archived-code-advanced

[misty glade]

Storytime - in our first month of operations we noticed really aggressive port scanning - TCP connections on every port in sequence from a variety of IP addresses around the world (obvious VPN/botnet). About 24hrs after the connection was made on the actual game port, we'd see a variety of malformed packets come in with all kinds of funny data. Invalid headers, mostly, but the valid headers contained just a slew of different kinds of requests.. HELO and GET index.html and stuff like that. I'm pretty sure we still get malformed header connections just around the clock.

I was confused, thought maybe it was azure health checks or perimeter stuff, they (MS) confirmed it wasn't, so I change the game port (at great pain) and the same cycle happened again

[plush hare]

yeah port scans are gonna happen no matter who your provider is

[misty glade]

yeah we still get them

[plush hare]

But yeah this is going to get pretty complex. Someone else will probably have better suggestions

[misty glade]

I think doing the microservices and location servers will be best. I know that's how EVE online does it, and I think diablo3 too.. Like, you'd have a chat server handle all chat, a zone server that would handle players in some fixed zone in real time, maybe an achievment server, etc.. It just gets wonky when one of those services go down and you're still able to play but chat isn't working

[plush hare]

You'd probably just be better straight up disconnecting clients when your blades go down. Or perhaps give them a warning. Or wait until it's like 2am to shut it down

I know some providers charge based on resources rather than time

[misty glade]

Yeah but even an "empty" blade is still gonna be consuming resources, not a lot, but some.. It might not be a huge deal to have a bunch of blades running empty, so long as they're not being too aggressive on database hits.

The more I think around this, though, the more I think that there's just not going to be an easy/magic solution - the restful api approach is interesting but probably creates more problems than it solves, and certainly each request is going to come with a lot of baggage. I think I'm just gonna need to code my way out of this and build the blade/lobby/microservice approach

Oh, the other "pro" for web/restful is that people with shit connections can have a slightly better experience - versus what happens now which is.. kinda crappy. We have to set a keepalive time (both for clients and server) so we know when one or the other disconnects, and it's a guessing game.. 15 second pings that don't come back within 60 is our current amount, preventing people from getting disconnected in tunnels (they just "catch up" when they come back to service) but not playing as a zombie and losing too much progress they never actually had because the connection went dead

[plush hare]

If that azure tunnel shit gives you too many problems, you should just switch providers

I don't know if azure is designed for realtime stuff like this. I think they're more geared towards the web market

https://us.ovhcloud.com/

ovh has like a 7 tbps backbone and automatically reroutes ddos to scrubbing centers. free of charge. No tunneling bs required

[green pagoda]

ok thanks

[gloomy mica]

is there any quick ways to determine what these byte[] is?

screenshot is from memory profiler

[gloomy mica]

solved, I can track it down from the reference object

[worthy vapor]

hello, i have multiple meshes of countries, and i wanted to generate a random point on the country, i have tried a lot of things with Chat GPT but every time, the repartition is uneven, clustered, or always on the border, maybe it is due to the special shape of my mesh but i'm starting to want to give up ...
(attched, two screens of the mesh used for Spain)

[tiny pewter]

there is a general algorithm for pick random point of any polygon

you need to trangulate your polygon first then select one triangle based on their area/ total area and pick one point in the triangle you have selected

maintain a ordered list of triangle based on their area can help you pick the triangle in O(log n) time

[tall ferry]

Is this not just weighted randomness based on area of triangles?

[tiny pewter]

yes exactly

[worthy vapor]

i also want to ingore triangles that are not facing upwards

[tiny pewter]

you can construct two vectors by three vertice of triangle and cross them
i dont know if there is built in api can help you do this

[worthy vapor]

i'll do some research, thank you for your aswerls

[tall ferry]

I believe u can use plane.normal or just calculate the cross product still

[vestal condor]

Quick C# question - let's say I want to have a delegate that is "public" in terms of adding to it but can only be privately invoked
Is there some syntatic sugar for this or do I just need to add boilerplate code?

public delegate void EnemyEvent(Enemy enemy);
private EnemyEvent OnEnemySpawn;

Is there a good way to add a method to OnEnemySpawn without boilerplate code and achieving the same outcome?

[dusty wigeon]

Use System.Action<Enemy> instead of EnemyEvent.

[vestal condor]

Hmm, maybe I was unclear

class Spawner 
{
  public delegate void EnemyEvent(Enemy enemy);
  private EnemyEvent OnEnemySpawn;
}

class Wave
{
  ...
}

I would like the Wave class to be able to add a method to OnEnemySpawn in Spawner but not be able to call OnEnemySpawn(...)
sort of like a write only rather than a read only

[dusty wigeon]

Add event before private EnemyEvent OnEnemySpawn; => private event EnemyEvent OnEnemySpawn; (And make it public)

Alternatively, you can implement manually the Observer pattern and control visibility base on your preference.

[vestal condor]

Aha, thanks I was looking for event.

[plucky laurel]

also look in Action<T> Func<T> Predicate<T>

will save you some typing in 90% of cases

public event Action<Enemy> onEnemySpawn;

[wise tartan]

So question, is it good practice to use async in Unity? From what I've read the problem is that Unity is designed to run on a single thread, and async operations can get a bit messy if you're dealing with Unity features specifically.

From what I understand Coroutines are supposed to replace that, but I feel like they're a little finnicky?

[plucky laurel]

async isnt threaded

yes you can use async, there are several implementations of it

[austere jewel]

2023 adds Awaitable https://docs.unity3d.com/2023.1/Documentation/ScriptReference/Awaitable.html

[plucky laurel]

Coroutines/ their better alternative MEC on asset store/ .net async / UniTask on github

[plucky laurel]

UniTask replacement?

[austere jewel]

Practically, yes

[wise tartan]

What does it add to working with async operations? I haven't studied async too much, so I'm still a little lost

Awaitable, that is

Right now you'd have to return Task to assess the state of an async operation, right? Is it a replacement for that, with more Unity integration?

