archived-code-advanced

[earnest canopy]

I debugged everything

The content of it is right

But i behaves differently in editor

[untold moth]

Content of what?

What exactly behaves differently and how?

We can't really help you without any details.
The compute shader itself behaves the same both in the editor and play mode.

[earnest canopy]

I have a computer shader that takes a density array and gives me a triangle array back

Marching cubes

[untold moth]

Go on.

[earnest canopy]

So if i just generate my terrain everything works

But if i modify the terrain with a brush in edit mode then my triangles also kinda lost (half of it)

[untold moth]

Ok, so the issue is not generation of the mesh, but editing it with a brush? Do I get it correct?

[earnest canopy]

Yes

But it works in playmode

That makes me crazy

[untold moth]

There are many differences between play mode and edit mode. Mostly related to lifetime of objects/buffers and the execution time of events and application loops.

But it's hard to say anything specific without more details/code.

[earnest canopy]

I checked the buffers
Everything is right

[earnest canopy]

But why does the generation work called with OnValidate But otherwise it is broke

Does OnValidate behave differently?

The code

[untold moth]

!code

[thorn flint]

Posting code

📃 Large Code Blocks

Use links to services like:
https://gdl.space/, https://paste.ofcode.org/, https://hatebin.com/, https://paste.myst.rs/, https://hastebin.com/

📃 Inline Code

Surround code with three backquotes. Not quotation marks.
To format as C#, add cs to the first line:
```cs
// Your code here
```
Add a comment with a line number if there is an error message.

[untold moth]

On validate is very specific callback. It shouldn't really be used for anything other than validating changes in the inspector.

[untold moth]

Does it work with OnValidate and breaks otherwise? And what is the "otherwise" that you're referring to?

[native beacon]

I mean, maybe you can't find any flaws in it but that is far from the same thing as it being flawless. The example it gives is basically the 'don't do this' section of any beginner tutorial on the subject.

[hot aspen]

anyone whos advanced with nav mesh agent that can answer my question?

I have an agent that goes to a destination. (this destination keeps getting set on update since the destination target can move.

is there a way to add extra motion to his path?> for example. lets say i want him to travel in a SIN wave pattern but still get to its destination. is this possible? so its not just a str8 line

[untold moth]

It's possible if you get the path and move the character yourself instead of delegating it to the agent.

[dapper cave]

Trying to push particles away from each other in Sjuriken with this job but instead they wobble at the emitter, even if I remove particle.velocity = newVelocity. Anyone can help me with that?
https://pastebin.com/50mFx3xy

[trail flax]

anyone have good videos on code architeture for 2d games?

[hot aspen]

and how owuld i move it myself not using the nav mesh agent and not using nav surface?

[untold moth]

You do use the nav surface and the agent to calculate the path. And you can move it in whatever way you want: transform, rb, character controller.

https://docs.unity3d.com/ScriptReference/AI.NavMeshAgent.CalculatePath.html

[earnest canopy]

From the brush

[untold moth]

What is "from the brush"? Where is the brush code executed from? Where is the entry point?

Share the code properly, so that we can see it.

[earnest canopy]

i fixt it

not unitys fault

my fault

so basically i used the wrong array type

used float instead of vector4

but why does it work in playmode but not in editor xd

[untold moth]

You never shared enough info for us to answer that question.

[earnest canopy]

ik sry

the problem is that i tried to save float4 in float and then i run out of space

[stone quarry]

if i am creating some kind of "open world" rpg game at what point do i need to worry about things like streaming content/advanced memory management in the world and what technique works best for Unity? as a solo developer i want to make a reasonably big open world, but i am not trying to break any records or create anything ridiculous. i've had trouble finding good answers to this kind of question, but i also don't know when exactly it's necessary. i don't have any rigid requirements; i can be flexibile in my approach. any good articles or examples? think valheim, but smaller.

is this something i can accomplish using addressables?

[regal lava]

you worry when you go past 5k units for when your camera starts to shake because of floating point precisional errors

[stone quarry]

haha ok, but i do want to structure the engine a bit before getting too deep. you know, plan for having to stream the content around the player in square chunks or something

it's hard to say how big relative to existing games because we all know how nuts they are today in scale. i guess mine would be considered "small" by comparison and the target platform is PC. i just wish i had a rough guide to say if i can jam it all in ram or if i should plan from the get-go to have world streaming.

if it's just addressables being used is it as simple as breaking the world into square chunks and streaming in the chunks surrounding the player's current position?

[potent pecan]

Yes, addressables can help with this. If you try and load a whole 5kmx5km game world worth of assets at once you're going to run out of memory really quickly.

[dusty wigeon]

As a solo developer, I'm not expecting you to have high resolution asset and I am pretty sure you potentially can have all your asset in memory. (For the PC platform)

What is the size of your world for reference in km ?

And how much variety you expect to have ?

Also, what your art style would look like ?

[stone quarry]

it would be low poly for sure. think valheim-esque with regard to graphics. so nothing crazy. i don't have a specific size determined as of yet, but i would definitely call it small by any modern standards, just big enough to feel like it's a world. i don't want a giant empty world; it has to have enough content to interact with and explore (i am thinking cave entrances that will load a new scene). so an overworld with cave/building entrances that load a new scene.

sparsely populated with enemies, but things like landscape objects are probably a big thing; trees, rocks, etc.

i am seeing some people just additively load/unload scenes together in something like a 3x3 grid around the player or trigger loading somehow. and addressables for intra-scene streaming if that is even necessary with my PS1 style low poly art

[dusty wigeon]

From what you say, I'm almost sure you can fit your whole world in memory. Keep your texture not to high resolution (and maybe you will have your terrain that needs to be loaded/unloaded)

[stone quarry]

yeah you may be right. i can probably throw together a prototype and profile it. my laptop is a decent gaming rig, but it's a few years old at this point. probably a decent base for performance testing.

i dont want to overcomplicate things for sure

[odd tulip]

Hello! I'm new to Unity Netcode. I have a snowball that I instantialize, which has a script of network behaviour.

When I enter the app and create a session (so I'm the server, I can instantiate and see others instantiate their snowballs just fine), but as a client I can not see mine or server's snowballs.
What am I doing wrong here? Can anyone help?

[undone coral]

~~https://docs-multiplayer.unity3d.com/netcode/current/basics/object-spawning/~~

In Unity, you typically create a new game object using the Instantiate function. Creating a game object with Instantiate will only create that object on the local machine.

[undone coral]

can you copy and paste your code so that the text is in here instead of the screenshot?

[thin mesa]

while this is accurate, they are instantiating it correctly. for n4go you instantiate then call Spawn on the NetworkObject component which is what they are doing

[undone coral]

i apologize i should read my own docs

my conclusion is that the api is horrible lol

[undone coral]

go ahead and share a more comprehensive snippet because ther'es nothing wrong with this code

[odd tulip]

Well thats all I shared, because when the gun is shot, it triggers just this function, and nothing else.

That's why it's all I shared. I don't know what else I'm missing!

I did notice there is a [ClientRcp]? Should I also make a duplicate of this function as ClientRcp and make that function execute as well??
Because The server can see both snowballs instantiated by themselves and the clients. But the clients see nothing.

[tall ferry]

client rpc would mean the server is telling the client(s) to run code, you shouldnt need it. You should check if the object is in the network prefab list, and try to see if it does exist on the client but is disabled/not spawned

[odd tulip]

It's already night time here, but I will run some more tests tomorrow. Its definitely in the network prefab list, I've checked it so many times from all the googling I've done about this. 😭

[tall ferry]

when you test, add debugs because it can be confusing where code is running (client or server). You should see if the server even runs this code in the first place, like is this a network behaviour?

[odd tulip]

It's a bit hard to do testing with unity editor, because I'm doing this in VR, and for some reason, pc and vr headset don't like connecting to the same session 99% of the time. VR to VR headset, no problem though.

But I can make debug show up in VR headset though.

On an off topic, has anyone tried using Vivox? It's another pain.. Their documentation to me personally was completely useless... I just don't get it. I got the sample project, ran it, and it didnt even work...

Is Vivox worth a go?

By the way, what would that mean if it is spawning but it is just invisible?

[odd tulip]

Yes, as mentioned above, the script is Network Behaviour, the prefab has network object component and network transform component, and the server sees the objects instantiated by the clients.

[thin mesa]

#archived-networking
but also it's probably just not in the right location on the clients rather than being invisible

[odd tulip]

ok, thank you, I'll check these things tomorrow

[loud cypress]

does anyone have a script that allows you to export an object's geometry to a file?

[sly grove]

https://docs.unity3d.com/Packages/com.unity.formats.fbx@5.1/manual/index.html

[loud cypress]

why does the exported fbx crash everything I try to open it with?

[sly grove]

¯_(ツ)_/¯

not really a code question btw

this is more #🔀┃art-asset-workflow

[loud cypress]

yeah ig

[loud cypress]

it just became one

the object only has geometry at runtime, so I need something that can export the geometry at runtime

[ashen shard]

You can use this https://github.com/atteneder/glTFast

[loud cypress]

works great... for everything else

idk why, but it seems like the script isn't properly interpreting the values sent back from the gpu for the triangles, so it renders fine, but the cpu side geometry is all messed up

i'm working on editing this to convert an sdf to geometry https://github.com/keijiro/ComputeMarchingCubes

[cobalt hemlock]

Hey guys, i'm working on implementing an undo and redo system in my application using the command pattern. Everything works, but i've come across this weird issue that the data in the stack data type changes slightly when doing a redo but only for redo.

The group and array classes for use the same Stack<List<gameobject>> to store the data, but for the array it seems the data changes slightly in structure after being added.

Does anyone know what could cause this issue?

I have attached an image with the issue highlighted in the visual studio editor when debugging the data.

The code used to set the gameobject list is the same for both groups and array.

[tiny pewter]

what is "doing redo" means (you show the undo only)
btw are you sure than everything is deep copyed when pushed into stack? otherwise if you put some managed type but still holding the reference and change the data, the data in stack also be changed (since they are pointing to the same memory block)

[cobalt hemlock]

Thanks for responding @tiny pewter , the implementation for undoing the grouping operation is the same as the array. Its when I start to do the undo operation on the array, the list of games objects when popped from the stack, doesn't go into the ref List<GameObjects>. It works for the group, but for array it doesn't even though the setup is pretty much the same.

I am trying to see what the differences are to change the structure in the sack (The bit i highlighted in red.) because the data is still there, but it seems to be nested one extra level, so it doesn't get added to the ref Array.

[cobalt hemlock]

So just re-rereading what you're saying, there could be something changing the data as its only there as reference.

[tiny pewter]

oh, mb
it is fine to cache the object that needed to be restored (otherwise no way to know which one is needed to be undo), but data still needed to be deep copyed.

[cobalt hemlock]

Hey thank you so much, its what you said about the reference data pointing to the same part of the memory that was the issue. I was adding the objects before fully completing the array process. so the because the data for re-parented to the array container, it mutated the structure in the stack.

So i just had to make sure when I add the items to the stack, it was done when the process is fully complete, so now I can toggle back and forward with the undo and redo.

I just have to make sure when its added, no further operation needs to be acted on the object.

@tiny pewter appreciate the help. Thank you.

@tiny pewter just one thing, is possible to deep copy a reference type object?

[tiny pewter]

https://stackoverflow.com/questions/129389/how-do-you-do-a-deep-copy-of-an-object-in-net
I suggest you dont use reflection if possible, write your custom solution

[cobalt hemlock]

Thanks for the suggestion, i'll through that.

[tiny pewter]

https://leetcode.com/problems/clone-graph/
deep copy is travelling on a graph

[cobalt hemlock]

Thank you.

[real blaze]

worth noting on unmanaged data it's as easy as a simple memory copy. marshal and memoryMarshal classes have the necessary tools for that.

[rancid oxide]

I need some help on bulding remotely using AWS, looks like in the lastet version my build requires too much RAM and it fails. Is there any way to change settings relative to the build so it uses less RAM?

[earnest canopy]

can someone help me to speed up my jobs in unity

[livid kraken]

Show your code, say where is the slowdown and maybe someone will know a way to speed it up. https://dontasktoask.com/

[pliant geyser]

So i have a weird case where Im trying to serialize an unknown type based on an enum, so pretty much the flow is:

• Select an enum in the inspector window
• A set of fields show up which have different fields

The way im managing this right now is that based on the enum, i grab whatever corresponding class (fetched from some other source), and the way im serializing this unknown type is that Im saving all field names and values into an array of a serializable class (With 2 fields being the field name and the field value as a string)

When I want to deserialize, I take this class and shove it into a Dictionary<string, string>, then deserialize that into the resulting class. I can provide some small example code to show off what im doing, but is there any way I can improve on this?

[sly grove]

Are these types all subclasses of a parent class or something? You might look into SerializeReference

[dusty wigeon]

Use SerializeReference maybe ?

[pliant geyser]

awkward example of what im pretty much doing
https://hastebin.skyra.pw/afiluyamuh.csharp

and atm it isnt a subclass of a parent

also what would serializereference do for me here?

[dusty wigeon]

It enable you to have different serialization base on the type. By example, you could have an abstract Field which would the be overriden by IntField, FloatField, etc.

All of them would then be serialized correctly and you wouldnt need to have a custom serialization structure.

Alternatively, you could simply use Json or XML instead of relaying directly on Unity serialization.

[pliant geyser]

yeah thats pretty much what im doing right now, just serializing the class instance into a json then deserializing later on

[dusty wigeon]

Finally, you have the option of using ScriptableObject as well.

