Jump to content

Alpha 20 Dev Diary


madmole

Recommended Posts

On 11/14/2020 at 3:22 PM, Jacoviz said:

@faatal Are you going to improve the controller rumble feedback? It would be cool to have the rumble on every melee hit and also when you fire with the guns. Also on vehicles would be cool to have a really low rumble when you are driving. Is there a way to mod this in?

Sorry. Don't know. I have never looked at our rumble code.

On 11/13/2020 at 10:05 PM, Picklefart said:

Would it be possible to add an option to adjust roaming horde size/frequency, similar to horde night variations?

We could, but probably better to wait for world events to get developed and have options for that.

On 11/13/2020 at 9:48 PM, Blake_ said:

@faatal here's a quick take on a19.3 after a couple of hours of testing :

Cool. Thanks for the feedback!

On 11/13/2020 at 7:53 PM, khzmusik said:

And is now here...

 

Just so we know - are there going to be more A19 patches, or is this it until A20? Asking for a friend who is me.

 

EDIT: Not asking for a timeline or release dates or anything, just want to know what to expect. Personally, I just want to know if I can start a new game and reach level 100 before A20 exp drops.

It could go either way. Probably depends if we find any major issues.

Link to comment
Share on other sites

on the subject of new character models and clothing, will character creation be changing?
as an aside from that, currently any hair color given to the character turns to black when you don a hat. such a small, annoying thing haha.

also thrilled the feral sense slider is coming back, that has been greatly missed by many 👍

Link to comment
Share on other sites

I have 2,000 hours in this game and literally my only complaint is the zombie combat system badly needs an overhaul.  I have introduced many friends to the game, but many don't stay because combat is so "jittery" and "unbearable.". The poor combat interactions make it hard for me to justify coming back to test out new alphas as well.  It is painful to struggle through.

 

Let me just say, I love everything else you guys have been doing.  It's truly remarkable how far you've come along. But please heed some friendly advice, combat needs massive work as it is a critical and ubiquitous interaction people have with the game.

 

Edited by Roland
There was no reason to quote the OP. Huge waste of space! (see edit history)
Link to comment
Share on other sites

4 hours ago, Blake_ said:

@faatal here's a question with a prologue: 

 

In a19.3 doors do not structurally support other doors. That is ok to avoid exploits I guess. I noticed that said band aid resulted in less creativity overall, with less cool multi stacked doors on some designs, so now I'm "unsure" if the change is intended to stay or if it's just temporary while the new doors are implemented. 

 

-Will the NEW doors be (physic-induced) blocks and also stackable on top of each other so entities can damage them and/or is the structural integrity between doors going to remain the same as this recent a19.3 change with/without the new door changes ?

 

Expected Result: That doors are cool and destroyable like in the Arlene test video, but also the entity logic is tweaked to think of doors as if they were just normal material blocks and therefore the structural integrity wouldn't need to be set to 0 for door on door, allowing the pre a19.3 freedom.

 

It's absolutely no great deal either way, but I think it's worth a thought.

 

Gentle Hugs.

I don't know yet. There is a good chance doors will retain zero SI.

Link to comment
Share on other sites

@faatal I've been further testing a19.3 and I have a few questions.

 

1-Does the loading-unloading of assets in the memory uses just VRAM and/or some CPU  ?

 

2-Does the process involve more dedicated memory usage than expected surpassing the total budget and using general purpose RAM ? How much VRAM does the game use on a minimum specs setting ? 

 

3-Is the RSS RAM feed in the dev console  the actual total amount of memory used by the game due to garbage collection and other factors?

 

4-If so, can it be cleared/emptied/reset in game ?

 

5-Wouldn't it be cool to have a CPU and VRAM and RAM tool (like  a very limited version of FRAPS) available through console commands ? 

 

 

I believe the speed of the Graphics card processing is at play here, though it shouldn't really amount to less than the necessary memory needed to the loading and unloading of resources.