[plucky laurel]

it returns an object that unity understands, ie wait for a frame, wait til some unity state is reached etc

so the state machine higher can suspend/resume the method at appropriate time, same as coroutines yield objects, as i understand it

with UniTask it injects itself into the playerloop, at specific points before/after update etc, then catches UniTask struct and makes decisions on when to continue the method

also its a struct, Awaitable is an object, which is weird

[austere jewel]

Awaitable Coroutines are async/await compatible types specially tailored for running in Unity. (they can be awaited, and they can be used as return type of an async method)
 Compared to System.Threading.Tasks it is much less smart and takes shortcuts to improve performance based on Unity specific assumption.
 Main differences compared to .net tasks:

• Can be awaited only once: a given Awaitable object can't be awaited by multiple async functions. This is considered undefined behaviour (considering it undefined behaviour allows for optimizations non-achievable otherwise)
• Never capture an ExecutionContext: for security purposes, .net Tasks capture execution contexts on awaiting to be able to handle things like propagating impersonnation contexts accross async calls. We don't want to support that and can eliminate associated costs

• Never capture Synchronization context:
  Coroutines continuations are run synchronously from the code raising the completion. In most cases, it will be from the Unity main frame (eg. for Delayed Call Coroutines and existing AsyncOperations)
  Methods returning awaitable coroutines with completion being raised in a background thread will need to mention it explicitly in their documentation
• Awaiter.GetResults won't block until completion. Calling GetResults() before the operation is completed is an undefined behaviour
• Are pooled object:
  Those are reference types, so that they can be referenced accross different stacks, efficiently copied etc, but to avoid too many allocations, they are pooled. ObjectPool has been enhanced a bit to account to avoid Stack<T> bounds checks with common get/release patterns produced by async state machines
• Can be implemented in either managed or native code:
  There is Scripting::Awaitables::Awaitable base class providing interop with this managed coroutine type, that can be extended so that Native asynchronous code can be exposed as an await-compatible type in user code.
  See AsyncOperationCoroutine implementation as an example

(that is a comment above Awaitable in the Unity source)

[wise tartan]

This is some dense stuff... thank you

https://docs.unity3d.com/2023.2/Documentation/Manual/AwaitSupport.html

{
    var apiResult = await SomeMethodReturningATask(); // or any await-compatible type
    JsonConvert.DeserializeObject<List<Achievement>>(apiResult);
}

async Awaitable ShowAchievementsView()
{
    ShowLoadingOverlay();
    var achievements = await GetAchievementsAsync();
    HideLoadingOverlay();
    ShowAchivementsList(achievements);
}```

i struggle to understand the `GetAchievementsAsync()` method from the documentation, it doesnt seem to have a `return`? am i just misunderstanding something about async?

[plucky laurel]

there is await in it

[austere jewel]

Ah, I was looking for those docs to link you, but found the code comment instead 😄

[wise tartan]

How is the apiResult returned to ShowAchievementsView through the await?

[austere jewel]

Because it would return Awaitable<string> or similar

when you await the Awaitable it practically evaluates and unpacks it, so you get the value it returned

[wise tartan]

but did GetAchievementsAsync return anything?

[plucky laurel]

unless i dont understand something, that second line is doc fail

[wise tartan]

ok good cause i thought i was going crazy

[wise tartan]

So if i understand correctly
await Awaitable.BackgroundThreadAsync();
or
await Awaitable.MainThreadAsync();
just basically toggle any following code to run on the respective thread?

[bleak citrus]

Pick random points in the bounding box until it lands on the mesh.

this performs poorly for pathological cases, like a country shaped like a very very thin donut

[novel plinth]

BackgroundThreadAsync refers to the threadPool it's the equivalent of Task.Run, kinda a bit weird Unity named it like that. while the latter is to continue/queue your code execution back to unity's main thread

[wise tartan]

Interesting, thanks

[worthy vapor]

i have managed to filter vertical faces and to pick a random triangle in the horizontal ones, my goal is now to pick a random point inside the picked triangle

[bleak citrus]

I have code that does that somewhere

it was used to sample a random point on a navmesh

https://answers.unity.com/questions/857827/pick-random-point-on-navmesh.html

note that this is not a uniform distribution, since the tris could have different sizes

Vector3 GetRandomLocation()
    {
        NavMeshTriangulation navMeshData = NavMesh.CalculateTriangulation();

        // Pick the first indice of a random triangle in the nav mesh
        int t = Random.Range(0, navMeshData.indices.Length-3);
        
        // Select a random point on it
        Vector3 point = Vector3.Lerp(navMeshData.vertices[navMeshData.indices[t]], navMeshData.vertices[navMeshData.indices[t+1]], Random.value);
        Vector3.Lerp(point, navMeshData.vertices[navMeshData.indices[t+2]], Random.value);

        return point;
    }

you know, i just realized that this might pick one vertex from one triangle and two vertices from another triangle

not sure

i am also not confident that this even produces a uniform distribution across the single triangle...

[last locust]

new rule no more refunds

Everythings free

[upbeat path]

Spoken like someone who has never put any effort into publishing anything

[last locust]

Anything advanced

[unborn relic]

hello, i am having trouble understanding something. if I fire some Jobs that take let's say 2 seconds to finish each, if I wait for them to complete with jobhandle.Complete(), the main thread is still going to be stuck for 2 seconds right? how do I avoid this? do I need to have a croroutine that checks if the jobhandle is complete, or yield return new WaitForEndOfFrame() if not? is there a better way?