[dusty wigeon]

Using Json would be better than using a dictionnary

[pliant geyser]

that was my old solution but it didnt scale all too well (SO)

hmm i guess this dictionary solution is a bit awkward

might be better to just fully commit to json then

[stark shore]

None of my callbacks are firing. Im not getting any errors or exceptions. Did I not subscribe to the events properly? What gives?

[sly grove]

What's the context here? Which network framework is this?

[stark shore]

Unity built in

https://docs.unity.com/ugs/en-us/manual/lobby/manual/lobby-events

Im using Lobby Events

[sly grove]

Dumb question but where are you running this subscribe function from

[chilly talon]

Does Unity support libraries written in .NET 8.0?
Or do I have to use .NET Standard 2.1?
I'm guessing not, but worth to ask.

Currently writing an interpreter and taking advantage of source generated regex.

[stark shore]

Sorry for the delay. I’m running it from a few other methods. One creates a lobby and the other joins an existing lobby.

[sage radish]

It has to be built against .NET Standard 2.1.

[chilly talon]

Damn. Figured as much but didn't know for sure.
Especially since Unity has a lot of dark magic in the backend.

Cheers for the answer.

I really wonder when Unity finally moves to CoreCLR

[sly grove]

Are you running it on the main thread?

[plush elk]

hey guys I need some sugestion, I am making a dungeon generator, I am making the rooms and with walls good, now I am trying to figure out the best aproach to make doors, I already have navmesh being done on the rooms

any ideas on how to aproach this?

[undone coral]

hey guys I need some sugestion, I am

None of my callbacks are firing. Im not

So i have a weird case where Im trying

[woeful sleet]

given how tightlipped they are about it on the roadmap, i wouldn't expect to see anything in alpha until the end of next year at the earliest

[chilly talon]

Right, my guess is late 2025.

[woeful sleet]

shipping the alpha then would mean net10 is already out, which means unity would be 6 versions behind everyone else at that point. seems very risky to put something this important that far in the future

[chilly talon]

I can agree with you there, but Unity has always been really late.
Wouldn't surprise me if it came really late, but I truly hope it will come anytime soon next year 🙂

[half swan]

The thing is, once they make this next jump, they will be able to keep up with .net more closely, it's the mono baggage holding them back now

[woeful sleet]

sure, but that doesn't change that there'd be years spent on the current language version and performance status. Given it'd be 3-5 years then for an lts release to happen, that's time the competition which already is on core clr can use to draw users away

[half swan]

I mean, it would change that. Not sure what you mean. Once they jump from mono

Edit: Oh, you mean the time it takes to GET to the point where they leave mono behind? Yeah, I get that

[woeful sleet]

yea

[novel plinth]

At one point Josh said, they've successfully internally implemented CoreCLR for in-Editor only.
I just want Unity to jump to NAOT bandwagon, all other engines already take advantage of this

[chilly talon]

For sure! I do wonder if Unity use Reflection though - If it does it could be a problem.

Probably not since the engine is mostly C / C++

[half swan]

The unity methods (start, awake, update, onenable, etc) are collected via reflection iirc

[chilly talon]

I thought they were C / C++ voodoo

TheMoreYouKnow 😛

[livid kraken]

NAOT?

[scenic forge]

https://learn.microsoft.com/en-us/dotnet/core/deploying/native-aot/

IL2Cpp is the solution for environments that disallow JIT like iOS, NativeAOT would also solve that problem.

[novel plinth]

also as long as the RYUJit supports the CPU it can run it no problem

meaning any future consoles (upcoming gens) would be easy to support

and lo and behold, https://github.com/dotnet/runtime/blob/main/src/coreclr/nativeaot/docs/reflection-free-mode.md

the god tier free reflection-mode

been seeing FNA folks bragging their NAOT ported games lately, damn, I'm so jealous

[ebon jungle]

Hey, I have a general question, does anyone happen to know classes from System.Runtime.Intrinsic would be useful in any way in Unity? Or is that something game-devs don't really use?

[tiny pewter]

sounds like SIMD

[compact ingot]

Those break platform compatibility, intrinsic means they are effectively executing instructions that are specific to a given CPU architecture. It’s one step lower level than the vector compute (simd) abstraction from the .Numerics namespace that makes that same stuff cross platform.

[hushed fable]

Josh said that they haven't looked at NativeAOT yet. NativeAOT probably isn't quite ready yet to replace IL2CPP, especially considering the browser platform which relies on Emscripten. NativeAOT-LLVM branch is active though, so there should be a path available. I imagine Unity would like to offload IL2CPP work to the .NET group.

[ebon jungle]

Yes

[ebon jungle]

Oh, okay. So is the Numerics namespace useful for game-dev ?

[tiny pewter]

we have unity.mathematics

[woeful sleet]

it would be, but every single gfx library ports brings their own math types. That means you have to do some pretty heavy computation before you outweigh the cost of copying all the values when its time to render

so usually its just for ai-heavy stuff or scientific compute

[ebon jungle]

Oh, okay, thanks!

[earnest canopy]

!code

[thorn flint]

Posting code

📃 Large Code Blocks

Use links to services like:
https://gdl.space/, https://paste.ofcode.org/, https://hatebin.com/, https://paste.myst.rs/, https://hastebin.com/

📃 Inline Code

Surround code with three backquotes. Not quotation marks.
To format as C#, add cs to the first line:
```cs
// Your code here
```
Add a comment with a line number if there is an error message.

[earnest canopy]

hey guys
can you help me to speed up my code?

https://hastebin.com/share/vekuhoguwi.java

[loud cypress]

why does this correctly render, but the vertices and triangles lists are full of garbage?

        _builder.BuildIsosurface(_voxelBuffer, _targetValue, 1.0f/_gridScale);
        MeshFilter meshFilter = GetComponent<MeshFilter>();
        meshFilter.sharedMesh = _builder.Mesh; //renders perfectly
        int[] triangles = meshFilter.sharedMesh.triangles; //full of nonsense
        Vector3[] vertices = meshFilter.sharedMesh.vertices; //full of nonsense

what's in the vertices:

then later on:

[long ivy]

that data exists on the gpu and you never copy it back to cpu

[loud cypress]

so is setting shared mesh just some gpu reference?

and how do i get the triangles back from the gpu?

[long ivy]

setting the shared mesh isn't just a gpu reference, but your data is on the gpu and not system memory so reading its geometry that's in system memory (triangles, verts) is not going to be correct

what did you try so far? You should start by reading the docs for the classes you're using to reference that data, now that you know it's on the gpu and you need to do something extra with it

[loud cypress]

frankly, I'm not sure how this stuff works. Most of this code is from https://github.com/keijiro/ComputeMarchingCubes

looks like I have to use ComputeBuffer.GetData

but I have no clue how to make it the correct format (or how to use it)

[plush elk]

hey guys need some sugestions

I am working on a map generatrion usig a grid system, my logic is, when there is a change in floor it meant I whent trought a wall, for example I start doing the corridor at the center of the room A to B so when I stop seeing floor it meant I whent trought a wall so a door is placed

and vice versa

now my problem is, how can I flip the door in the correct side of the square grid and the direction

[left quiver]

https://hastebin.skyra.pw/mepehivabo.pgsql

would anyone be able to help me with my unity2d level editor script? this was mostly generated by chatgpt through many iterations but it doesn't save my changes from 2+ iterations previously.

i'm sleeping now so please ping me if you reply, thanks!

[plush elk]

chatgpt sucks for code

at least the free one

try to debug the code little by little and try to undertand what it is doing

[left quiver]

alr ty haha

[wet burrow]

So I'm trying to unpack a crash report and got this error:
Got a UNKNOWN while executing native code. This usually indicates a fatal error in the mono runtime or one of the native libraries used by your application.
Anyone experienced with this sort of thing and how to better troubleshoot it (e.g. unpacking the conventions of what's being shown in those reports)? I don't want to just upload the whole file because I'm under NDA and trying to be on the safe side of not accidentally revealing stuff

[upbeat path]

Windows?

[tiny pewter]

did you read the stacktrace?

[wet burrow]

yes

[wet burrow]

trying to decipher it. So it's just essentially the 'hierarchy' of functions that led to the error? E.g. A was called from B which was called from C, etc.

[upbeat path]

Take a look at the Windows Event Logs, they may contain additional debug info

[wet burrow]

I'll copy this in, I realised it isn't giving anything serious away

=================================================================
    Managed Stacktrace:
=================================================================
      at <unknown> <0xffffffff>
      at UnityEngine.Object:DestroyImmediate <0x00094>
      at UnityEngine.Object:DestroyImmediate <0x00022>
      at Obi.DynamicRenderBatch`1:Dispose <0x00182>
      at Obi.ObiClothRenderSystem:Dispose <0x00052>
      at Obi.ObiRenderer`1:UnregisterRenderer <0x0015d>
      at Obi.ObiActorRenderer`1:DisableRenderer <0x000de>
      at Obi.ObiClothRenderer:OnDisable <0x00093>
      at System.Object:runtime_invoke_void__this__ <0x00087>
      at <unknown> <0xffffffff>
      at UnityEngine.GameObject:SetActive <0x0009b>
      at GarmentDesign.Mannequin:Undress <0x000e2>
      at GarmentDesign.Mannequin:InitializeOutfit <0x00022>
      at <DismantleGarments>d__43:MoveNext <0x0006a>
      at UnityEngine.SetupCoroutine:InvokeMoveNext <0x0004f>
      at <Module>:runtime_invoke_void_object_intptr <0x0008e>
      at <unknown> <0xffffffff>
      at UnityEngine.MonoBehaviour:StartCoroutineManaged2 <0x0009b>
      at UnityEngine.MonoBehaviour:StartCoroutine <0x00072>
      at <LoadSequence>d__9:MoveNext <0x0017a>
      at UnityEngine.SetupCoroutine:InvokeMoveNext <0x0004f>
      at <Module>:runtime_invoke_void_object_intptr <0x0008e>
=================================================================

[upbeat path]

DestroyImmediate?

[wet burrow]

no idea what it's used for

I'm trying to decipher a heap of code someone else made

but it's probably good to check where DestroyImmediate could be running from

thanks for the lead @upbeat path

[upbeat path]

DestroyImmediate should not be called outside of Editor code. And, btw, why are you not working from the source code

[wet burrow]

hang on what do you mean by the second line?

[upbeat path]

This stack trace is obviously from compiled code

[wet burrow]

I was running the program in play mode in the editor and it crashed, and in the bug reporter window it listed crash log files that I opened

[upbeat path]

then the stacktrace should trace source, that is not the case here

which would suggest it is not crashing in your code but in a plugin

[wet burrow]

no idea, there's some super complicated code that I've only just begun to figure out, the guy who made it has dropped out of the project due to mental health issues and also being a dad

[upbeat path]

you are missing the point. What I am saying is it is not source code in your project generating this error

[wet burrow]

okay I think I need to double-check what a plugin specifically is to make sure my definition is correct and I know what to specifically look for

I searched for 'plugin' in the crash log and got some repeated lines:

Refreshing native plugins compatible for Editor in 85.86 ms, found 3 plugins.
Preloading 0 native plugins for Editor in 0.00 ms.

oh right

[wet burrow]

wait hang on so what bit actually tells you that?

[upbeat path]

all of it

[wet burrow]

so is the whole layout different?

[upbeat path]

yes

[wet burrow]

ah

[upbeat path]

methinks you are considerably out of your depth here

[wet burrow]

this is my first industry job and I have nowhere near enough support from people who actually know what they're doing

but this stuff isn't going to get done otherwise

so I have to learn it

[upbeat path]

Don't feel too bad, this stuff is very tricky even if you do know what you are doing

[wet burrow]

it'll help me get ahead in the industry

[wet burrow]

Aha! I found a use of DestroyImmediate in an OnDestroy() function

a thing that's runtime and not editor

and is the base class of the class that seems to be running OnDestroy() at the time the editor crashes

[upbeat path]

this is where to look

at UnityEngine.GameObject:SetActive <0x0009b>
      at GarmentDesign.Mannequin:Undress <0x000e2>
      at GarmentDesign.Mannequin:InitializeOutfit <0x00022>
      at <DismantleGarments>d__43:MoveNext <0x0006a>

you have a coroutine DismantleGarments which calls GarmentDesign.Mannequin:InitializeOutfit which calls GarmentDesign.Mannequin:Undress which tries to call SetActive on a null or destroyed gameObject

[wet burrow]

yep that's already the code I've been checking for several hours

using breakpoints and stuff

[upbeat path]

Aren't stack traces wonderful things?

[wet burrow]

but for some reason it took me this long to really look in-depth at the crash logs

[wet burrow]

yeah I use them all the time in the editor

to check where a function is being called from

[upbeat path]

if you change DestroyImmediate to Destroy it will probably throw an error as well so wrap it in a null check

[upbeat path]

that's the one

[wet burrow]

I may need to actually change it and try it tomorrow morning

because it's past midnight

[upbeat path]

you certainly need to change the DestroyImmediate to Destroy

[wet burrow]

thanks for the help though, even if that doesn't specifically solve it there are a bunch of other DestroyImmediate instances I need to snoop out and double-check

as in ones that aren't used in the editor

[upbeat path]

replace all of them

[sudden rapids]

Hi everyone, I'm trying to rotate an object in a way that its transform.up would be aligned with the normal of the ground it is standing on, and at the same time indepedently rotate it around the y axis in order to reach a certain target.
When I do these rotations separately on my object, they work perfectly, but when I add them together, one of them always decides not to work properly.
In this implementation, the y rotation kind of works, but the normal doesn't.