 

The thing is, I have microstutters when the yellow warnings pop up (symptom). They are of various levels of severity depending on just how much stuff is being processed and I wonder about just how much do they tax the system in that short amount of time. They look like RAM microstutters, so I was just surprised given that my map is 720p and according to the console consumes just about RSS 3800 Mb of memory (with the other two feeds below 1200MB)  , so I guess I'm either mistaken and they do use some CPU or they are indeed too taxing for a minimum GPU to handle and systematically surpass the budget every time. ((EDIT: Interesting, RSS feed DOES go from 3800MB to 4300MB in just 1 hour of gameplay in 720p half size textures, a bit too much if you ask me; and it never empties to avoid that surpassing within that timeframe))

 

Maybe a more spreaded (lighter) loading and unloading would solve the issue, yet Is that even possible at this point ? 

 

Wandering hordes still give bumbs (lesser bumbs), but I believe that it's the spawner itself that might be causing problems. Apparently it's just too heavy to activate a spawn with a bit of complex logic.

 

Unity 2020.1 and 2020.2 might hold some answers for stability in these cases, though I don't believe in night and day situations this far into the dev cycle.

 

A penny for your thoughts. Thanks in advance. 

 

 

Edited by Blake_ (see edit history)
Link to comment
Share on other sites

10 hours ago, Blake_ said:

@faatal I've been further testing a19.3 and I have a few questions.

 

1-Does the loading-unloading of assets in the memory uses just VRAM and/or some CPU  ?

 

2-Does the process involve more dedicated memory usage than expected surpassing the total budget and using general purpose RAM ? How much VRAM does the game use on a minimum specs setting ? 

 

3-Is the RSS RAM feed in the dev console  the actual total amount of memory used by the game due to garbage collection and other factors?

 

4-If so, can it be cleared/emptied/reset in game ?

 

5-Wouldn't it be cool to have a CPU and VRAM and RAM tool (like  a very limited version of FRAPS) available through console commands ? 

 

 

I believe the speed of the Graphics card processing is at play here, though it shouldn't really amount to less than the necessary memory needed to the loading and unloading of resources.

 

The thing is, I have microstutters when the yellow warnings pop up (symptom). They are of various levels of severity depending on just how much stuff is being processed and I wonder about just how much do they tax the system in that short amount of time. They look like RAM microstutters, so I was just surprised given that my map is 720p and according to the console consumes just about RSS 3800 Mb of memory (with the other two feeds below 1200MB)  , so I guess I'm either mistaken and they do use some CPU or they are indeed too taxing for a minimum GPU to handle and systematically surpass the budget every time. ((EDIT: Interesting, RSS feed DOES go from 3800MB to 4300MB in just 1 hour of gameplay in 720p half size textures, a bit too much if you ask me; and it never empties to avoid that surpassing within that timeframe))

 

Maybe a more spreaded (lighter) loading and unloading would solve the issue, yet Is that even possible at this point ? 

 

Wandering hordes still give bumbs (lesser bumbs), but I believe that it's the spawner itself that might be causing problems. Apparently it's just too heavy to activate a spawn with a bit of complex logic.

 

Unity 2020.1 and 2020.2 might hold some answers for stability in these cases, though I don't believe in night and day situations this far into the dev cycle.

 

A penny for your thoughts. Thanks in advance. 

 

 

This is not an exact science, since Unity does a lot of things in the background that we don't have control over.

 

1 Depends on the asset, but everything uses some CPU and many assets use system RAM or VRAM or both. Unloading also tends to be delayed because many resources are not specifically tracked by the game and flushed by Unity by UnloadUnusedAssets, which is slow, so not called very often.

 

2 Only the gfx driver and Windows really know how much VRAM is in use and VRAM use is often padded to a higher amount that what it actually needs. Microsoft is working on tech to report more info back, but then Unity has to support it to, so maybe in a few years.

 

3 Not sure. There are various pools of memory in use, which grow and shrink.

 

4 Using the mem console command does call UnloadUnusedAssets, but it won't clear memory being used by resources, since many resources are loaded once and they would never come back. Also the Mono mem heap can never shrink, short of restarting the game.