[real blaze]

you want to asynchronously wait for jobs to finish? if so, yeah you can just wait for them in a while loop. either async or coroutine.

if it's possible however, you can move your whole method to another thread, and just stop the thread untill all jobs are done using that Complete(). this will have less overhead, if you need to do a lot of work in the jobs. which it seems you are

[unborn relic]

I am trying to Physics.BakeMesh my 65k verts meshes at runtime (so I can't use the example provided in the docs, because my meshes are generated at runtime), which don't exactly take 1 sec, but they do take about 200ms on my machine. I am not sure I understood the last solution you proposed, would you mind elaborating a bit? Thanks for the help!

[real blaze]

the last solution was to create another Thread to do what you're doing. and in that thread, start all the jobs, and then wait for them to Complete() right there. but now that you mentioned you're doing Physics.BakeMesh, I don't think you can do that in a separate thread

[sage radish]

You can, it's one of the methods they made thread safe.

[dawn zephyr]

hey guys would a zenject scenecontext with an installer with dontdestroyonload work in order to inject things in another scene?

That's my installer, maybe im doing something wrong, just picked up zenject

public class Installer : MonoInstaller
{
    [SerializeField] private DatabaseManager _databaseManager;
    public override void InstallBindings()
    {
        Container.Bind<IDatabase>().To<DatabaseManager>().FromInstance(_databaseManager);
    }
    private void Awake()
    {
        DontDestroyOnLoad(gameObject);
    }
}

And the class

public abstract class DatabaseDependant : MonoBehaviour
{
    [Inject] protected IDatabase _database;
    protected abstract void LoadData();
    protected abstract void SaveData();
    protected virtual void Start() => LoadData();
    private void OnApplicationPause(bool pause)
    {
        if (!pause)
        {
            return;
        }
        SaveData();
    }
    #region DEBUG
    private void OnApplicationQuit()
    {
        SaveData();
    }
    #endregion
}

my database doesnt get injected for some reason

[sage radish]

A SceneContext is only meant to contain bindings for one scene (hence the name). You can use a ProjectContext for project wide bindings. There are also many different approaches for dealing with multiple scenes, all described in the readme.

[vestal condor]

I'm having this issue where reloading script assemblies takes really long 30-60s after changing any bit of code at all
any idea what could be causing this?

[dull kestrel]

That's unity reloading the domain. You can disable it.. BUT if you're using any static variables they will not be reset.

[novel plinth]

nah, dont disable it unless you know what you're doing

[novel plinth]

Sometimes it makes sense on a pretty huge project..
Make sure you tell Unity only to compile parts that are changed. That said, read up on asmdef and how to use it in your project

[vestal condor]

alright I’ll take a look

It didnt take this long back then though, not that I recall at least

[dull kestrel]

If you have any add-ons like fishnet they do things during domain reload too, which made mine take ~10 minutes (but I had also not fully set it up).

[vestal condor]

hm what’s fishnet?

i do have this dependency hunter thing but i don’t think it runs unless i tell it to

maybe i should investigate that

[dull kestrel]

It was just an example.

[dusty wigeon]

Pretty sure you can profile also. On our project (Not divided by assembly (-_-)), we have 10s. Sometimes it goes in the 30s and a good restart (Computer) works.

[bleak citrus]

this has absolutely nothing to do with unity

[austere jewel]

This is a Unity server, so no, if it's not Unity-related it's off-topic

[dawn zephyr]

so i made a project context with an installer on it

public class Installer : MonoInstaller
{
    [SerializeField] private DatabaseManager _databaseManager;
    public override void InstallBindings()
    {
        Container.Bind<IDatabase>().To<DatabaseManager>().FromInstance(_databaseManager).AsSingle();
    }
}

but still when I load the next scene everything that has the

[Inject] IDatabase

doesnt get injected

[dawn zephyr]

Man it's killing me it's just one line of code I can't wrap my head around why it isnt working

[flat mountain]

Is there a way to get the currently active material whilst rendering Editor IMGUI?

Trying to figure out which material is being used so that I can use it in my line rendering and retain the clip (which Internal-Color, as I learnt the hard way, doesn't do)

[timber flame]

I have big problems to design an architecture for my data in a city building game (simulation survival game) like oxygen not included. In my game, there is different buildings, plants, etc.
I can inherit them from a base class BuildingDefinition and then specific child type ResourceBuildingDefinition, PlantDefinition, etc.
The problem is
1- It is possible to add another hierarchy.
2- Sometimes, it is hard to add some fields to base class or put them in individual child classes. For example Hitpoint. Maybe, some buildings have hitpoint or not (cannot damage). I know I can add a boolean field canHit to base class BuildingDefinition but...
the design of this game is really incomplete and I do not know which architecture is the best in this situation, component based or inheritance. I know component based approach is more flexible and I generally prefer it even for data (scriptable object definitions)

Building Data
Damageable component
Construction component --> build time, required resources, etc.
Specific components for each building
BaseComponent --> name, icon, etc.

[dawn zephyr]

maybe a damageable interface?

[timber flame]

First you prefer component or inheritance. Do you have any experience about it to design your data?

They are data attached to that scriptableobject not mono.

public class BuildingDef:ScriptableObject{
   public IComponnet[] Components;
}

[dawn zephyr]

try the serializereference attribute

[timber flame]

I know it

My question is not how I can implement it

It is about architecture if somebody has any experience to structure their data using component list or inheritance with different children

[plucky laurel]

@timber flame you have odin?

with odin you can directly serialize your components in the scriptable, with polymorphism and all

at runtime you clone it from the def

i dont fully understand what you already have, but it looks like you decided to use composition for defs

so list of defs that are responsible for various aspects of the thing

i dont think there is a game that would not rigidify at some point, where all the excess modularity would only hinder your design time

composition sounds great if you have a good way to edit all of that, like [InlineEditor], and a custom window with all your data split into tabs, etc

if its done well you wont have to spend time in the project panel

[timber flame]

Yes, dude.

So, you like me prefer to organize your data as component list?