Any help would be greatly appreciated! 😃

Code snippet:

//Calculate the rotation based on the Y axis rotation
Quaternion yRotation = Quaternion.Euler(0, Time.deltaTime * currentAngularVelocity, 0);

//Calculate normal
Vector3 v1 = rightLegTargets[0].transform.position - leftLegTargets[^1].transform.position;
Vector3 v2 = rightLegTargets[^1].transform.position - leftLegTargets[0].transform.position;
Vector3 normal = Vector3.Cross(v1, v2).normalized;

//Calculate rotation to align transform.up with the normal vector
Quaternion normalRotation = Quaternion.LookRotation(transform.forward, normal);

//Combine the normal rotation and the Y-axis rotation
Quaternion combinedRotation = normalRotation * yRotation;
transform.rotation = combinedRotation;

[native jasper]

To rotate around the normal, you want Quaternion.AngleAxis. This will make your transform's "up" direction align with the normal vector (you can imagine the axis, normal, skewering the transform, and then twisting it by myAngle):
Quaternion normalRotation = Quaternion.AngleAxis(myAngle, normal)
The second part you describe isn't clear to me: and at the same time indepedently rotate it around the y axis in order to reach a certain target. I think what you want is to calculate that myAngle above in a way that points (roughly, since it's tilted by the normal vector) to some target? But I'm not sure what you mean by "independently rotate".

Quaternion.LookRotation isn't behaving the way you expect here, it will rotate the transform off the normal axis. The up vector param in Quaternion.LookRotation is not the same conceptually as the axis param in Quaternion.AngleAxis. I very rarely pass anything here, I just let it default to Vector3.up, unless I am inverted for some reason and want to pass Vector3.down.

[sudden rapids]

Hi, thanks for replying.
I changed my code to this now:

 Vector3 v1 = rightLegTargets[0].transform.position - leftLegTargets[^1].transform.position;
 Vector3 v2 = rightLegTargets[^1].transform.position - leftLegTargets[0].transform.position;
 Vector3 normal = Vector3.Cross(v1, v2).normalized;

 
 float angleY = Time.deltaTime * currentAngularVelocity;
 Quaternion normalRotation = Quaternion.AngleAxis(angleY, normal);
 transform.rotation *= normalRotation;

This results in my object rotating toward the specified target, but the transform.up isnt the normal, instead it just stays as is. (for clarification, the target is a sphere somewhere in the scene, and my object needs to rotate by a constant angular velocity value every frame to face it)

Am I doing anything wrong here?

[sly grove]

This is way over complicated. Just do Quaternion.LookRotation with:

• first parameter your look direction projected on the normal plane (Vector3.ProjectOnPlane)
• second parameter is the normal

[plush elk]

hey guys I am doing a map generator, I think I need to use the proffiler to try to debbug the problem but I dont undertsand much about it, the problem is that after I generate the map, unity becomes slow untill I delete the list of rooms I created, I already checked the methods that use the list but I dont find the problem, unity slows doen even after the generation finishes

[dire bramble]

could you show us some code?

[plush elk]

ufff

its some 450 lines of code

I can create some paste bins to share you interested?

[dire bramble]

create a single pastebin and link it here

[plush elk]

whats the best site? pastebin.com?

[sudden rapids]

It probably is over complicated, I'm pretty much a noob when it comes to quaternion shenanigans.
I understand your suggestion, it's just that I'm not sure how I'm supposed to gradually rotate toward the target every frame when I must give the vector direction toward the target as a parameter.
When I put the vector that points toward the target here, it works, but sometimes it causes my object to spaz out ;-;

Vector3 v1 = rightLegTargets[0].transform.position - leftLegTargets[^1].transform.position;
Vector3 v2 = rightLegTargets[^1].transform.position - leftLegTargets[0].transform.position;
Vector3 normal = Vector3.Cross(v1, v2).normalized;

transform.rotation = Quaternion.LookRotation(Vector3.ProjectOnPlane(<direction here>, normal), normal);

[dire bramble]

yep

[dire bramble]

oh... quaternions... the EVIL

[plush elk]

https://pastebin.com/yHQaQJXN
https://pastebin.com/JE2fUn7w
https://pastebin.com/6MsxkdTp
https://pastebin.com/vuKsX1ve

[dire bramble]

you should definetely try to profile it

there could be many things that could slow it such as asynchronous operations, instantiate issues and bad handled algorithm

[plush elk]

but then it does not show in what onGui is using all of these resources

[dire bramble]

you gotta look for the profiler modules

not the hiearchy view

so you can see frame by frame what is causing what

[plush elk]

if I run the map generator while I am running it is less slow, in scene view is slower, could it be that the async functions when used in the editor are slower?

this last print was taken while in game mode, the spikes persist after the map is generated, but some time after. they disapear

[dire bramble]

thats caused by the generation of the map

[plush elk]

it must be, I have a inspector button to remove the lists of rooms and the moment I do it the performance go back to normal

but I am not finiding what is still using the lists that is my problem, thats why I wanted to use the profiler to see if I could find who were still running

nah I dont know the cene runs at 360 fps

I think its the editor

but I dont know wht

nah I take it back, the fps count goes up again if I delete the list

yeah... have no clue

I think I know what is the problem, I have a list<Rooms>, and each class rooms have a list of<Rooms> that are connected and each of those Rooms from the list have is own list of connected<Rooms>

yeap, problem found 😄

I just replaced the list of connected Rooms to conneded IDs

[sly grove]

Gradual rotation is a separate concern. Figure out the correct target rotation (for example with my suggestion) then interpolate towards it with Quaternion.RotateTowards or Quaternion.Slerp

[undone coral]

Thread

Hi everyone, I'm trying to rotate an

[ivory glen]

Hey guys, with a WebGL build is there a way to interface with the website code?
I've used a .jslib file but I want to import functions from javascript that is included with the website
outside of the Unity build.

[devout hare]

#🌐┃web but make functions in jslib that call those functions that you want

[ivory glen]

Thank you, I've posted in the webgl channel. I have made functions in jslib but not sure how to access the functions outside of the unity build

[hushed fable]

Keep questions visible in a single channel at a time.

[vague flame]

Hello
I have question regarding disabling domain reload when working with c# scripts
I lose a lot of time waiting to test features, so I was looking to disable it (except if there is another way to save time)
So my question after looking the Unity Docs, except for static stuff, is there any side effects ?

[livid kraken]

Except for statics not really. Another way to speed up compilation time is to split your codebase into relevant assembly definitions.

[vague flame]

ok ty

i looked in the docs about assembly definitions, tell me if I read wrong (i didnt got the time to read all the page ofc) : By "default" Unity compile all user created scripts in one assembly file ? (idk if its right term)

so thats why it takes time, because even if you wrote 1 ligne somewher, it compiles ALL other scripts that are in the default assembly def

[flint sage]

Basically yeah

[vague flame]

in the docs i quit didnt understood how the "modularity" works when using AD

except for the compilation time & storage, whats the difference between creating multiple scripts and working with modular methods

[vague flame]

to keep this example of my case, to save time should i just create a AD in my script folder with everything related to my player (mouvements, inputs) ?
so i save time when compiling, because when i reload it will only reload this AD and not all the other unity stuff (if nothing was updated)

[wet burrow]

So yesterday I was investigating a bug that causes the editor to crash, and in the editor log file I noticed the error: Got a UNKNOWN while executing native code. This usually indicates a fatal error in the mono runtime or one of the native libraries used by your application.
@upbeat path pointed out the stack trace and some instances of DestroyImmediate being used at runtime, which were likely responsible. I tracked down the instances causing it and replaced them with regular Destroy calls. Problem is, the crashes are still occuring and now there's absolutely nothing in the 'managed stacktrace' part of the editor log.
Anyone have an idea of what's happening here?

[untold moth]

Is there anything at all relevant in the log?
There should be some error and/or stack trace even if not from managed code.

[wet burrow]

here's everything from where the log starts talking about the crash

=================================================================
    Native Crash Reporting
=================================================================
Got a UNKNOWN while executing native code. This usually indicates
a fatal error in the mono runtime or one of the native libraries 
used by your application.
=================================================================

=================================================================
    Managed Stacktrace:
=================================================================
=================================================================
Received signal SIGSEGV
Obtained 17 stack frames
0x00007ff78bc42a58 (Unity) delete_object_internal_step1
0x00007ff78bc3f1d0 (Unity) DestroySingleObject
0x00007ff78bf5d7fd (Unity) DestroyObjectHighLevel_Internal
0x00007ff78bf5d394 (Unity) DestroyObjectHighLevel
0x00007ff78bd7a77b (Unity) DelayedDestroyCallback
0x00007ff78bd7c664 (Unity) DelayedCallManager::Update
0x00007ff78bf8c5a9 (Unity) `InitPlayerLoopCallbacks'::`2'::PostLateUpdateScriptRunDelayedDynamicFrameRateRegistrator::Forward
0x00007ff78bf736bc (Unity) ExecutePlayerLoop
0x00007ff78bf73793 (Unity) ExecutePlayerLoop
0x00007ff78bf793d9 (Unity) PlayerLoop
0x00007ff78cebd018 (Unity) PlayerLoopController::UpdateScene
0x00007ff78cebb1ff (Unity) Application::TickTimer
0x00007ff78d306b6a (Unity) MainMessageLoop
0x00007ff78d30b41b (Unity) WinMain
0x00007ff78e63ab6e (Unity) __scrt_common_main_seh
0x00007ffe8ccb7344 (KERNEL32) BaseThreadInitThunk
0x00007ffe8dbe26b1 (ntdll) RtlUserThreadStart

[untold moth]

This is probably still related to destroying an object.

What components are there on that object?

[wet burrow]

If you mean the thing that was being destroyed using DestroyImmediate, it's a Mesh

actually unity's garbage collector should handle deleting that stuff automatically

so I'll comment out the regular Destroy call as well

and see what happens

[untold moth]

Meshes are not collected by gc

[tiny pewter]

probably double free

[wet burrow]

It worked!

I commented out the 'destroy' call and the function was carried out perfectly

but I guess I still need to figure out how the GC works properly

[upbeat path]

Glad you resolved it

[wet burrow]

trying to look for this info

[untold moth]

The rule of thumb is that it would clean up (C#)objects when they are not referenced by anything. Assets are a bit different though. Unity keeps them alive in the background. You can call unload unused assets for it to unload them iirc.

[austere jewel]

Calling that is unnecessary if things have been correctly destroyed, it's practically just an expensive fallback

[wet burrow]

hang on in that case what do you mean by 'correctly destroyed'?

[austere jewel]

UnityEngine.Object types should have Object.Destroy called on them when you're done with them (or DestroyImmediate in edit mode)

[wet burrow]

yeah the problem is that running Destroy or DestroyImmediate on the mesh causes the whole program (and editor) to crash

[austere jewel]

It's probably due to when you're calling it, or it's a bug

[wet burrow]

yeah that'll be a massive pain in the arse to figure out

I'm trawling through somebody else's (very complicated) code

I'll have to deal with it tomorrow

thanks for now and goodnight @austere jewel @untold moth @tiny pewter

[hybrid belfry]

Hello. Is there a way to mimic the behavior of Debug.Log, where double clicking the Console Entry takes you to the Call location? I'm implementing a simple Color method but can't get that functionality to remain 😕

[upbeat path]

you can get the correct StackFrame from the StackTrace class
https://learn.microsoft.com/en-us/dotnet/api/system.diagnostics.stacktrace?view=net-8.0

[fallow dune]

I would really love some help. I have a FSM that controls my AI and For the life of me I can see the Navmesh being set but the NPC is just doing his own thing.
I can't get the NPC to reset to initial potion .
https://gdl.space/cavupetiho.m

[untold moth]

What potion?

[native jasper]

I'm not sure exactly what your problem is (there's a million little settings in the inspector for the navmesh, agents, etc that can cause undesired behavior) but id recommend working from the beginning with the simplest script you can create and then build up your functionality, testing your additions along the way. And use gizmos and debug lines to visualize what your agent is doing

[brazen girder]

Can I use Import Package dialog window from code?

[native jasper]

probably a better question for #↕️┃editor-extensions

[humble gale]

may i hop in vc with someone to help me fix a very specific problem in screenshare?

[fallow dune]

Thanks for gaettinb back to me
THis is what hte NPC is doing

You can clearnly see the destination being set yet it walks away from the destination and it looks like it's walking into an invisible wall

[sly grove]

looks to me like your Animator is controlling the player

it's probably overwriting whatever the NavMeshAgent is doing

[fallow dune]

I am not sure I follow. The NPC sees the player and all I am doing is resetting it back to 'Spawn position'

[sly grove]

disable your animator

see what happens

[fallow dune]

sure. 1 sec

[fallow dune]

well that seemed to work

but now the question is,, WTF

[sly grove]

Again, your animator is overriding the position

you've made the classic mistake with the animator a lot of people make

your Animator should not be animating the root object

The visual/renderer piece of your character should go on a child object

the animator should animate that child object's local position

the root object should have the nav mesh agent and move itself that way, unimpeded by the animator

[fallow dune]

Spot on. I appreciate this very much. Thank you .

[cold canyon]

Hello. When I rotate my camera with any interpolation-based (?) method on high "speed" rotation and with enough "smoothness time" it just stops. I tried LerpAngle and Quaternion.Lerp, same result

transform.rotation = Quaternion.Euler(
            Mathf.SmoothDampAngle(transform.eulerAngles.x, _targetV, ref _cameraVelocityEulerV, time),
            Mathf.SmoothDampAngle(transform.eulerAngles.y, _targetH, ref _cameraVelocityEulerH, time), 0);
        
        transform.position = target.position - transform.forward * distance;

Video: https://streamable.com/vrppq2

[fallow dune]

sorry to be a bother but your last suggestion was really solid. I am struggling to get the NPC to rotate correctly.