 

5 We have the Unity Profiler. For you, FPS is a good indicator of CPU use. Use console command "gfx g fps 200" to graph it.

 

Our micro stutter is generally from CPU use spiking on a single frame. I've been working to find spikes and spread work across frames. Some of those changes have already been done for A20, but more can still be done.

 

We are moving to Unity LTS releases, so A20 should be on 2019.4. We won't try 2020 LTS until after A20.x in second half of next year.

 

Edited by faatal (see edit history)
Link to comment
Share on other sites

Hello Guys,

 

I would like to know if by chance will you be doing anything with the Terrain Blocks?

 

Examples are:  When placing a Snow Terrain block on sand it TURNS into sand but even though it looks like sand you get snowballs from it if you dig it.  Can we make it like it used to be where if you put snow on forest ground it remains snow and NOT turn into forest ground just the looks of it and not the properties?

 

Making a museum POI for my servers I made scenery that had different biomes in them.  Using the Prefab Editor it was perfect snow in it's own place,  Sand in it's and so on.  Well when the POI spawns in with the map creator it all looks good but those terrain blocks change depending where the POI is spawned in at.

 

This is a real bummer.  The Museum POI lookes great but my scenery areas all change terrain types.  Anyway of preventing this or at least is there a way to keep the look of these blocks the way they are suppose to be even when placed on different terrain?

 

Think the game is awesome, been playing since A2 still enjoying every moment of it.  Keep up the good work!

 

SeabeeMan....

Link to comment
Share on other sites

17 hours ago, faatal said:

This is not an exact science, since Unity does a lot of things in the background that we don't have control over.

 

1 Depends on the asset, but everything uses some CPU and many assets use system RAM or VRAM or both. Unloading also tends to be delayed because many resources are not specifically tracked by the game and flushed by Unity by UnloadUnusedAssets, which is slow, so not called very often.

 

2 Only the gfx driver and Windows really know how much VRAM is in use and VRAM use is often padded to a higher amount that what it actually needs. Microsoft is working on tech to report more info back, but then Unity has to support it to, so maybe in a few years.

 

3 Not sure. There are various pools of memory in use, which grow and shrink.

 

4 Using the mem console command does call UnloadUnusedAssets, but it won't clear memory being used by resources, since many resources are loaded once and they would never come back. Also the Mono mem heap can never shrink, short of restarting the game.

 

5 We have the Unity Profiler. For you, FPS is a good indicator of CPU use. Use console command "gfx g fps 200" to graph it.

 

Our micro stutter is generally from CPU use spiking on a single frame. I've been working to find spikes and spread work across frames. Some of those changes have already been done for A20, but more can still be done.

 

We are moving to Unity LTS releases, so A20 should be on 2019.4. We won't try 2020 LTS until after A20.x in second half of next year.

 

Thank you for the quick response. I kind of expected some of the memory usage answers, which are not really great news as they depend on Unity improvements.

 

I don't expect huge performance out of my Single Channel 8gb DDR2 1666Mhz RAM (833Mhz double rate)  for those intensive juggling tasks, yet I had to ask in order to separate CPU from GPU situations.

 

I will be using gfx g 200 for my next test to help things further.  FPS do drop, but not always . That "not always" alone could hint at a bigger problem. 

 

EDIT: Further testing revealed that 99.9%  of the performance problems (for a Windows with Nvidia GPU system ) are CPU related and happen in these cases:

 

-Sleeper/entity spawn, respawn and relocation after reentering the area. More than 1 sleeper can be expected to drop fps in a 4C/8T 2.7ghz system.

 

-Music uses CPU. It has a generally low CPU overhead on its own, but can worsen the experience if added to any other CPU situation, doesn't matter how low. It has a hard to spot impact on my system. I spotted it nevertheless, but never alone. Too much stop, start, pause, transition in a short timeframe generates spikes that make the system susceptible if there are other CPU bottlenecks close by.

 

-Particles, Collapses, many entities on screen, big POIs with too many Zds/rooms and also in any situation with many vertex calculations. Texture Atlas reading also uses CPU yet it is almost at its prime currently, so it probably can't get better than this in many cases.