[plucky laurel]

i approach each case individually, where it makes sense to modularize it, i do, if the cost of maintaining bunch of modules is too high, i will not

when im weighing the costs i take into account things like - ease for designers, amount of additional objects you have to maintain (drawers, runtime representations),
amount of cases that would result from modularity, i.e. if its something like a script for a spell, is it reasonable to construct it from basic atomic blocks, similar to visual scripting?
maybe it is, but this will result in an explosion of maintenance cost

so i tend to lean towards plain pods as much as possible, you can migrate data later if needed, but from the get go creating yourself a bunch of maintenance which benefit cant be easily predicted in the long run, is akin to premature optimization, overengineering

in 90% of cases it never goes further than a pod

[timber flame]

because in our game that does not have any stable design now. In my mind, component based is more flexible.

No, I do not want to break them and create them from atomic blocks

Probably, the number of components do not exceed 10.

[plucky laurel]

look at it like this, right now you have say a building that is modular, then you adhere to that paradigm and eventually you will have everything in the project modular, thousands classes to maintain, architecture to handle it all, draw/edit it all, you have to remember which components you have for each thing, what goes where, what to use when etc

just envision the end state of the project if it was this way, and if you are ok with it, go for it

[timber flame]

but at this moment, I do not know for example every building has damage. Every building has ... and so one to use inheritance with base data

[plucky laurel]

yes but every building damage can be described as bool isDamageable

[timber flame]

Yes, I can define all data for all of them with default values if they do not have them

[timber flame]

Yes, something like it but messy

public class Buidling: ScriptableObject
{
   public bool IsDamageable;
   public int HitPoint;
   public Resource[] Inputs;
   public Resource[] Outputs;
}
public class Plant:Building{
   // specific properties for plants (grow rate, conditions, etc.)
}
public class Military:Building{
  // attack point, rate, etc.
}

Inputs and outputs can be resource or something different completely.

[plucky laurel]

i like this approach as a base one, you have the power of the type system, you can infer what kind of entity you spawn by the type of the def itself

theres a lot of things you can also add to the class, that are specific for handling that type of entity

you use type system to bind a runtime entity to the def type, it is very clear and simple to use

[rotund horizon]

DOTS Best practice

[timber flame]

Yes, the problem is which data should be put in base. If it can be generalized for all types or not.

[plucky laurel]

you can figure it out as you develop, data is serialized by property name, so it doesnt matter if you move some field in the base class, it will be preserved

even when you remove the property for the type it still hangs around in the .asset

[jagged fiber]

Hey guys! A small shader question.

I have a vertex function where I set UV.
That UV goes to the fragment function.

The question:
I'm reading a book where it says "vertex function data goes straight into the fragment function".

Is it really true? Why, then, in the fragment function I get hundreds of different UV values for each pixel, while having just 3 vertices (meaning I set UV data just 3 times in the vertex function).

Some interpolation happens in-between to get per-pixel UV and probably some other calculations.
Where can I read about what happens in-between vertex and fragment functions?

[regal lava]

#archived-shaders

Probably get some better answers there

[drifting vessel]

Does anyone here have experience with asynchrony? Specifically, trying to make REST API calls, so I don't want it locking the main thread.

[urban trellis]

UnityWebRequest is asynchronous

[bleak citrus]

where did this quote come from?

well, two staff members said it's under the same license as the engine

so there's your answer

[drifting vessel]

Yeah, I was hoping to use HttpWebRequest, since it's agnostic

[upbeat path]

did you not see '/Il2CppOutputProject' this includes the output from IL2CPP not IL2CPP itself

pretty sure that's just support code for the runtime

build your own il2cpp project and look at the code generated

[untold basalt]

Hey folks! I’m trying to make a Camera move using DOTS but it’s not working, and I’d like to understand why. I have a MovableComponent (just a tag) and a pretty small system:

{
    protected override void OnUpdate()
    {
        Entities.ForEach((ref LocalTransform transform, in MovableComponent _) =>
        {
            transform.Position += new float3(5f, 0, 0) * SystemAPI.Time.DeltaTime;
        }).Run();
    }
}

I attached the MovableComponent to the main camera and to a cube. The cube moves, but the camera doesn’t, even if I can see the camera localTransform being updated in the inspector, which makes me think I’m missing something. I also tried making the camera a child of an empty game object and adding the tag to the GO instead, but it doesn’t work either.

Of course, I could just use Camera.main.transform.Translate, but I want to understand why it's not working. I searched everywhere but I couldn’t find an answer. Any ideas?

[sand wraith]

In unity some in built components like "Transform","Rigidbody" etc can be added only once to an object.
I am going through components of a gameobject runtime and if the components can be added only once, i should ignore that..
for custom components i can check if it has [DisallowMultipleComponent] attribute and ignore. But for built in components, it not possible..
Is there a way to detect it for built in components?

anyone?

[spiral swift]

I have a classification model that takes 3 values

not sure how to get the output tho

        InputTensor[0] = Kills;
        InputTensor[1] = Hits;
        InputTensor[2] = Distance;
        Worker.Execute(InputTensor);

        Tensor outputTensor = Worker.PeekOutput();
        return int.Parse(outputTensor.ToString());```

this is what I did for the output, but it gave me an error

[sand wraith]

hi I am still not able to find any solution. any help would be great..

[long ivy]

What's wrong with the obvious solution of keeping the exceptions in a list and checking that?

[sand wraith]

actually do we have an official list of components that can be added only once? couldn't find one in docuumentation. This was my first approach to the problem

[long ivy]

probably not, but it wouldn't be hard to write a little editor script that tries to do just that and have it generate that list for you

[sand wraith]

ok

a warning dialogue box is popping up when i try to add one of those exceptions.. so unity is detecting it somehow..