The NPC spawn rotation is stored on start and upon getting back to it's position I try and reset the position but the AI is doing it's own thing again.

[undone coral]

have you tried cinemachine?

you are sophisticated. you will really like it

[acoustic vapor]

Any ideas for a fix? bullet is not accurate on 60 degrees camera. It's all good when camera is 90 degrees. But my game is on 60 degrees

the projectile launches from rangedAttackPoint and the direction is changing based on mouse cursor

this is my RangedAttackScript:

using UnityEngine;

public class RangedAttackScript : MonoBehaviour {
    private Rigidbody _rigidbody;
    public int damageAmount = 0;
    public GameObject rangedAttackPrefab;
    public Transform rangedAttackPoint;
    public float bulletVelocity = 30f;
    public float bulletPrefabLifeTime = 3f;

    public void InitializeRangedAttack() {
        GameObject rangedAttack = Instantiate(
            rangedAttackPrefab,
            rangedAttackPoint.position,
            Quaternion.identity
        );

        rangedAttack
            .GetComponent<Rigidbody>()
            .AddForce(rangedAttackPoint.forward.normalized * bulletVelocity, ForceMode.Impulse);
        Destroy(rangedAttack.gameObject, bulletPrefabLifeTime);
    }
}```

the codelet that handles player facing rotation:

```cs
private void HandleRotation() {
        float rayHit;
        Vector2 mousePosition = _inputActions.PlayerActions.MousePosition.ReadValue<Vector2>();
        Plane playerPlane = new Plane(Vector3.up, transform.position);
        Ray ray = Camera
            .main
            .ScreenPointToRay(mousePosition);

        if(playerPlane.Raycast(ray, out rayHit)) {
            Vector3 targetPoint = ray.GetPoint(rayHit);
            Quaternion targetRotation = Quaternion.LookRotation(targetPoint - transform.position);
            transform.rotation = Quaternion.Slerp(
                transform.rotation,
                targetRotation,
                100 * Time.deltaTime
            );
        }
    }

[humble gale]

using UnityEngine;
using UnityEngine.EventSystems;

public class Slot : MonoBehaviour, IPointerClickHandler
{
    public bool isBusy = false; 
    public  GameObject itemObj;

     private void Start()
    {
        // Check if this object has a 2D collider, add one if none exists, and set it to trigger
        Collider2D collider2D = GetComponent<Collider2D>();
        if (collider2D == null)
        {
            Debug.LogWarning("No 2D collider found on this object. Adding a BoxCollider2D...");
            BoxCollider2D boxCollider2D = gameObject.AddComponent<BoxCollider2D>();
            boxCollider2D.isTrigger = true;
        }
        else if (!collider2D.isTrigger)
        {
            Debug.LogWarning("2D collider is not set to trigger. Setting it to trigger...");
            collider2D.isTrigger = true;
        }

        // Check if EventSystem is present in the scene
        if (FindObjectOfType<EventSystem>() == null)
        {
            Debug.LogWarning("No EventSystem found in the scene.");
        }
    }
    public void Update(){
        if(isBusy){
            //sets its position to my position
            itemObj.transform.position = transform.position;
        }
    }
    public void AssignItem(GameObject x){
        itemObj = x;
        isBusy = true;
    }

    public void OnPointerClick(PointerEventData eventData)
    {
        if (isBusy)
        {
            Item item = itemObj.GetComponent<Item>();
            // Change isEquipped to true when the object is clicked
            item.isEquipped = true;
            Debug.Log("ITEM EQUIPPED");
        }
    }
}

When i click it nothing happens 😦

[thin mesa]

not really an advanced issue. but do you have an event system in the scene and a Physics Raycaster 2D on your camera?

[humble gale]

oh uhm, its an item that drops that once collected becomes static and is put above a ui slot, what im trying to click here is the ui slot right below that item.

AFAIK all it needs is a collider 2D no?

[thin mesa]

that's all this object needs. you still need those other things i asked about

also if this is a UI object, why does it have a collider on it?

[humble gale]

its not ui, it gets placed in the ui when the item is collected but back to a Dropped item in world when its not

[thin mesa]

i wasn't referring to the item, i was referring to the slot

[humble gale]

oh uhmm

its cuz i asked chatgpt to help me find the issue and it said nothing happened when i clicked cuz it didn;t have a collider

[thin mesa]

i don't know what you mean by "i asked unity to help" but your ui objects do not need colliders

as long as it has some sort of graphic it will be clickable as a ui object

[humble gale]

oop i meant chatgpt *

i can screenshare if you want!

[thin mesa]

well chatgpt is very frequently wrong and now you see part of why

[humble gale]

yep! I never ask it to do code i mostly use it to get an idea of "how to"

but it hasn;t been helpfull

[untold moth]

It's also probably due to you asking wrong questions. You need some understanding of what you're doing anyway to ask the correct questions for it to be helpful.

I've no idea why this conversation is in advanced code channel...

[white pulsar]

i'm trying to find a way to change how the 2DAseprite importer handles importing assets, mainly what i'm looking for is to moddify the source code so the textures it generates are basically the Aseprite file, but with the data from the "Exported" PNG.

What i mean by this is basically have it so the generated texture looks exactly like a regular .png export, instead of what the importer ddoes, where it pads the texture's size to make sure all sprites fit

i'm trying to do this mainly for these small, tiny texture files that i plan to use for VFX using custom shaders and particle systems

I've had some luck by forcing certain aspects to be the same as the aseprite canvas width and height, issues arises whenever i try to do this for a sprite that doesnt fully encompass the canvas (pic related). where instead the generated picture just looks like a mess of pixels

[pseudo trout]

Looking for some insight on this issue. If anybody has any help, please let me know! I've never manipulated Time.time in this manner

[austere jewel]

myTime += Time.deltaTime * movementScale

Even better, just use cinemachine and blend between noise profiles https://docs.unity3d.com/Packages/com.unity.cinemachine@3.0/manual/CinemachineNoiseProfiles.html

Generally if you're doing camera stuff and you're not using Cinemachine, you should be

[bitter magnet]

that's not really going to be different since that's just a manually created time.time

[austere jewel]

why would it not be different

[bitter magnet]

because it already uses time.time??

[austere jewel]

It's not using Time.time, it's using a scaled Time.deltaTime

[bitter magnet]

this is basically the bob code right here

[pseudo trout]

never used cinemachine before but heard its very powerful. guess i should be

[pseudo trout]

speghetti at this point

[bitter magnet]

yeah unfortunately

[bitter magnet]

also it used to be * time.time but now it's + time.time as a test

it did cause less jittering but

it also y'know doesn't look right

[pseudo trout]

jittering is still jittering

[austere jewel]

I have no idea what you're talking about, but the original question was how to scale time so it doesn't jump around

[pseudo trout]

anyways i'll look at cinemachine

[austere jewel]

The answer is to track time yourself, and added the scaled time difference, ie. add scaled deltaTime

[bitter magnet]

that wasn't the original question but ok

[austere jewel]

I have no idea who you even are, so 🤷 you do you

[bitter magnet]

wth does that have anything to do with anything ☠️

[austere jewel]

I don't know why you're telling me that my interpretation of the question is wrong, when you're not even the one who asked the question

[bitter magnet]

I'm the one who wrote the code, I work with squidy

[pseudo trout]

im just trying to ask questions not start an argument 😭

[austere jewel]

Then maybe rephrase the question so it becomes clear why my code answer wouldn't work, because you're probably just misunderstanding it

[bitter magnet]

the question was how to move along the graph at the speed of movement speed instead of scaling the graph based on movement speed

[pseudo trout]

to make the question more clear, I want time to track normally (literally time.time) BUT i want it to go faster based on movement speed. (Time.time * movementSpeed). the issue is that i want it to be additive. so for example, I move 1 unit per second for 5 seconds. that means the current x coord on the wave would be 5. after moving that amount i move 2 units/sec for 3 seconds. it should add 6 to the previously elapsed time. making it 11. does this make it more clear? @austere jewel

that way i can just plug in the x coord to move along a static graph

[austere jewel]

Hence my answer.

[pseudo trout]

ok thank you sorry for the confusion with ender

[upbeat path]

dont cross post, but yes it is widely known

[lusty basalt]

I wasnt sure about which category to post in, can delete

[upbeat path]

definitely not advanced

[rancid oxide]

I'm trying to make a simple shooting star (its green so u can see it), I've added the trail to the point I move in arc but I noticed that the trail start transparent, then turn color and after that goes transparent again

I need it to be attached to the ball i generate

How can I keep the color in this section?

There is no fade to alpha so I can't figure out where is the issue

these are the Trail settings

[sage radish]

Looks like you might be using a particle texture, which is a circle. That's what a stretched out circle looks like.

[lusty basalt]

try setting the texture scale really high or really low and see if you get a fully collored low-poly line

[rancid oxide]

maybe I should also change the texture mode?

also what material should I use then?

look like standard material doesnt work on particles

[rancid oxide]

this is with a scale of 10

[humble leaf]

This isn't a coding question. Use a texture that is solid.

[rancid oxide]

omg sorry I thought to be in the vfx channel .-.

[regal olive]

Hi, working on a Baldur's Gate 3 like quest. I have movement working using NavMesh. I am trying to get the path that is being taken to display on the screen. Right now what I'm doing that is not working is using a child object attached to player, giving it a line renderer and a script that tries to display the line. Anyone tackled this that can help me out? (Script available upon request) Thanks!

[lament salmon]

That's how most people do it, in what way does it not work?

Make sure your linerenderer is in the correct space (local/world)

[timid sinew]

hey i was wondering if anyone knows a way that Unity supports F#(even if its a separate open source library made by the community) that doesnt force me to use a ddl file?

its not really a big thing for me but still

[sly grove]

Very unlikely

[prisma ravine]

I'm trying to instantiate a particle effect prefab (ParticleSystem) in code. Everything I've been able to find says I should be able to do it like this:

var particle = Instantiate(particlePrefab, parent.transform.position, Quaternion.identity);

but this throws an exception:
InvalidCastException: Specified cast is not valid.
(wrapper castclass) System.Object.__castclass_with_cache(object,intptr,intptr)
UnityEngine.Object.Instantiate[T] (T original, UnityEngine.Vector3 position, UnityEngine.Quaternion rotation) (at /home/bokken/build/output/unity/unity/Runtime/Export/Scripting/UnityEngineObject.bindings.cs:295)

If I wrap it in a try/catch, it seems to work as long as I have "play on awake" selected, and "stop action: destroy" does actually destroy the instantiated prefab, but after I instantiate a certain number of these, the client crashes.
Anyone know why I'm getting that exception?

[thin mesa]

not really an advanced issue. most likely you changed the type of your prefab variable and just need to drag the reference in again because it's still serialized as a GameObject reference

[prisma ravine]

Dang, that was it. Thanks!

[radiant spear]

Does anyone know what returns from Touch.Radius? I'm receiving wildly different values from different devices that have similar specs. Does it differ per brand/device? For instance on a Google Pixel 7a I get on average 130 which translates well to pixels, but on a S23 Ultra I get an average of 2, which might be millimeters according to some threads.

In the docs it's described as "An estimated value of the radius of a touch", but not in what units.

[undone coral]

Does anyone know what returns from Touch

[timid sinew]

unfortunate friend thanks for the answer, i prefer to mix OOP and Functional Programming

due to the inherent benefits when it comes to certain aspects of Game Developments for each one

but i guess i wont be able to use the cool other language for it lol

[lofty inlet]

for(int i = 0; i < 5; i++){
objArr[i] = 5;
}

vs

objArr[0] = 5;
objArr[1] = 5;
objArr[2] = 5;
objArr[3] = 5;
objArr[4] = 5;

which is faster?

[devout hare]

If there's any difference it's so vanishingly small that you can't even measure it. Don't make your life harder with useless micro-optimizations

[lofty inlet]

that was just an example
what if the array was 1million in size

I just didn't want to type the second block with 1million lines

[devout hare]

Then typing out a million lines to avoid a loop would be even sillier

[lofty inlet]

can you please answer the question (if you know it)

[devout hare]

I did

[lofty inlet]

the million object one?

[devout hare]

The answer is the same regardless of how many lines you have

[lofty inlet]

there's no way that's true

I just realized the second one has got to be faster
since you are not iterating over the i variable

unless the for loop somehow caches the array

[devout hare]

Well actually if you have millions of separate rows then in practice the computer might need to use extra resources to handle the whole thing so it might even be slower

[lofty inlet]

ohh like the actual file size

true

hmmm so you're saying they're about the same if the array is like 20 objects or less

same speed that is

[tiny pewter]

unrolling is faster

and you hard code the value, save one register

and no branching if you hard code all the things

[lofty inlet]

cool thanks

[flint sage]

The runtime (or compiler) might even do loop unrolling for you

So it's hard to say if there's a definitive answer

[lofty inlet]

damn

[flint sage]

Besides, these questions are so vague that there's no way to properly measure since it depends on a 100 different variables

[novel plinth]

for-loops perform bounds checks to ensure that the loop index stays within the valid range. So the hardcoded on would be faster.
But you should not care about this... it's too silly

[flint sage]

Go and measure your actual use case, on your actual target hardware

And the loop is obviously a lot safer, which makes the performance pretty irrelevant unless this is the hottest of paths

[upbeat path]

if we are talking clock cycles option 2 is considerably faster with very large arrays. The compiler will be able to determine an exact address whereas in a loop that address will need to be evaluated at runtime

[flint sage]

Except.. it could do some optimization since the runtime knows you're iterating an array (and thus touching all elements) instead of in arbitrary order

Meaning it might be able to load things more efficiently into the cpu cache

[upbeat path]

wishful thinking, the compiler is not that clever

[flint sage]

I mean we're talking about hypotheticals at this point, since you can't talk about this in a generalized manner since it depends on the platform etc

[fickle mountain]

Anyone has experience toying with Camera.projectionMatrix or Camera.worldToCameraMatrix?
I'm trying to have an effect that change the world's scale, without the hassle of actually scaling the level itself.
Pretty sure what I need is somewhere in one of those two, but no matter what I try I can't get the effect I want.

[sly grove]

