line of sight, so it would draw a player from safety to get rid of them before the rest come.
The crew drawn by this zombie would be cloned models, but follow the same logic as the Foreman.
Destroy area.

 

Would this add a lot of overhead.

 

Well a random behaviour is good  : some zombie prefere jump over things, some of them will try to go around and some "feral" will destroy everything to get to you

Link to post
Share on other sites

Not necessarily, while randomness makes fighting out in the world more interesting it makes base building and trap setting a lot harder. This is where the tower defense elements and survival fps elements can clash if not dealt with carefully.

Link to post
Share on other sites
28 minutes ago, ShellHead said:

Not necessarily, while randomness makes fighting out in the world more interesting it makes base building and trap setting a lot harder. This is where the tower defense elements and survival fps elements can clash if not dealt with carefully.

Yep, just make in logical. The biggest problem i see is water. Zombies can into water? it can look stupid. They can't? player will use this too offten ehh

Link to post
Share on other sites
2 hours ago, Matt115 said:

Yep, just make in logical. The biggest problem i see is water. Zombies can into water? it can look stupid. They can't? player will use this too offten ehh

Yeah, you can’t build around Every insane “meta” strategy the players come up with, the best you can do is make the optimal strategy a marginal improvement over just playing however.

i say this as a World of Warcraft veteran.

Link to post
Share on other sites
27 minutes ago, ShellHead said:

Yeah, you can’t build around Every insane “meta” strategy the players come up with, the best you can do is make the optimal strategy a marginal improvement over just playing however.

i say this as a World of Warcraft veteran.

well this same problem with cod XD 5 sec to kill orda gg. So  7dtd need some "impromvents" with water or add drowned zombies

Link to post
Share on other sites
6 hours ago, dcsobral said:

On the feral sense thing, I'm sad that it is just about how easily zombies detect you. When I first heard of it on dev stream I thought it was about how zombies approach the player. This sophisticated path finding being used has reduced the diversity of effective bases a lot,

 

Or in other words, base building was trivial then. So you could build anything and it would work if it just had enough HP to occupy the zombies long enough or left the zombies running around without attack vector (stilt bases and underground bases were practically out of reach of zombies). 

 

6 hours ago, dcsobral said:

 

 

and it is a bit absurd -- not even google maps is that good!

 

The now-abandoned idea of player's making a "trail" would have been my favorite pathing solution,

 

Which doesn't work because players would simply "remove" the trail by using a drawbridge or wood frames that they remove after entering the base. The trail was never described as a blood moon feature to my knowledge, it was always about some zombies following you when you were moving about.

 

6 hours ago, dcsobral said:

 

 

and my second favorite would be path to player through line of sight. I don't like how zombies will see you and then move away from you because they know how to path to you to avoid obstacles.

 

 

That is the old pre-A17 method. As I said above this AI can be trivially gamed.

 

Link to post
Share on other sites
1 minute ago, meganoth said:

That is the old pre-A17 method. As I said above this AI can be trivially gamed.

 

If pathing was decent pre-A17, I'd agree. Pathing was bad, and zombies went more or less in a direct line (unless running in circles). Pathing to a player that can be seen or to a location a sound is coming from but only through blocks that are within LoS of the zombie is not something that was ever done.

 

Regardless, this is a strawman argument. The AI was no more trivially gamed then than it is now, minus a few exploits that have been removed since. Furthermore, it doesn't matter how many exploit bases exist: if a single one exist within common knowledge, then any player who desires an exploit base will make one. The argument above boils down to "many exploit bases existed before, and now there's less exploit bases", to which I reply: who cares?

 

What I do care about is that there were many more ways of making non-exploit bases before than now, but I am not advocating going back to pre-A17 AI. Go back and read again without the "he wants A16 AI" filter on your monitor. The current AI leads to a single optimal approach, and one which becomes quite boring since 7d2d isn't going the way of Night of the Dead with multiple traps with which to design a killing maze.

 

Side note about the trail thing: yeah, players could remove the path if it was ever implemented for horde night, just like players could remove the path in pre-A17 AI and like players can remove the path on current AI. Absence of a path is a problem common to all pathing approaches.

 

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...