[timber flame]

public class Building: ScriptableObject{
  //name, id, icon, description
  public IRequirement[] Requirements;  // preconditions and requirements to be able to construct it
  public IEffector[] Effectors;
  
}

It is neat and really flexible but probably over generalized.
Sometimes, one specific effector is just for one specific type of building. Also, about requirement.
I mean it is hard to use some of them more than one time.
It reduces code duplication (e.g. implement an effector or requirement just one time)

The alternative way is to define properties one by one in each building type. It can cause duplication.

public class Building: ScriptableObject
{
}
public class PlantDefinition: Building
{
   public ElementData RequiredWater; // required
   public ElementData RequiredSoil; // required
   public SeedData Seed;  // required

   public bool CanFertilize;
   public ElementData Fertilizer;  //optional requirement
   public int BuilderLevel;  //requirement
   public Season PlantSeason;    // requirement

   public ProductData Product;  // product id and amount  // effector
   //-------------------------
   //????
   public Season HarvestSeason;
   public float GrowRate; // grow rate per cycle
   //...
}

[sleek marsh]

u mispelled ProductData

[severe topaz]

over generalized is what you want for interfaces anyway, the object itself handles what it should do if it receives appropriate command.
use dictionary with enum as key to check if the command is appropriate. or slap on another interface.

[obtuse arch]

everything has its pros and cons, Every-time you will face a scenario where a practice or principle is regarded as decent one but will not be applicable to your design. Forcefully applying it the greatest mistake one can make.
First and foremost finalize on design rather than hopping from one req to other, Then architect your codebase according to the expected axis of changes keeping in mind the domains where frequent changes are required and what makes it easy to reduce production time and helps you ship quickly.
Each decision to make a code/module accessible for design will eat up tiny part performance somewhere or will introduce an overhead which is unavoidable. In short there is no fixed design. Build Test Refactor is the way to go. Fixating on a principle/design/architecture is the worst thing you can do at early stage of development similarly as optimizing prematurely.

[timber flame]

I have decompiled codes of oxygen not included. As I guessed, they have defined building definitions with all fields inside !

[severe topaz]

@timber flame that doesn't mean it's the best solution for your problem though. and you don't know if its' devs ever felt like they majorly fucked up with tech debt.

i'd be more interested in how games like path of exile handle their stuff because it doesn't look like they locked themselves in with any gear at all

[timber flame]

Exactly. To me if I am a designer, it is really hard to change and update many arguments while many of them are redundant and should be set to the default value

I think in cult of the lamb, they have used same approach, define all fields in one class.

like defining them in a table (db)

[severe topaz]

if we take path of exile as an example, every item that is not a currency (e.g. armor, weapons and flasks) is essentially treated the same. almost none of them behave the same, except that they all have similar rules for how currency is used on them and how crafting works on them.
that's why IMO it's a good example of how modular stuff should work.
if you already know exactly what you want and it doesn't need this much flexibility, and you're not planning to expand, then do whatever you want tbh 🤷‍♂️

even then they probably curse at themselves for the spaghetti they made daily

[regal lava]

im convinced POE is all spaghetti considering how that game performs

but that's probably because the bloat of affixes and statistical stuff that's been being crammed into their calculations over the years

[severe topaz]

@regal lava yeah, too much pointless crap that is barely noticeable anyway. the only way to gain performance is to kill faster so that there are less calculations to perform.

but the modularity is 👌

[cobalt hemlock]

Hey guys, i am loading some data from a database. There is a line in the code that assigns a boolean value to isOn for a toggle (Please see attached).

When happens, it triggers a ArgumentException: Object of type 'UnityEngine.Object' cannot be converted to type 'UnityEngine.GameObject'.... exception, which when I look at the step through has nothing to do with my code after the boolean is assigned.

If i comment it out, all works fine.

Please see the attached images. If you need any more information let me know.

[dusty wigeon]

Your error might be something else. It is just happening when you set a boolean because it triggers something that make the error apparent.

[upbeat path]

at a guess there is a custom inspector/editor attached to requestData_SettingsParameters

[cobalt hemlock]

This is what I have attached to the requestData_SettingsParameters (see attached).

[dusty wigeon]

Readying a bit, it might be cause by UnityEvents that are not properly set.

What the Event on the Toggle looks like ?

[cobalt hemlock]

Ohhh, it it could be something due to the scene that is being loaded, and after loading, the data is then applied to the scene. so there might be some misconfiguration in the scene that is triggered with the assignment.

[dusty wigeon]

Maybe, I would try to remove the Events and see if it works.

[sly grove]

It's something in your toggle listener

Actually seems like an object is trying to be passed to the toggle listener as a parameter that is supposed to be a GameObject but is not

I'd scour your toggle inspectors

Basically a wrong object is set as the parameter or something of that nature

[cobalt hemlock]

This is how the Toggle is setup.

[dusty wigeon]

The None could be the cause.

[cobalt hemlock]

I think it could be the missing game object on the dimension trigger.

[sly grove]

Could be yes

BTW if you want to toggle your toggle in code without triggering the listeners, it has a SetWithoutNotify function you can call instead

[cobalt hemlock]

Yup, i completly missed that...I checked the logo one, but not the dimension one. Apologies. Thanks for the help, I do i appreciate it.

[cobalt hemlock]

Ahh, I noticed that today, but It something I could definitely use. Thanks for the hint.

[cobalt hemlock]

Thanks for the help trouble shooting, I should have done a bit more checking before asking.

[worthy lodge]

Got some issues with GetVertexData if someone has experience with it

https://forum.unity.com/threads/nativearray-deallocated-after-a-single-use.1450384/

[raw lily]

Is there a difference between changing the localScale of a transform, that contains a sprite renderer OR changing the .size of a sprite renderer?