Are you just trying to zoom the camera?

[fickle mountain]

Not really. Changing the scale. Having the camera render the perspective of a 1x1x1 object as if it was 1000x1000x1000.
It's to make an effect as if you were turned into a mice or a giant (think Alice in Wonderland, or CS maps like De_rats).
Right now I'm just scaling the entire level, but that break a ton of things.

[sage radish]

Is this in VR or something? Because without stereo, there's nothing you can do to the camera projection to make the scale look different. The only difference would be how far the camera moves relative to the world.

[sly grove]

Just kinda sounds like zooming/moving the camera closer/further combined with a depth of Field postprocessing effect

The bigger concern would be modifying physics behaviors

[timber flame]

I want to distort a flow texture in my voxel based water shader according to directions.
For each water voxel, a direction is calculated. My approach is to change texture uv based on the direction and its magnitude. What is your idea?
Also, because the directions is calculated for each voxel, all pixels belonging to that voxel will have the same direction. To resolve this problem and change direction smoothly, I need an interpolation technique. Passing all directions of neighbors and interpolate between them based on the distance

[fickle mountain]

I do have VR to tackle later, haven't researched that part yet. And yeah maybe there is just no way to do that in "flat" rendering, which would explain why no amount of weird formula is doing something I'm satisfied with. Crap.

[lament salmon]

Modifying depth of field can also contribute to the scaling effect

[brazen girder]

Has anyone done code generation? Here I have a SO, and I need to generate fields with the same values as in this SO in a regular empty class. What can be used for this? Does anyone have any thoughts?

[upbeat path]

Reflection is your friend

[sage radish]

For something like this, where the generated code is dependent on a Unity asset, it would be easiest to just write the code generator by hand, as in piecing the source code string together and then writing it to a .cs file in your project.

The bigger question is when to trigger the generator. You could try to implement an automatic trigger, but you probably don't want Unity to recompile every time you change a single field.

[brazen girder]

Well, in the file .cs it was like this

    /* App constants */
    public const string AD_INTERSTITIAL = "ad_interstitial";
    public const string AD_REWARDED = "ad_rewarded";

1. clicked on the button in the editor
2. the data was pulled from the scriptable object
3. and it became like this

    /* App constants */
    public const string AD_INTERSTITIAL = "new_value";
    public const string AD_REWARDED = "new_ new_value";

[upbeat path]

shall we take this to a thread?

[brazen girder]

Okay

[inland knoll]

Hello! I'm trying to make an enemy npc who can essentially hide behind objects/walls, perhaps something like you'd see in GTA with the cops. However, I'm currently struggling on figuring out how to do this. The first thing that came to mind was something like in BTD 6 (aka on the picture) where it creates this red area in places where the troop cant reach. I'm wondering that maybe I could do this for the player, get the two points where an obstacle originates from and set it as a cover point for the enemies. Another idea is to just put gameobjects to the places where the enemies could take cover, however this feels very inefficient, plus I'm certain that GTA does this at runtime.

[gusty kernel]

I got a similair problem like that once. I fixed it by drawing a line from the player/target to the enemy. If the line touches an object first (you can work with layernames or tagnames), then the enemy knows it's safe and won't the player/target shoot.

[inland knoll]

It's not exactly about the shooting part. It's about finding the nearest cover to the enemy which the player is close to. I'm trying to make them like go against a wall or behind a box and peek out & shoot like in GTA.

[gusty kernel]

Oh. In that case would I place zones which are the shooting/hiding spots. Then the enemy will search the closest zone (using 2d/3d distance calculations in a loop) and will move to that place. It'll stand on a specific distance from the corner and maybe you can trigger an animation which shows the shooting.

If a spot is already taken, then you have to make sure that the hiding spot is removed from a list containing all hiding spots. Then other enemies will search for another spot. (You prob don't want 2 enemies glitched into eachother)

I would do it like this, but maybe there are other more effective ways.

[undone coral]

the projection matrix does not really change the sense of scale. an ant has a FOV that is greater than a human's, which is what a projection matrix is. however its "camera" (eye) is located very close to the ground, and it can get very very close to things

[undone coral]

try to first write the function signature for what you want to do

[inland knoll]

So like write pseudocode?

[undone coral]

you can start with pseudocode, but i think there's a lack of clarity of what values go in, what value goes out

by the time you write this method signature you will have 90% solved your problem

[timid sinew]

have you tried getting the vectors of all possible points and calculating from there?

[inland knoll]

I haven't done anything yet. That question was more of a "How would I go about doing this" question, since I didn't really know what to google nor did I have any good ideas of my own

[timid sinew]

does your system already include something where it calculates where exactly(all the possible points) the NPC can go for cover?

or do you need to do that as well

im unaware what gta does but the more duct tape solution is to place areas on the prefab which indicate where an NPC can possibly look for cover

[inland knoll]

Nope, I don't have that yet.

[timid sinew]

the other way would probably be

if i were to take a guess

[inland knoll]

On bigger maps with all sorts of angles and stuff tho, this seems very inefficient to me

[timid sinew]

some sort of equation including the rotation of both characters and their vector 3s and the possible corners and/or something that has to do with raycasts

id have to think about it a bit more if i actually wanted to make a solution though

heres a solution i found atleast

In real-time calculate corner areas wide enough for the NPC to hide and that aren't exposed to the player. Choose the nearest for the player to hide.

https://gamedev.stackexchange.com/questions/4374/whats-the-best-practice-to-have-npc-enemies-taking-cover-in-an-fps-game

[timid sinew]

he basically says the same thing i say

[lament salmon]

I personally achieved that by reading the NavMesh triangulation and generating cover points according to the edges. This happens at runtime when the game starts/map is generated but can be baked in editor too.
Then at runtime I get some near points and score them by visibility, cover, distance from self, distance from target etc.

If you don't use a NavMesh then you could just brute force it by picking random points around the player, raycasting from those points towards the target player. Discard visible positions and score them by distance etc.

[inland knoll]

How would you know if the point is at a corner or just in the middle of a wall?

[lament salmon]

Does it have to be a corner?

[inland knoll]

The enemy cant peak if it's in the middle of the wall I mean

[lament salmon]

Yeah ok, do you use a NavMesh?

[inland knoll]

I do

I think I could actually even be fine if I get a point in the middle of a wall, then generate a new path from there to the player and stop at the point before the point where a raycast would hit the player

[lament salmon]

I loop over the edges in the NavMesh and bake a "visibility score" to a few directions from each generated point

[inland knoll]

That's also pretty interesting

I'll see what I can do tomorrow

[lament salmon]

I like to visualize this kinda stuff, easier to develop:

I also bake different firing poses (lean left/right, center) which is visualized here by the green stick figure 😄

The rectangles on the ground are the generated points.
Green points have a line of sight towards their cover direction from any pose. Red ones are bad for shooting, they just provide cover

So I generate as much info as I can when baking, and do some additional checks at runtime

[jagged fiber]

How can one exclude libraries from build in a safe way? (To avoid "The type or namespace name X could not be found") .
I excluded certain server-only things from client, but an android build crashes with "The type or namespace name X could not be found" exception (I don't use the components, that use the excluded libraries on the client).

When launching in the editor, the client build runs just fine.

[upbeat path]

select the library in the project view and then in it's inspector select the build platforms

[jagged fiber]

So I right-clicked on the folder that will be excluded, what's next?

[upbeat path]

not the folder the library itself and who said anything about right click?

[jagged fiber]

As far as I understand the "Project" view is a set of folders

[upbeat path]

and folders contain files

[jagged fiber]

yes, so how do I get to the inspector you mentioned?

[upbeat path]

This is one of mine.
In the folder select the library, that shows the inspetor

[jagged fiber]

Oh, I see. Doesn't cover my case though unfortunately, as platforms there are just platforms, not "Subtargets" (e.g. any platform could be a client or a server). Is there any other option?

will look into "Define Constraints" there

[upbeat path]

but you said you wanted to exclude for Android, so select include for all platforms except Android

[jagged fiber]

No, I already completed exclusion as I said, I want to avoid an exception that says "Something is missing".
The exception is thrown for no reason right at launch (I don't invoke the methods from the missing library)

the exception is thrown in real android builds, but when building "android" and running in editor, it runs fine with libraries excluded correctly

[upbeat path]

sounds like you need to increase your code stripping level

or switch to IL2CPP backend

[glacial wedge]

Hi. my game lag in editor

why I have 100ms of update?

50ms of shaderGUIUtility?

with deep profile

[jagged fiber]

thanks, I tried it, didn't help, unfortunately.
Seems like I'll have to exclude "using" and calls to excluded libraries via #if UNITY_SERVER even from components that are never added to game objects in client builds for Android specifically😦

[austere jewel]

Look at what class is calling that.
Your material count is also increasing disturbingly fast.

It may be easier to see in the profiler's hierarchy view

[sage radish]

The editor might be spamming warning logs about that vector property not existing in that material/shader.

[glacial wedge]

the property exist, no warnings.

I found out that i was wrongly call an instantiate now it is running better, but still not enought i'm still checking

[glacial wedge]

why "first call"?

and shouldn't be better use gpu instancing instead of srp batcher for render my multiple tiles?

but trying to use the materialpropertyblock with gpuinstancing material it still try to use srpbatcher with tons of batches caused by "different materialpropertyblock"

[untold moth]

First call from the job. There was nothing to batch it with, since it was the first batch apparently.

What are you trying to achieve anyway? Sounds like you're doing premature optimization.

[glacial wedge]

but it wasn't

[untold moth]

How do you know that?

[glacial wedge]

because the previus call draws half of the tiles

[untold moth]

Well, there are no guarantee that they are in order and/or batched in the same job.

80 draw calls /batches is very little. I don't understand what you're trying to do here.

[glacial wedge]

I want to use materialpropertyblock, but if i do 80 becames 500

[untold moth]

Why do you want to use it?

[glacial wedge]

i read that for multiple simple objects like my floor tiles it works better

[untold moth]

It would work good if you have several tens of thousands of these objects. And then again, it would just offset the work to the gpu. I bet that there are other things that actually affect the performance more(like culling and stuff)

[glacial wedge]

true

[untold moth]

You still didn't explain what you're trying to do..?

[inland knoll]

How did you get the edges of the navmesh like that? Did you just call Navmesh.GetClosestEdge on each vertex and compare it to the position of the vertex?

[lament salmon]

https://docs.unity3d.com/ScriptReference/AI.NavMesh.CalculateTriangulation.html

Wait it's been a while lemme check

[inland knoll]

Oh that returns the edge points?

[lament salmon]

Actually yeah seems like I used GetClosestEdge also

[inland knoll]

Alright, I'll cook something up with that then

[lament salmon]

Yeah, lemme know if you have troubles I can compare our approaches then

I totally improvised this

I check all the edges in each triangle and do a FindClosestEdge on that edge's center

Then I do a dot product check between:
A. The normal of the NavMeshHit from FindClosestEdge
and
B. The normal perpendicular to the edge we are currently iterating
If the dot is > some threshold (I just used 0.9) then it is a valid outer edge

@inland knoll Hope that helps

Hacky but it works

[inland knoll]

Oh, I was just planning on calling FindClosestEdge on each of the vertices and checking their distance.

I don't exactly get the point of getting the normal and the dot stuff. What would be a not valid outer edge?

[lament salmon]

But a vertex is the same distance away from two different edges

Since it is shared by those edges

[lament salmon]

An edge that is between triangles but not a solid/unpassable edge

[inland knoll]

But it would return the distance as 0 if it's on an edge? I don't think it gets the distance to the center of the edge

[lament salmon]

You only want the outer edges. So the green ones here

You don't care about the passable edge between the tris

[sly grove]

when I try to change the value, The set is not being called.
What do you mean? How are you trying to "change the value"? What function are you calling? What set are you expecting to run?

[inland knoll]

Ohh

[lament salmon]

@inland knoll FindClosestEdge finds outer edges only, not edges between triangles

That's why I use it to check if it's on the outer edge

[lament salmon]

Makes sense?

[inland knoll]

Yep

Alright, so we're finding the outer edges not only the vertices, right?

Since I was trying to get only the vertices before. I guess edges are actually a lot better to get, lot more info with that

[lament salmon]

Yeah, you want to work with edges

They are just a pair of vertices anyway

You can then generate some points along those edges and check their cover score and such

[viscid remnant]

I fixed the problem, sorry for not detailed more about, I will take care in the future to make more explicit what I need help for

[inland knoll]

Is this what you meant? ```cs
private bool IsOuterEdge(Vector3 vertexA, Vector3 vertexB){
Vector3 center = vertexA + (vertexB - vertexA)/2;

    NavMeshHit hit;
    if(!NavMesh.FindClosestEdge(center, out hit, NavMesh.AllAreas)){
        return false; //For some reason, found no edge
    }

    Vector3 perpendicularDir = GetPerpendicular(vertexB - vertexA);
    float dot = Vector3.Dot(hit.normal, perpendicularDir);

    if(dot > 0.9 || dot < -0.9){ //Since getting the perpendicular direction might have given us the negative of what we want
        return true;
    }

    return false;
}

private Vector3 GetPerpendicular(Vector3 vector){
    return new Vector3(vector.z, vector.y, vector.x); //Switch the X and Z coordinate.
}

[lament salmon]

Yeah but make sure to normalize the perpendicularDir

So that dot product is in the -1..1 range

[inland knoll]

Oh right

[inland knoll]