 

-Loading and unloading of assets uses CPU , external profiling reveals that the overhead it's not constant and comes in spikes/bursts, too big for confort. Better systems should not notice these spikes, yet I do, every single time (microstutters). I don't have a clue on how to fix this. I mean maybe unplugging the power supply , blowing and plugging it again or also punching the screen while swearing in Navajo. I got nothing.

 

-Wandering hordes still generate stutters. At least they do not freeze the system for 5 seconds anymore. Just the 1 second it takes for fps to instantly drop 30-45 frames out of nowhere, it is noticeable but playable. Maybe a bit more spread would help..... 1 and a half seconds between spawns ? lol.

 

In order to clean test all of this I did a full wipe/formatting of my system on top of my OCD for absolutely no programs in the background (except the profiling). No EAC nor Gamesparks either.

 

I hope this helps to confirm currently known issues.

 

Salutations.

 

Edited by Blake_ (see edit history)
Link to comment
Share on other sites

Let me ask this very serious question, I'm very fond of the momentum that this game is moving at; believe me it's great. 

 

That being said, why don't we have a usable hearse? I as well as a few other players that I know think this would be a perfect addition to the game. Also, preacher zombies to go with the church POI's would be pretty sweet. 

Link to comment
Share on other sites

Wow! Alpha-20 looks good so far. I'm looking forward to seeing all the coming details. 

 😀

First day back to the forum. 

I was reading where crawlers can now crawl thought one meter holes. I hope that means that players once again have the ability to go through one meter openings. 

Edited by sillls (see edit history)
Link to comment
Share on other sites

A lifetime ago... I built educational software simulations in ML(animation subs) and C++.  I mostly built biomes filled with wildlife with primal AI (thirsty, hungry, sleepy, defending).  Managing the threading priorities (what is noticeable) in the event manager is key, along with giving the entities faster options (e.g. less accurate approximations, or delaying an action) when the number of active entities gets too high, or FPS for example drops. 

 

So, in this game, I would imagine a scenario, like 100 zombies getting hit by a concussion grenade and falling down.  If they all try to follow the exact same AI script to get up all at the same time, not only would it break realism, it would over tax event handling.  My desire would be to have them recover at variable rates and "decide" to act when it is in the best interest of making the simulation realistic, within the constraint of the capability of event management ( some number of entities would choose to delay acting  for the benefit of the others).  But, they also need to be smart enough to not be random, in that they delayed an action rather than skip it.

 

This game compounds what I dealt with in my simulation problems in a number of ways.  I never had to deal with the structural integrity and physics of every object in the world.  I didn't have to load/unload regions of the map (in effect freezing time while unloaded and possibly reacting to changes when unfrozen), and I never had the issues of multiple players driving at high speed in vehicles through the world changing the decisions of hundreds of AI entities and physical objects. 

 

So I would just say in sum, good job to the developers who are working on this software and I'm looking forward to seeing how you resolve the issues.   Probably most people would not understand the challenges that need to be resolved or the vast number of compromises that need to be made for performance.  

Link to comment
Share on other sites

2 hours ago, Gazz said:

The vehicles in 7D do count as underwater transport.

Fixed for you :)

 

Btw, @faatal, or you Gazz, are you going to improve the point of view transition, of when we change vehicle steering mode between mouse to keyboard and vice versa, with the left mouse button.

Sometimes it changes a bit of the angle and is comfortable to continue, but sometimes it is like I am looking from a selfie stick point of view instantly, and a world class of slalom happens. Not a big deal, just wonder about it.

 

Great job so far boys, keep kicking butts. 

Edited by beerfly (see edit history)
Link to comment
Share on other sites

I thought zombies were supposed to dig in A19, but I built a "trap base" near bedrock with a long ramp going down to me, and they didn't show up on BM.

I guess there's a limit on how far pathing can reach out? :pout:

 

@faatal : could you check this for A20 and see if it's possible to have them path to you from 60-100 blocks away? (at least during the Blood Moon!) :yo:

Link to comment
Share on other sites

On 11/2/2020 at 6:49 PM, madmole said:

The drone has a headlight mod.