[sly grove]

Didn't people already answer this for you in #📲┃ui-ux

Read the docs for .size

it's not what you wan't in 99% of circumstances

the difference, obviously, is that one changes the object's scale and the other doesn't

meaning its children will also be affected etc

[raw lily]

Ohh, I assumed no one answered since I just read the guy saying it wasn't the right section. Will do, thanks! 🙂

[sly grove]

Mao answered you

Also Carwash's answer was also an answer of sorts

[raw lily]

Yeah, it tends to be obscure whether my UI is actual UI or 3D in my project, I don't think I should use an Image there but will see if there are any upsides I might I've missed the first time I decided on SpriteRenderer.

[sly grove]

SR is not a UI element

if you are making UI, don't use a SpriteRenderer

[solid lodge]

public class BuildingSystem : MonoBehaviour
{
    public GameObject[] placeableObjects;
    public float placementHeight = 0f; // The desired Y position for object placement
    public float movementMultiplier = 1f; // Multiplier for cursor movement distance

    private void Update()
    {
        HandleInput();
    }

    private void HandleInput()
    {
        for (int i = 0; i < placeableObjects.Length; i++)
        {
            if (Input.GetKeyDown((i + 1).ToString()))
            {
                PlaceObject(i);
                break;
            }
        }
    }

    private void PlaceObject(int index)
    {
        GameObject objectToPlace = placeableObjects[index];

        Vector3 mousePosition = Input.mousePosition;
        mousePosition.z = -Camera.main.transform.position.z;
        Vector3 worldPosition = Camera.main.ScreenToWorldPoint(mousePosition);
        worldPosition.y = placementHeight; // Set the Y-position to the desired placement height

        Vector3 objectPosition = new Vector3(
            worldPosition.x * movementMultiplier, // Adjust the X-axis position
            worldPosition.y,
            -worldPosition.z * movementMultiplier // Invert the Z-axis position
        );

        Instantiate(objectToPlace, objectPosition, Quaternion.identity);
    }
}

So what the code should do is place a object depending what key I press and it kinda does it

1. when placing on the x axis it goes the other way the mouse is
2. the objects dont place exactly where the mouse it instead it bundles up together kinda (Tried to fix this using movement multiplayer)

what are the steps to fixing this?

photo of what i mean bundling up

[prime kite]

i have a 2d world made of chunks

each chunk contains 10x10 square objects (every time i move in a direction i instantiate 3 chunks and destroy 3 chunks)

instantiate is really slow

do you guys know what the besy way to optimize this is?

[tiny pewter]

object pool
but this cost lots of ram of your machine since it has to hold the object cant be free()ed in heap

[dusty wigeon]

It does not cause that much of strain on memory because GameObject and component are relatively small. (The asset (Texture, Audio, Meshes) are most of the time what is the cost)

In theory, the user could even reuse the same object without needing additional objects as he destroy others.

[vestal condor]

running into this issue where a collider is being shown as existing in editor but actually doesn't exist in code

i was working with a prefab variant that previously had its collider removed
i updated it to be no longer removed, and indeed, instantiating it shows that it is no longer removed

but it's literally acting like it's not there - i can't even collide with it despite all the layers being correct

I've run into a similar issue before where i was colliding with a non-existing collider after removing it (and verifying that it is indeed not there in editor)

anyone has any clue why this happens? deleting all my streaming assets and restarting editor doesn't seem to help

its like the editor and game are desynced

[radiant oasis]

hey who needs help

[radiant oasis]

@vestal condor

i can help

[vestal condor]

nvm restarting machine seems to have worked

just a super weird bug with unity editor

[timber flame]

To show stats easily, do you have something like below for each definitions?

  // AttributeType contains all attributes (enum or a collection of strings)

  // A definition (character, building, etc.)
  public override Attribute[] GetAttributes(){
     return new Attribute[]
        {
          new Attribute(AttributeType.Power,power)
          //...
        };
  }

or instead of define properties as primitive types with values, define them as Attribute type.

public class BuildingA: ScriptableObject{
    public int HitPoint;
    //...
}

public class BuildingA: ScriptableObject{
    public Attribute HitPoint;
    //...
}
[Serializable]
public class Attribute{
    public AttributeType Type; // or directly AttributeDefinition
    public int Value;
}

[Serializable]
public class AttributeDefinition: ScriptableObject{
    public AttributeType Type;
    public string Name;
    public Sprite Icon;
}

[tender ore]

Unity randomly micro freezes, Profiler shows editor loop 98% at this moment, what can I do? Tried different 2022+ versions

[sly grove]

The editor has random overhead

it won't be there in the build

most of the time, it's not worth worrying about

[remote drift]

Is there any publicly (or internally) accessible method to convert instance ID to Object?

Specifically I want that for Material

[flint sage]

https://docs.unity3d.com/ScriptReference/EditorUtility.InstanceIDToObject.html

[remote drift]

what about runtime?

[flint sage]

Nope

[remote drift]

😩

[dusty wigeon]

You should not use instance ID for any feature. It is suppose to be use to debug.

https://docs.unity3d.com/ScriptReference/Object.GetInstanceID.html

The ID changes between player runtime and Editor sessions. As such, the ID is not reliable for performing actions between the Editor and runtime sessions, for example, loading an object state from a save file.

[novel plinth]

There's ExposedReference, then you can implement your own persistent id

https://docs.unity3d.com/ScriptReference/ExposedReference_1.html

never really used it, but the Playables api uses that like everywhere

[remote drift]

I need to use it to pass Objects in unmanaged way (burst)

[novel plinth]

oh.. then idk about that 😄

never really used it in such manner

[dusty wigeon]

I have no experience with burst, however you should see how others are doing it.

[remote drift]

others are Unity and they just have private extern methods 😅

allthough

that ExposedReference might be it (or not)

[novel plinth]

it should, if you just want to look for the reference to the object directly

the rest is up to your impl

[dusty wigeon]

Also, can you just use directly a reference ?

[remote drift]

I can't pass it, as direct reference is managed and can't be compiled with burst at all

in the end I will use it obviously

but the whole setup for usage I want to be burst compiled

(decals rendering)

[dusty wigeon]

So, you cannot pass a pointer to an object ?