It's kind of working, however a lot of edges all over the place too. Gotta do some tweaking. I believe one problem is that if a random edge in the middle of the play area is aligned with a wall which is the closes edge on the navmesh, then it takes that as a valid edge, I think a simple distance check would do tho.

[inland knoll]

Wouldn't an edge qualify as an outer edge if the center's distance to the closest edge is 0? This would eliminate the need for the dot would it not?

[lament salmon]

Looks like I also did a final NavMesh.SamplePosition check from the edge center before adding the edge. I didn't comment it so not sure what it solved but it's probably needed 🤔

[lament salmon]

Looks like I tried that first, but commented it out and used the dot product.
I can't recall the reason. You can try it though

[undone coral]

it's really frustrating that unity doesn't expose the underlying navmesh data structure. they are using recast (https://github.com/recastnavigation/recastnavigation) which is open source anyway. it's a real dick move.

@inland knoll it might be easier to adopt granberg's A* because then you can query this stuff directly on a navmesh.

[lament salmon]

Actually, I think the reason is that NavMeshHit doesn't actually give you the indices of the edge's vertices. And the NavMeshHit's position is not in the center of the edge.
So you don't know if it's the correct edge or not. Had something to do with edges close to each other or something.

[undone coral]

if you need to find the edges of geometry (i think you mean the corners), there are standard algorithms for that. you can use it on the triangulation.

[inland knoll]

OOh

Now we're talking

[lament salmon]

@inland knoll The dot product's point is to check if the point is along the edge. You don't know what is the center of the edge that was hit

[lament salmon]

There we go!

[inland knoll]

Why do I need the center of the edge which was hit?

Oh wait, you were talking about a different thing than I thought, mb

[lament salmon]

Yeah. Are you currently using the dot product approach?

[inland knoll]

No, this is the distance check approach

[lament salmon]

You can try with a distance check, but there's a reason I changed to the dot product, but I can't recall what exactly

[inland knoll]

This one was with the dot

[inland knoll]

At least it's working for now. I'll keep the dot check commented out if I'll need it

[lament salmon]

My issue might've been that I had multiple overlapping NavMeshSurfaces or something

Use the distance check if it works for you

[undone coral]

@inland knoll well in case you need something more robust, you can use geometry3Sharp and load the navmesh triangulation into it. then, simply enumerate the edges and ask it directly if it is a boundary edge
https://github.com/gradientspace/geometry3Sharp/blob/8f185f19a96966237ef631d97da567f380e10b6b/mesh/NTMesh3.cs#L866

it looks like you are going to want to do more advanced queries on your navmesh surface eventually so this is probably going to be helpful to you. it explicitly supports unity and i think it is used by many games

var mesh = new DMesh3();
foreach (var v in vertices) {
    mesh.AppendVertex(v);
}
for (int i = 0; i < indices.Length; i += 3) {
    mesh.AppendTriangle(indices[i], indices[i + 1], indices[i + 2]);
}
var boundaryEdges = new List<Index2i>();
foreach (int eid in mesh.EdgeIndices()) {
 if (mesh.IsBoundaryEdge(eid)) {
  var edge = mesh.GetOrientedBoundaryEdgeV(eid);
  boundaryEdges.Add(edge);
 }
}

@inland knoll pretty much all it would be once you've triangulated the navmesh

but... what is this for?

[inland knoll]

This is the start of the conversation

[lament salmon]

@inland knoll Yeah I just checked, I was using dot product because some of my navmeshes were overlapping so the distance check could catch a different edge

[undone coral]

i see... it is good you shared an actual screenshot of your game

it was kind of confusing

you want npcs to take cover

that's what i'm hearing

in something that looks like XCom

generally those places are designed. like you can't take cover on an exterior wall

[lament salmon]

What's wrong with automating it?

I personally did it because the maps are procedural

Of course it's good to have the option to hand-author them too

[undone coral]

you can procedurally determine places that you can get cover. first the navigable areas are painted by the designer for stuff that actually makes sense to take cover on. let's call that the "coverable area." then the coverable area is voxelized, and you query which voxels can be seen by the player (raycast from voxel to player) and remove them. then the npc navigates to the voxel closest to the npc.

[undone coral]

nothing it's definitely doable

[inland knoll]

I have already thought about and discarded that idea. It'd be pretty tedious having to go and map out all the positions, and walls and stuff, mostly on bigger maps, plus its more adventureous for me to do it like this.

[undone coral]

procedurally i guess there are a lot of ways to do this. at a high level:

find all the boundary edges (exterior and interior) of a navigable mesh surface
compute a convex edge for the whole navigable mesh. remove all the edges that are both in the mesh and its convex counterpart.
 this leaves all the concave edges of the exterior of the navigable surface and the interior edges.
using the concave exterior edges and interior edges, place voxels inside the mesh bordering these edges.
keep the voxels that are wholly contained in the mesh. this is your coverable area, procedurally determined.
whenever an npc needs cover...
  raycast from each voxel nearby the npc to the player.
  if the raycast hits the player, remove it as a potential location.
  choose from the remaining voxels heuristically and move to it.

[undone coral]

yeah i mean, are you going to add geometry3sharp (or geometry4sharp the updated fork) to your project?

[inland knoll]

Probably not, I'd rather make these things on my own, however some time in the future I might, if I'll remember this.

[undone coral]

there you go.

i mean i can only help you do stuff that makes sense. right now your decisionmaking is "whatever @inland knoll feels like." it isn't sensible. you can't in the same breath reject an external library - i got news for you, Unity is an external library - and then also complain about it being "pretty tedious to go and map out all the positions"

you have to do stuff that makes sense

[lament salmon]

What's your issue? I guided him to a working solution

How does it not make sense?

[undone coral]

it is a long journey to figuring out how to remove the exterior edge or interpret it correctly for taking cover

[foggy verge]

hi everyone ima need some help

[undone coral]

if the user is going to author it from scratch.

[foggy verge]

anyone here knows how to balance games

[undone coral]

anyway you'd probably agree that the user should use a library

i'm not saying you're right or wrong about anything it's good that there's something that works

[thin mesa]

game balance is a design issue, perhaps you want to describe your game and balance issues in #archived-game-design if you need help with that

[lament salmon]

Why? What's wrong with the current solution?
I'm literally using it in my own game and it works well

You are implying that it is faulty

[foggy verge]

nah i actually need help in formulas

[lament salmon]

And that there's a need to rely on 3rd party libraries, which many people want to minimize, including me

[remote pumice]

you probably should give some more details and how exactly this is related to code..
balancing in itself is basically tweaking numbers

[undone coral]

i think there are some more steps between what is shown in that screenshot and procedurally finding locations for cover

[foggy verge]

how do u come up with formulas like this (1-sign(floor(x/200001)-1/2))/2*10*(139+1.55^139*1.145^360*(Product[1.145+0.001*ceiling(i/500),{i,501,x}])+((1+sign(floor(x/200001)-1/2))/2)*ceil(1.545^(x-200001)*1.24e25409+((x-1)/10))

[undone coral]

are you trying to make an idle game

jk

you can write about it #archived-game-design

[foggy verge]

{
    Block, // converts damage into defense more effective on shields
    BlockPct, // block 3% of damage, 
    Defence, // Block +X damage 
    Immunity, // every X turns turn immune of x cooldowns
    Defect, // defect x% of damage back 
    Parry, // 1% chance to completly block the attck
    Motivation, // take x% less damage for first 5 secs 
    Shield, // + x health
    RockAttack, // every 2 attack throw rock to enemy 
    EarthQuake, // slow both you and enemy down but Both take more damage
    Dust, // enemy attacks are x% slower
    HeavyWeapon, // attack very slow but do large damage
    Dig, // Every 10 cooldown dig inside for 2 cooldown
    BuriedTreasure, // on death recive X amount of gold 
    Mineral, // gain x ruby per victory 
    LevelBoost, // all ice of this weapons +x levels
    FullBlock, // fully block the first hit of the battle 
    RandomTypeBoost,
}``` this is like one out of 10 blocks

and i need to balance the skills

[lament salmon]

Yes, I personally generate some points along the edges and bake some visibility and cover data into them. I explained it to them before

[foggy verge]

yes lmao

[lament salmon]

Then at runtime you can query those points and score them according to distance, visibility, cover etc.

[undone coral]

how do you deal with cover on the exterior edge of the navmesh

[foggy verge]

heard of inventory idle?

[undone coral]

which is only cover if it's on a concave part of the edge

[lament salmon]

Detecting cover? Simply a bunch of raycasts when baking

It is a bit more complex than it sounds, I admit

But doable

Each cover point has info about its cover angles and available shooting angles

[undone coral]

the player is the green star, the npc is the red star. without dealing with the exterior edge correctly, all the blue areas would be cover. if the npc navigates to the closest cover area, it will be in kind of a bullshit place

i agree it's complicated.

[lament salmon]

I don't just score the cover point by distance when querying.
I calculate the visibility to the target and cover amount from the target

You can see some of my personal debug info in this #archived-code-advanced message

[undone coral]

i think without explicitly dealing with the exterior edge, you will have issues with one or both of the situations above

there's no heuristic.

[lament salmon]

Tbh I'm not sure what you mean with 'dealing with the exterior edge'

[undone coral]

the exterior boundary of the traingulated navmesh ground surface

let's say you omit the exterior edge
okay, because you've omitted the exterior edge, the npc does not see the cover that makes the most sense

so there are two situations that need to be dealt with. if you treat all the areas around edges of the navmesh surface as potential spots of cover, you must deal with the exterior border of the navmesh specially.

it cannot be dealt with heuristically.

there is definitely something there with whether or not the edge is concave. but it's an open surface in 3d, so i can't imagine querying that ad-hoc. you'd have to use a library to determine concave edges of a 3d surface

[lament salmon]

I think I see what you mean now, the very exterior edge
I don't generate any cover points on edges that don't have any cover, detected by raycasts

So they are basically ignored

[undone coral]

i think you're back to the first illustration then

as an issue

there are points on edges that have cover detected by raycasts in that illustration

you'll see that the outermost border / exterior edge of illustration 1 indeed has areas of it that are not visible to the player according to raycasts.

[lament salmon]

The whole idea is to have points next to solid geometry

Not just points that are not visible to the player

[undone coral]

okay... well then you are in the issue in illustration number 2, which is that sometimes the exterior border of the navmesh could or could not have "solid geometry," where solid geometry is meant to be a heuristic that is supposed to include the part of the level's exterior border that is concave.

it really depends on definition of solid geometry. maybe the whole level is made out of kitbashed stuff

you'd still have to write a bunch of heuristics for solid geoemetry, you couldn't do it in a general way

maybe that geoemtry that is making the exterior border of the level isn't "solid"

i don't know. one thing i know for certain is, if i want to query facts about geometry, i'm going to use a library.

@inland knoll anyway, you know, you use libraries. you better start believing in libraries, because you're in one (unity). if you don't want to use libraries, it's your game, write stuff that is error prone. i mean i'd use the word buggy, but maybe you don't make the errors and don't write the bugs. i don't know.

you can blub through a bunch of heuristics.

that's a perfectly good way to do this

it will be faster and less buggy to hand design the valid cover areas

you're going to want to exclude valid cover areas from ever being used as cover anyway

just stylistically

so you will need to do it by hand eventually.

i have literally never met someone on this chat room who has had a valid reason not to use a good external library. not even the people trying to minimize load times for webgl, because it's always going to take longer than 6s, so you will have lost most of your audience due to loading no matter what you do.

i guess education was valid

that's pretty much it

if the academy required that you not use libraries

[inland knoll]

If I were to only use libraries, I wouldn't learn. The more I handle these kinds of situations on my own, the more capability I'll have in the future. If you wish to praise libraries as greek gods and ignore other possibilities such as making it yourself, you go do that. I just rather do these things on my own, learn and get experience. What if one day, there isn't a library that would do the work for you? Also, I'm not on a deadline, so I can just make progress as I like and make the things how I want them to be. Another good thing about doing it myself would be that I actually know what's going on and I can modify the code, or use it as a base for future things more easly.

[undone coral]

yeah i mean, you can write a mesh analysis library

i don't know how much there is to learn from that experience.

you can do this

[lament salmon]

Still not certain what you mean. It detects solid geometry on the exterior edges like it should.

Rectangles are cover points

[undone coral]

okay well, there are definitely cover points there that will not make sense most of the time

if ever

[lament salmon]

Super minor issue

[undone coral]

i think it will happen all the time. i mean look at my illustration

[lament salmon]

My thoughts exactly.

[undone coral]

in this situation, if the npc who is visible to the player right now is seeking cover, the closest cover is in kind of a bullshit place

i am actually not sure if there is a general way to deal with this, which is why i think people hand design valid cover areas

[lament salmon]

That's why the points have their visibility angle data cached. Like I said many times.

You query that at runtime

[undone coral]

while i don't know how that solves the problem in the illustration, supposing it does, surely you see that "an agglomeration of heuristics" is harder to get right, in every sense, than "specify the areas by hand". even in a procedural level generation setting, i would probably generate cover areas directly and build geometry around it, or something like that, then do it as a procedural pass afterward

i mean you're a reasonable person of course you agree

just because something exists - just because it's even your code and it exists today, right now - doesn't mean it makes sense

[inland knoll]

Hold on, how would your approach of setting the cover areas by hand fix the issue? You'd still have the exact same problem, but now you don't have the data to figure out the most optimal cover

[lament salmon]

I just recreated your illustration with my tester script of my system, and it works just like I want it to.