Is this exclusive to the drone or can the sledge and turret also have this attached? I've been hoping for the ability to hook a light up to the turret basically since I originally got it. They don't support the weapon flashlight mod.

Link to comment
Share on other sites

On 11/19/2020 at 11:29 PM, sillls said:

Wow! Alpha-20 looks good so far. I'm looking forward to seeing all the coming details. 

 😀

First day back to the forum. 

I was reading where crawlers can now crawl thought one meter holes. I hope that means that players once again have the ability to go through one meter openings. 

None of that has been done yet, so it remains to be seen how it will work.

Link to comment
Share on other sites

On 11/20/2020 at 2:09 PM, Jost Amman said:

I thought zombies were supposed to dig in A19, but I built a "trap base" near bedrock with a long ramp going down to me, and they didn't show up on BM.

I guess there's a limit on how far pathing can reach out? :pout:

 

@faatal : could you check this for A20 and see if it's possible to have them path to you from 60-100 blocks away? (at least during the Blood Moon!) :yo:

The pathing distance limit is horizontal. They can only path about 40m out. Farther and they just go in a straight line to the edge of the path grid.

I dug to bedrock in Navezgane and BM day 7 spawns came above me and started digging down. They do a lot of moving and stumbling in addition to digging, so it can take them a while if you don't have a ton of zombies.

On 11/20/2020 at 10:50 AM, beerfly said:

Fixed for you :)

 

Btw, @faatal, or you Gazz, are you going to improve the point of view transition, of when we change vehicle steering mode between mouse to keyboard and vice versa, with the left mouse button.

Sometimes it changes a bit of the angle and is comfortable to continue, but sometimes it is like I am looking from a selfie stick point of view instantly, and a world class of slalom happens. Not a big deal, just wonder about it.

 

Great job so far boys, keep kicking butts. 

That is code, so would be me, but doubt it will change anytime soon. I tend to forget that is even a feature.

Link to comment
Share on other sites

7 hours ago, faatal said:
On 11/20/2020 at 9:09 PM, Jost Amman said:

I thought zombies were supposed to dig in A19, but I built a "trap base" near bedrock with a long ramp going down to me, and they didn't show up on BM.

I guess there's a limit on how far pathing can reach out? :pout:

 

@faatal : could you check this for A20 and see if it's possible to have them path to you from 60-100 blocks away? (at least during the Blood Moon!) :yo:

The pathing distance limit is horizontal. They can only path about 40m out. Farther and they just go in a straight line to the edge of the path grid.

I dug to bedrock in Navezgane and BM day 7 spawns came above me and started digging down. They do a lot of moving and stumbling in addition to digging, so it can take them a while if you don't have a ton of zombies.

Thanks for the reply @faatal but, honestly, this explains only part of why the zombies didn't find me during horde night.

 

This is what I did: before my first BM I created a slanted tunnel going down at 45° for (admittedly) longer than I had expected, surely more than 40 blocks horizontally. It's a wide entrance since it's 3x2 all the way down. I dug it like that so that the zombies would easily fit in, but as you confirmed, they didn't come. Next day I found some of them tunneling down in a nearby POI but apparently quite far from my horizontal position that night (no idea why).

 

So after that utter failure on my part I decided to use the old "pit trap" design. So I dug a completely vertical 3x3 tunnel down to the end of the 45° ramp. So, my latest design has 2 points of entry: the first through the ramp and the second (vertically) from the "skylight" with a ladder all the way up.

 

Well, next BM night, even after this last change, I stood right at the center of the vertical pit with no obstacle above me, but the zombies still didn't come. Not even one fell down or tried to use the ladder to reach me (there's a gap of a few blocks at the end). Next day I went up and checked the ground near my pit and found no evidence of any digging! Is this what you'd expect? Just trying to help. I'm sniffing some erratic behavior or hard "limit" on the AI part... Thank you!

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
×
×
  • Create New...