🤔

[tiny pewter]

Need Fixed until the job is end

[remote drift]

anything that is GC managed is not burst compatible

[dusty wigeon]

You can fixed object can you not ? https://learn.microsoft.com/en-us/dotnet/csharp/language-reference/statements/fixed

[remote drift]

no

[dusty wigeon]

Like I said, I never done burst, so I do not know. But I feel like using GetInstanceID is definitely the worst way.

[remote drift]

nah, it is the way to go for a lot of API

I was surprised to see that Unity NavMesh api accepts instance ID

which allowed me to set the field via ptr

I really wish there was a way to make draw calls natively via instance ids

or maybe some wrapper for it

which has set/get property with actual reference, while storing object as instance id

(this is how NavMeshBuildSource type works)

[novel plinth]

@remote drift if you don't mind with reflection at all, there's protected api iirc called FindObjectFromInstanceID

[remote drift]

🤔

[novel plinth]

it's under UnityEngine.Object

[remote drift]

I'd love to avoid indirection

nope

ExposedReference is not it

[novel plinth]

not quite sure what you mean by that, you wont see anything there as this is resolved internally

[remote drift]

it has managed field

that struct is managed

thus - not burst compatible

(besides, it's unclear what's it used for even)

[valid tapir]

Hi, Is there a faster way to get Layer Index of Mesh Renderer, I have about 19k of them and I want to sort them by layer which on PC takes about 4ms, 3ms when I pass data to the JobSystem, and do the comparison there, the 2.79ms is spent on getting gameObject.layer. I can't pre cache this because the parts of a scene are being loaded and unloaded outside of my scope as Scenes.

[sly grove]

crosspost^

[candid meteor]

sorry

[opaque elbow]

Hi, I'm having an issue with implementing the IEqualityComparer interface, I have the following unit test, when I try and run it with the comparer the test fails, but when I try to compare each memeber individually without using the comparer the test fails.

    
    public class SaveData
    {
        class GameDataComparer : IEqualityComparer<GameData>
        {
            public bool Equals(GameData x, GameData y)
            {
                if (ReferenceEquals(x, y)) return true;
                if (ReferenceEquals(x, null)) return false;
                if (ReferenceEquals(y, null)) return false;
                return x.gold == y.gold && x.howardCoins == y.howardCoins;
            }

            public int GetHashCode(GameData obj)
            {
                return HashCode.Combine(obj.gold, obj.howardCoins);
            }
        }
        
        [UnityTest]
        public IEnumerator SaveSystemTest()
        {
            var gameObject = new GameObject();
            gameObject.AddComponent<DataPersistenceManager>();
            var saveDataTest = gameObject.AddComponent<SaveDataTest>();
            var gameDataComparer = new GameDataComparer();

            yield return new WaitForSeconds(1f);
            
            
            DataPersistenceManager.instance.fileName = "unit_test.save";
            DataPersistenceManager.instance.NewGame();
            DataPersistenceManager.instance.SaveGame();
            DataPersistenceManager.instance.LoadGame();
            
            GameData data = new GameData();
            data.gold = 999;
            data.howardCoins = 999;

            Assert.AreEqual(data, saveDataTest.data, "Saved game data isn't correctly loaded", gameDataComparer);
        }```

[sly grove]

AreEqual is not going to use your GameDataComparer

it's going to test object equality by looking at the GameData class.

If that class implements IEquatable, it will use that

otherwise it will use .Equals

which by default checks for reference equality

oh wait - sorry I just noticed you passed in gameDataComparer as a parameter to AreEquals 🤔

Is that actually a thing??

[scenic forge]

Is your SaveData supposed to just be an immutable stateless data container?
If that's the case, I'd suggest using record which have all of these implemented for you correctly.

[sly grove]

Or here rather, don't see IEqualityComparer as an option here:
https://docs.nunit.org/articles/nunit/writing-tests/assertions/classic-assertions/Assert.AreEqual.html

[opaque elbow]

It is mutable during gameplay. Also can you please provide more info or a link : )

[opaque elbow]

See this, https://docs.unity3d.com/ScriptReference/Assertions.Assert.AreEqual.html

[sly grove]

Gotcha ok.
So can you revisit your description? it says:

when I try and run it with the comparer the test fails, but when I try to compare each memeber individually without using the comparer the test fails.

So ... it's failing in both cases?

[opaque elbow]

sorry typo

[scenic forge]

https://learn.microsoft.com/en-us/dotnet/csharp/language-reference/builtin-types/record

[opaque elbow]

the test is passes when not using the comparer

[scenic forge]

You can't mutate records, but you can create a new record based on existing one:

var newSaveData = oldSaveData with { gold = newGold };

[opaque elbow]

yeah but my code structure can't fit using it

[sly grove]

what does "not using the comparer" look like?

Can you show the code?

also have you taken any debugging steps?

Debug.Logs etc?

[opaque elbow]

[UnityTest]
        public IEnumerator SaveSystemTest()
        {
            var gameObject = new GameObject();
            gameObject.AddComponent<DataPersistenceManager>();
            var saveDataTest = gameObject.AddComponent<SaveDataTest>();
            var gameDataComparer = new GameDataComparer();

            yield return new WaitForSeconds(1f);
            
            
            DataPersistenceManager.instance.fileName = "unit_test.save";
            DataPersistenceManager.instance.NewGame();
            DataPersistenceManager.instance.SaveGame();
            DataPersistenceManager.instance.LoadGame();
            
            GameData data = new GameData();
            data.gold = 999;
            data.howardCoins = 999;

            Assert.AreEqual(data.gold, saveDataTest.data.gold, "Saved game data isn't correctly loaded");
        }

[scenic forge]

Bummer, records are perfect for use cases like these.