[undone coral]

lol i'm not 100% sure! this is a really complicated problem!

[undone coral]

can you move target to the right gradually? try to reproduce my illustration

[inland knoll]

In that case you had no reason to fight the approach which gave better data

[undone coral]

you haven't actually recreated my illustration. it's essential that there be green dots on the border of the lower cargo container that are definitely not visible to the target but closer to self than the ones in the upper container

you have to reproduce my exact illustration

target should appear to be directly above self, but still visible to target

but i believe you

[lament salmon]

Directly above

[undone coral]

okay well

that's kind foa weak choice

[lament salmon]

(Ignore the missing edges, was messing with some thresholds)

[undone coral]

there are a lot of spots that are closer that would be better cover

than the place it chose

[austere jewel]

The system seems to be working great, well done Osmal

[lament salmon]

Thank you

The heuristics do need some more tuning, but these bots are already wrecking my ass in game

[undone coral]

i mean this would be the best spot no?

[austere jewel]

If you found any issues, I imagine it would be minimal effort to write an extra heuristic to prune them out if you required

Or hell, just use this as a fantastic starting point to authoring

[lament salmon]

You can't shoot from there, my dude

[undone coral]

same with this illustration

although it is already not visible to the target

[undone coral]

i think we are moving the goalposts

[inland knoll]

Were not just looking for the furthest cover point from the player, but somewhere where the enemy could shoot

[lament salmon]

Did you really miss the part that it's a shooting position with cover, not just a cover position?

[half swan]

Yes, you certainly are

[austere jewel]

Seeing as you set up the goalposts I don't think it matters lol

[undone coral]

well you can't shoot from the position it chose in that illustration either

[half swan]

Yes it can
It's at a corner...

[woeful kraken]

how bout this - Osmal likes it. There. End. Finito.

[undone coral]

in #2 in this picture

i mean it "can" "shoot" form there

the npc can move

[inland knoll]

However a step to the right would provide that ability. The point is that you can be in cover, but at the same time you can also shoot by peaking out

[austere jewel]

Even if you authored this cover you would still need a step in code to validate the things you are now arguing about

[lament salmon]

My NPC's can lean to left and right

That's why it looks like that. The green stick figure resembles the shooting pose

[undone coral]

i'm not really arguing one way or another. i guess if i wanted to implement a cover system, i would look at a pre-existing game, and implement it exactly, since that's kind of what players expect

i wouldn't make any decisions about which heuristics to use, i wouldn't choose the areas procedurally, and i wouldn't ad hoc anything. i already think that this cannot be done generally.

[half swan]

"And now let's set the goalposts way over here"

[undone coral]

it's like arguing over the best way to move fps characters. i don't have any opinions about that.

i would just do exactly what my audience expected from playing counterstrike.

[austere jewel]

Well, I'm absolutely certain there are games out there that automate this

[undone coral]

...maybe, but i don't think that would make sense

[austere jewel]

Well, you're wrong

[undone coral]

just because something exists doesn't mean it makes sense

i'm not sure why that is an unorthodox thing to say

[inland knoll]

I'm about 100% sure GTA uses it

[undone coral]

you know it's true

[undone coral]

i would go ahead and message them and ask

i have no idea honestly

[inland knoll]

You think they mapped out every single place where you can hide? Behind cars who were moving at runtime and stuff

[undone coral]

i think they might rely on a much simpler heuristic

but i don't know. i really don't know. iw ould ask them

anything is possible. they make beard growing simulator games inside their games. if you wanna do what GTA V does, by all means, do it

i don't think it's easy to make counterstrike in unity either. it's possible, it's just not practicable for an individual with a limited lifetime.* heuristic after heuristic is possible but it is not practicable

it can exist with flaws, and you can pick and choose which flaws you care about, but in which case, you know, try to do it as simply and as lazily as possible

back to square one: use a library.

i'm not sure you have strong opinions about which flaws you do or don't care about

my instinct is to look at exactly what xcom does - this is turn based right? - because it's written in python, it's right there for you to see not sure yet

because that's what players expect

i mean i can shoot from this blue rectangle, it seems like it's the best place for cover

i don't know what to say, it made a bad choice

but perhaps i'm not understanding what "shoot from here" means

i think the closer it looks like my illustration, the more simultaneous bad choices will appear

[lament salmon]

Its shoulder would be visible to the target from there, when taking cover.

[undone coral]

i am not sure what the game is, but if it were an FPS, you know, every player would choose blue rectangle. if it were a turn based tactical, i am not sure.

surely the most sensible option, if its shoulder was visible, would still be a square currently marked red near the blue rectangle, it's going to be sensitive to how much "moving" around a corner is valid

doesn't this successfully illustrate the problem with having a lot of heuristics?

how could it not? everyone would choose the blue rectangle for cover. an extremely simple heuristic (closest not ivsible location between me and the player) would also choose the blue rectangle, although it might be an inferior firing position, i'm not sure if in aggregate the simplest heuristic chooses worse firing positions on average

it's all very finnicky

[lament salmon]

It also tries to keep some distance to the target, that's why it most likely prefers the point on the right

[undone coral]

maybe the secnario is unrealistic. i think it's worth reproducing my original illustration exactly, which has self barely visible to target

rather than right there in the open

i don't know

[lament salmon]

Anyway, I'm not here to convince you that it works. It works really well for me, and I'm happy with it. If you want to only use libraries and ChatGPT, then you do you.

[undone coral]

i hope this is interesting

obviously i just want the game to be good

my goal isn't to find flaws personally, it's just to say, okay, it's possible to think about this like, a tiny bit, and come up with a pattern of terrain that is very challenging for heuristics

clearly that's a very challenging scenario

everyone looking at that illustration surely agrees the blue rectangle is not just a little superior* but a lot

[tall ferry]

🤷‍♂️ if talking about enemy AI and making a challenging game, it could be better that its unexpected where it will go.

As the player, if you can predict where it will be always then there is no challenge.

[austere jewel]

if I was a real person I definitely wouldn't choose to take cover right in front of my target in a way that caused me to run in a direct line towards them

[undone coral]

that caused me to run in a direct line towards them
looking at the actual screenshot, the difference between the exposed parts of the two paths would maybe not be very large. anyway, self is fully exposed in this scenario, and i think the most interesting scenario is still the one that was illustrated.
i don't know, maybe i'm not looking at the right thing, or maybe i'm looking at the wrong details, i am surprised that is a "direct line" in the screenshot. it's not my game.

this has been a very informative conversation.

[eternal musk]

    public static bool ReportIfNull<T>(T component, Object reporter) where T : Component
    {
        if(component == null)
        {
            Log($"{typeof(T).Name} could not be found!", reporter, LogType.Exception);
            return true;
        }

        return false;
    }```

I've only worked with generics a couple times, but I'm trying to learn them more. Writing this method that logs a message if a component was found null just for practice. When I print ```typeof(T).Name``` it just prints Component. Can I print the *exact* Component that was passed into the function?

[urban warren]

Can you show show how you are calling it?

[tall ferry]

.GetType(); should get what you want

[eternal musk]

Ah just realized my mistake lol I feel so stupid

        Component test = null;
        if (Logger.ReportIfNull(test, m_Processor)) return;```

This is how I was testing it

It prints exactly what I want, just was testing it incorrectly. My mistake!

[urban warren]

I was wondering if that was what you were doing haha

[eternal musk]

Lol I've been programming for too long today

[urban warren]

In this case, GetType() would not have worked as there is nothing to call it on. As component is null when you would want to get the type.

[tall ferry]

ah i am dumb, i mentally skipped over the whole null thing

[desert hare]

Hey, did anyone work with addressables before? I am looking to build a moddable game and, whatever I tested was working just fine until I made two new projects, one for building the mods and another one which is the actual project that loads created mods.
However in those last projects I am getting the same errors when attempting to load any mod from any of the three projects, nothing has changed internally so i am very lost to tracking down the issue, they're pointing out to a .bundle file that couldn't be imported or something like that

 /// <summary>
        /// Called when the addressable finishes loading.
        /// </summary>
        /// <param name="obj"></param>
        private void OnCompleted(AsyncOperationHandle<IResourceLocator> obj)
        {

            //find prefabs in the addressable with the tag specified in the first parameter
            IResourceLocator resourceLocator = obj.Result;
            resourceLocator.Locate(findByLabel.labelString, typeof(GameObject),
                out IList<IResourceLocation> locations);

            //if there are loactions in the adressable spawn them
            if (locations != null)
            {
                foreach (IResourceLocation resourceLocation in locations)
                {
                    GameObject resourceLocationData = (GameObject)resourceLocation.Data;
// NOTE: Apparently when debugging, the next line causes errors, but i am still not sure what to do.
                    AsyncOperationHandle<GameObject> prefab = Addressables.InstantiateAsync(resourceLocation);                        
                    prefab.Completed += OnModInstantiated;
                }
            }
        }

this is getting very overwhelming to work on ngl

[long ivy]

How were you testing before? Your paths are all very fishy, including spaces and bundles existing inside Assets folder

[desert hare]

Inside the editor and then building the project and testing with the built project as well

did the same for the other two projects and well, errors

they are all located in the same directories

i mean, the same root folder where i save all my unity projects

[untold moth]

There's also a chance that something else is opening/locking the file preventing it from being accessed🤔

[wooden cedar]

Some of them have spaces, and a lot of unity command line tools can't work with long file paths. You really should ensure those files are alphanumeric only, no spaces, and shorter (like 12 characters).

[desert hare]

I'll try bringing the built project on the desktop or something

[wooden cedar]

I'd go shorter, like C:\dev (just because of the long filenames). Desktop will still be a long path. It's also important to mention no non English characters in the path (which happens a lot when you use My Documents or Desktop with a non English username)

I wouldn't worry about it normally, but worth a test if getting odd permission errors

You can also manually browse to parts of that path and see if it is iniecting spaces where they shouldn't be (just copy paste the path into explorer)

[desert hare]

well, the most intuitive thing i can do is make the paths relative to the game's build

it may work for development but also hate to distribute files like that

[wooden cedar]

Yeah it's mostly a short term thing. Making sure the paths exist. If they do, they a shorter path. Just troubleshooting the steps of the problem first

That way we know if it is even the issue 🙂

[desert hare]

makes sense, i'll try

[wooden cedar]

Because if none of that helps, then no sense worrying about any of it

[desert hare]

Same errors, different directory

I feel like it may be something in the settings but I don't really understand them

I have modified the same fields between all projects

I'm noticing something weird tho, the backward and forward slashes

I know that could be an issue but it is weird that it worked on the first project whatsoever

the worst thing is that i can't load a simple stretched cube, while in the previous one i was able to load a complete model with textures

ok i have noticed something

it works if i dont move the bundles from their original build place

Well this is terrible, found this inside the "mod" json file, I'll have to do more research on this

[runic tendon]

I'm loading adressable scenes additively using Addressables.LoadSceneAsync(). For heavy scenes this is freezing the editor which doesn't make sense as the addressables are supposed to be asyncronous. Anyone have any idea why this could be the case?

None of the gameobjects in the addressable scenes are addressables atm.

[sudden stone]

anyone know how to work with .awb/.acb files

[urban warren]

I have two methods. Is there any way for me to have the first one call the second one to set the ref value (I could switch it to a out if needed)

public void Input<T>(string key, ref T value)
{
  Input(key, newValue => value = newValue);
}
public void Input<T>(string key, Action<T> setValue) { }

The above code doesn't compile of course cause you can't reference a ref parameter from a lambda.

[scenic forge]

No, lambdas could be stored and invoked at a later time, by which point the ref parameter may not exist anymore.

[tiny pewter]

using out should be possible?

the scope is limited and only be applied when you call it

[urban warren]

Nope, out results in the same compile error

[urban warren]

I know, but I am asking if there is a way to get the same basic behaviour 😛

[scenic forge]

If you must use lambda then no.

You could extract the logic inside the Action overload, and make both overloads call that

[urban warren]

The Action is stored and called at a later date :/

Oh well, I guess it isn't the end of the world. Just syntax sugar at the end of the day really 🤷‍♂️

[finite sundial]

I'm following the tutorial of xzippyzach, I'm using this code! https://github.com/xzippyzachx/UnityFirebaseDatabaseTutorial/blob/main/FirebaseDatabaseTutorialCompleteProject/Assets/Scripts/FirebaseManager.cs

I have my logout button from another scene instead of a UI panel
so the only thing I changed would be instead of using the UI manager, I'm accessing the scene

    {
        auth.SignOut();
        UIManager.instance.LoginScreen();
        ClearRegisterFeilds();
        ClearLoginFeilds();
    }

I have this

{
   auth.SignOut();

   // Wait for the sign-out operation to complete
   StartCoroutine(WaitForSignOutAndLoadScene());
}

private IEnumerator WaitForSignOutAndLoadScene()
{
   // Wait for a short time to ensure the sign-out operation completes
   yield return new WaitForSeconds(0.5f);

   // Now, load the scene
   SceneManager.LoadScene("FirebaseLogin", LoadSceneMode.Single);

   // Clear fields or perform other actions as needed
   ClearRegisterFields();
   ClearLoginFields();

   // Reinitialize Firebase in the FirebaseLogin scene
   ReinitializeFirebase();
}

[unique mountain]

anyone here has setup github actions for unity ?

[bleak citrus]

what is your question?

(this also isn't really a code question)

[unique mountain]

ok then tell where shall i ask this

[half swan]

#💻┃unity-talk

[random summit]

Guys do you know how can i start a client with an URI instead of an ip and a port on Netcode for GameObjects 1.6.0?

on Mirror i would do this

[urban warren]

#archived-networking Would probably be a better channel to ask in.

[random summit]

thanks

[vague flame]

quick question
if i have lets say a recursive function that causes sometimes a stack overflow, does running it in background in a coroutine or another thread can make it "slower"/reduce the risk/totaly avoid the risk ?

[dusty wigeon]

You can convert the recusion into a for loop in almost every case as far as I know.

[vague flame]

its a mix of a for and recursive

[dusty wigeon]

You can convert the recursive part in a while or foop loop

Share the code and someone might have an insight

[vague flame]

its not mine but a friend

[dusty wigeon]

Then share his code

[vague flame]

i will try to simplify, but so coroutine doesnt make it not stack overflow ?

a stack overflow isnt a result of to much operations in 1 frame ?

[sly grove]

No

It's a result of running out of stack memory

Due to too deep of a callstack

[vague flame]

so if you have no recursive there will not be to much callbacks ?

[sly grove]

Generally stack overflow is a sign that you had infinite recursion

It's a logical error

Similar to an infinite loop

It means you need to fix your logic, not switch to another style of iteration

[dusty wigeon]

I most say that I have seen stack overflow in the past that was legit, but it was in iterating a huge graph.

[vague flame]

ok ty guys

[sly grove]

Yeah that's super rare

But possible

[vague flame]

after reviewing his code we found a way more easy solution to avoid a big for loop and a recursive :)

but i note everything you said as future advices

[cyan gale]

can anyone help me I have this script were im trying to spawn objects on terrain based on texture
but i get an index out of bounds error when i hit play and nothing spawns

[half swan]

!code

[thorn flint]

Posting code

📃 Large Code Blocks

Use links to services like:
https://gdl.space/, https://paste.ofcode.org/, https://hatebin.com/, https://paste.myst.rs/, https://hastebin.com/

📃 Inline Code

Surround code with three backquotes. Not quotation marks.
To format as C#, add cs to the first line:
```cs
// Your code here
```
Add a comment with a line number if there is an error message.

[regal olive]

not sure what I did, but at least it compiles now,
does it do what you wanted? no clue what so ever...

public void Input<T>(string key, ref T value)
{
    T localValue = value;
    Input(key, (T newValue) => localValue = newValue);
    value = localValue;
}
public void Input<T>(string key, Action<T> setValue) { /* ... */ }```

[urban warren]

Thanks, but no, that doesn't sadly.

[regal olive]

I'll flip W arry you one day 🤌

[hoary yew]

Hello everyone! I'm currently trying to predict my targets position using rigidbody velocity, i already have a method for prediction regular projectiles and it works great however, I'm really struggling to find a way to predict where an enemy will be at the end of a parabola or how to make it so the parabola always ends at my predicted position. This is the method that the mortar projectiles use

private float progress;
    protected void MoveAlongArc(Vector3 startPosition, float arcHeight, Vector3 targetPosition, float stepScale)
    {
        progress = Mathf.Min(progress + Time.deltaTime * stepScale, 1.0f);
        // Calculate a parabolic value for vertical movement, creating an arc effect
        float parabola = 1.0f - 4.0f * (progress - 0.5f) * (progress - 0.5f);

        Vector3 nextPos = Vector3.Lerp(startPosition, targetPosition, progress);

        // Adjust the y-coordinate of the next position to create the parabolic arc
        nextPos.y += parabola * arcHeight;

        transform.LookAt(nextPos, transform.forward);
        transform.position = nextPos;

        if (progress == 1.0f)
        {
            //damage target
        }
    }```

And this is the method I use to predict positions regularly (In a separate script (: )

 protected Vector3 PredictTargetPosition(GameObject target, Rigidbody targetRigidbody, float projectileSpeed)
    {
        
        // Estimate the time of flight based on the distance and projectile speed
        float timeToImpact = Vector3.Distance(transform.position, target.transform.position) / projectileSpeed;

        // Predict the future position of the target using the current position and velocity
        Vector3 predictedPosition = target.transform.position + targetRigidbody.velocity * timeToImpact;

        return predictedPosition;
    }```

Any help would be greatly appreciated!

[dusty wigeon]

The implementation of: https://www.youtube.com/watch?v=aKd32I0uwAQ&t=228s

I also have the following if you interested

[hoary yew]

Hey thanks for this! This is insanely helpful for me. I actually have watched this video twice now but math is not my strong suit and I couldn't find the document explaining it like you have linked here!

[dusty wigeon]

I did the implementation, the actual document is https://docs.google.com/document/d/1TKhiXzLMHVjDPX3a3U0uMvaiW1jWQWUmYpICjIDeMSA/edit

[hoary yew]

Yes I saw this commented in your code (:

[honest furnace]

does some1 know a salution my player is a prefab and i have: public GameObject inventoryObject; where i need to put a ui on but i cant if i do i get a type mismatch does any1 know a salution?

[bleak citrus]

Sounds like you're trying to reference a scene object from a prefab

https://unity.huh.how/references/prefabs-referencing-components

[inland knoll]

Just a lil question, how did you make the panel which displays the info in the editor? Or is this in-game? Because I haven't been able find anything quite like that yet. Is this like a custom editor window?

[lament salmon]

Yeah it's a custom editor for a component that is just for testing different AI queries

[inland knoll]

Oh, alright!

[lament salmon]

That particular panel is drawn with Handles.Label with a custom GUIStyle IIRC

[left quiver]

https://hastebin.skyra.pw/dikixatiwu.csharp
https://hastebin.skyra.pw/ekuwafukuh.csharp

does anyone know why this loads the scenes i want but doesn't fully unload and reset the scene i just left? i want to be able to leave the game scene then load it again and for the scene to act as if it's the first time loading it.

[austere jewel]

What's not resetting?

[left quiver]

I'm not sure really. If I load the scene for the first time everything works fine but if I go to my main menu and return to the game my player can't move etc. I want it to load like it loads the first time around.

[austere jewel]

Any static variables will not reset, and anything that is DontDestroyOnLoad will also not reset.

[left quiver]

alr, will have a check

[grand dagger]

hey so what im trying to do is make an object stay in its absolute position, and what i mean by that is for it to stay in one place on the screen not the game window, here is some poorly drawn pictures in paint to help understand.

[silk trench]

You can do it but you have to write platform specific code by the looks of things

That being getting the position of the window

Then it's just a little bit of math from there

[tribal ridge]

I've got a snippet of code here that I can't figure out. It's in a while loop, but it breaks out at a yield return statement. I can't figure out why it does that. If anyone can explain how it could be breaking out of that while loop I would really appreciate the help ```
while (true)
{
destLoopIterations++;

            checkMatchesAgain = false;

            var destroyItemsListed = field.GetItems().Where(i => i.destroyNext).ToList();
            if (destroyItemsListed.Count > 0)
                yield return new WaitWhileDestroyPipeline(destroyItemsListed, new Delays());

            yield return new WaitWhileDestroying();
            yield return new WaitWhile(()=>StopFall);
            yield return new WaitWhileFall(); <- It's breaking out here !!!!!!!
            yield return new WaitWhileCollect();
            //yield return new WaitWhileFallSide();
            var combineCount = CombineManager.GetCombines(field);
            if ((combineCount.Count <= 0 || !combineCount.SelectMany(i => i.items).Any()) && !field.DestroyingItemsExist() && !field.GetEmptySquares().Any() &&
                !checkMatchesAgain)
            {
                break;
            }

            Debug.Log("got here 6");

            if (destLoopIterations > 10)
            {
                foreach (var combine in combineCount)
                {
                    if(combine.items.Any())
                        combine.items.NextRandom().NextType = combine.nextType;
                }

                combineCount.SelectMany(i => i.items).ToList().ForEach(i => i.destroyNext = true);
            }
        }```

it's breaking out at WaitWhileFall();

But it shouldn't be able to do that right?

The whaitwhilefall function is does not have anything that would break the loop that I can see

There's no errors saying something went wrong when i run it either

I was thinking maybe I didn't understand yield return statements enough or something

[north tundra]

Not sure if this is considered enough to be advanced, but can anyone help me figure out why my Perlin noise implementation is giving me strange results?

The Perlin implementation itself is here: https://hastebin.skyra.pw/kepajolujo.java

If anyone needs info on the way I implemented random numbers, I can provide that, but based on my tests it doesn't seem to be the issue

When I apply the noise to the Y value of verts on a plane, I get this:

[untold moth]

Did you try using unity perlin noise instead?

Just to see if the issue is with the noise itself or the way you map it to the vertices.

[tired fog]

I've been working with Jobs and Burst compiler for a while and I feel like i never know enough to optimize it, does anyone have any good tutorials /documentation to better know about the methods available and different techniques to apply?

[tiny pewter]

https://docs.unity3d.com/Packages/com.unity.burst@1.6/manual/docs/OptimizationGuidelines.html
havent heard that burst compiler allows you set branch predicator....

[fickle mango]

yielding shouldn't cause the loop to break. Maybe the gameObject gets deactivated and that's why the coroutine stops?

[north tundra]

It’s with the noise itself, cause built-in looks fine.

[sacred epoch]

!code

[thorn flint]

Posting code

📃 Large Code Blocks

Use links to services like:
https://gdl.space/, https://paste.ofcode.org/, https://hatebin.com/, https://paste.myst.rs/, https://hastebin.com/

📃 Inline Code

Surround code with three backquotes. Not quotation marks.
To format as C#, add cs to the first line:
```cs
// Your code here
```
Add a comment with a line number if there is an error message.

[sacred epoch]

Hello, i need urgent help, highly appreciate if some can help

https://gdl.space/uxaqipoxil.cs
https://gdl.space/niyijodewu.cs

these are 2 code for 2d chess game, my Rook chess piece, highlight the legal moves correctly but when a chess piece comes in the path it highlights that tile as well, how to stop highlighting tile if there a chess piece comes

[tiny pewter]

i think it is not advanced topic
BTW Chess is turn based, and i assume the highlight function is showing the valid movement (the tiles) of each chess, so why there can be chess moved while not during its turn....
Or i misunderstand, the function cant handle highlighting correctly....

Actually i suppose there is some abstract class eg ChessPiece and each type of chess piece eg rock, knight has an overload method to highlight different valid destination
rock:
started from its position(x,y) having four for loops from x-1 to 0, x+1 to 7, y-1 to 0, y+1 to 7, check if there is any other pieces (at that point the loop must breaks), highlight that tile only if the piece is opponent

[sacred epoch]

cheers dude, thanks it worked

[static vessel]

Hello. I have a question about Vivox Participant Tap.

I'm using a script to create participant tap on joining the channel but I want it to set their positions to relevant player objects in order to make 3D Audio. How can I achieve this?

(I tried to make it in positional channel but it doesnt work for me)

[untold moth]

Is there any reason you don't want to use the built-in noise?
I'm not very strong with the noise algorithm itself. There probably aren't many people that are.

[north tundra]

For this particular project, I plan on using different types of noise in different combinations to create landscapes, and so I figured it might make more sense (for standardization and cleanliness) if I implemented all of the used noise algorithms myself. I have considered giving in and using the built-in Unity noise, but the thing is that the Perlin algorithm isn’t actually too complex or difficult to understand. I understand it myself to an acceptable degree, it’s just that my implementation went wrong somewhere and I can’t identify where.

I used this page to drive most of my development: https://rtouti.github.io/graphics/perlin-noise-algorithm

It’s in JavaScript, so I had to adapt it to C#. Presumably the bug is occurring because it got lost in translation somewhere.

[untold moth]

The simple advice would be: don't reinvent the wheel.
Even if you need to use different noise types and/or blend between them, there's absolutely no need to implement this particular noise yourself.

Also, if you really understand the implementation, it shouldn't be too difficult to apply it in C#.

But generally, looking at the result screenshot, I'd just guess that you're supposed to sample points within some range and you're going over it.

Perlin noise always has the same value at the corners.
And it doesn't seem like the implementation accounts for sampling outside the specified range. Didn't look to deep into it.
My final verdict:
Use an existing implementation(the one provided by unity or one that is thoroughly documented and doesn't require translating to a different programming language).

[obsidian glade]

you may want to show the shuffle you used as well - it looks odd to me that you're creating an array of size 1024, only setting 0..255, then "shuffling" it. On face value that's shuffling a lot of 0's into the 0..512 range you're using.

it's important how the js code builds that array, shuffle then double, and it's not size 1024

[obsidian glade]

you've also flipped this:
float xf = xCoord - X, yf = Y - yCoord;
in the js code xf,yf are in [0,1], here you'll have yf in [-1,0] - it's not clear exactly what effect this will have, but it likely wont play nicely with the ease function that is supposed to be mapping [0,1] -> [0,1]

[brittle charm]

I have a bunch of game objects that are moving upwards on Y axis let's say. How can I make camera track the top one? It may change as objects move with different velocities at different moments. Simply going through list of transforms and getting one with max Y each frame seems inefficient

[tiny pewter]

Sad, but this is the only way, since maintaining a ordered structure requires at least nlogn time complexity

but getting maximum can be simply multi threaded

[compact ingot]

How many objects?

[brittle charm]

Actually just 16 so that way is actually okay, but I was wondering if there's a better method for larger amount of objects

[compact ingot]

Well, so long as you don’t sort them and just find the top one you stay in O of N and iteration without moving memory around and simple float value comparisons (finding a max/min) is so quick as to be irrelevant at your order of magnitude (10-100 objects)

[brittle charm]

Yeah you're right

when tf did i become optimization obsessed nerd trying to O(1) every algorithm

[harsh lion]

You could even run it every few frames (not evrty frame) and lerp if there is a change. At that point it becomes completely negligible