[rancid kite]

Hellooo

Finally i find here yepp

[opaque elbow]

Yes, but I don't want to create a new object every time its a save game

Also I have an interface so objects can modify data

[rancid kite]

Ops sorry here is advanced

[opaque elbow]

yeah but the IEqualityComparer seems to not like it

[midnight vale]

Hi folks. I have two sprites with polygon colliders attached to them and I need them to detect a mouse hover. If these two sprites overlap is there anyway to "order" them so that a raycast from the mouse position will hit the topmost poly?

Both are in the same layer btw

[opaque elbow]

I missunderstood you, SaveData is just the script for the unit test

[sly grove]

z position.

Also you should/could use the event system for this

instead of doing a manual raycast

IPointerEnterHandler

[scenic forge]

Ah my bad, I meant to ask about GameData.

[midnight vale]

@sly grove Alright. Thanks I'll look into that 🙂

[opaque elbow]

Funny because I thought you were actually asking about GameData XD

[sly grove]

See how the physics 2d raycaster can sort things in different ways too:
https://docs.unity3d.com/Packages/com.unity.ugui@1.0/manual/script-Physics2DRaycaster.html

[midnight vale]

How would you implement an event system for that, like you mentioned?

[sly grove]

i.e. 2d sort order and render order

you don't implement the event system

you use Unity's existing event system

[midnight vale]

You mean IPointerEnterHandler?

[sly grove]

https://docs.unity3d.com/Packages/com.unity.ugui@1.0/manual/SupportedEvents.html

[sly grove]

yes which is why I said IPointerEnterHandler above

[midnight vale]

Ok now I get it. Thanks 🙂

[opaque elbow]

@scenic forge @sly grove found a solution, actually needed to implement https://learn.microsoft.com/en-us/dotnet/api/system.object.equals?view=net-7.0 instead of using the IEqualityComparer

thanks

[sly grove]

right that's kinda what I was hinting at originally...

or implementing the IEquatable interface

[opaque elbow]

yes

Idk why unity suggests doing it the other way arround if it isn't going to work XD

[scenic forge]

Ah, for completeness, here's a reference implementation for value semantics:

public class Stuff : IEquatable<Stuff>
{
    public string Foo { get; set; }
    public string Bar { get; set; }

    public bool Equals(Stuff other) => other is not null && Foo == other.Foo && Bar == other.Bar;

    public override bool Equals(object obj) => Equals(obj as Stuff);

    public override int GetHashCode() => (Foo, Bar).GetHashCode();

    public static bool operator ==(Stuff lhs, Stuff rhs) => Equals(lhs, rhs);

    public static bool operator !=(Stuff lhs, Stuff rhs) => !(lhs == rhs);
}

[grim dew]

Is anyone here familiar with unitys extension of statemachinebehaviours SceneLinkedSMB? Stumbled across a video that uses it to make a state machine for AI and would like to implement it in my game as well, but I have some questions about. Would love to pick someones brain on it who's familiar with it. https://www.youtube.com/watch?v=vxgxcDihld4&ab_channel=DapperDino is the video I found it from and here's the pastebin of the script: https://pastebin.com/zP9s2y8p

[plain granite]

is there any alternative for TypeCache.GetFieldsWithAttribute that i can use at runtime? i have a script that replaces the value of all fields that have a specific attribute, but because TypeCache is only available in the editor, i can't use it in a build

[compact ingot]

(maybe more of an opinion and not quite an answer) personally i'd make a state machine specific to my system's needs (may have multiple FSM implementations specialized to those), i'd skip the visual editor for anything that doesn't absolutely need it (terrible debug experience). If i need a visual behaviour editor i'd use Node Canvas (AssetStore). A FSM imo shines especially in two situations: either you want to view all the transition rules on one "page" in code for the ultimate separation of transitions from the effects of those transitions, or you want to securely encapsulate state's implementation into separate files/classes with lots of padding and contracts that get discussed extensively in meetings. IMO the best go-to/default FSM to start with is this adaptation of stateless for unity which can be used to implement arbitrarily complex systems, it does allocate on each transition so it's not ideal for being called each tick on many agents but one can easily adapt it to be non-allocating: https://openupm.com/packages/games.corundum.stateless-for-unity/
The core of such a declarative FSM engine can be implemented with maybe 500 lines of code.

[grim dew]

Part of the reason I would like to do it the way provided above is because I already have used the animation controller already for the animations, and the scenelinkedsmb seems like a reasonable extension to that without using different packages. For right now I am just looking to understand the rationale behind it so I can modify it if I need to.

[compact ingot]

in my mind its a semi-terrible contortion of the design intent of that state machine

[grim dew]

To be honest that was my initial thoughts as well, but I'm also thinking in the instance in which the animationcontroller state machine aligns exactly with the state machine you want, would it not make more sense to do it the way I've described?

[compact ingot]

sure if its exactly what you want

[dusty wigeon]

Having done a small amount of PlayerController, I can guarantee you that not every Animation State result in an state of your player and not every state of your player result in an Animation State. By example, you might have a Fall animation and a Fall High animation. I also suspect that you will have issues with transition/responsiveness due to the blending of the StateMachine; Something you do not need/want in the actual state of your Character.

While it seem like a good idea at the moment, those disparity will cause you a headache on the long run. That being said, I still use those StateMachineBehaviour to synchronize the Animator and the StateMachine of the character whenever it is necessary.

[bleak citrus]

agreed: it's often true, but not always

[steel snow]

can unity's jobs use delegates ?

[olive python]

I split screen dynamically at runtime so there is ghosting sometimes. How do I refresh unity rendering

[sage radish]

They didn't mention they were using ECS.

[real blaze]

right... my bad. I'll remove that

[thorn flint]

You can format your multiline code block with C# highlighting! Just add cs to the start of it. Highlighting example below!

class Fluffs : FluffyCat
{
    public void PetCat ()
    {
        Debug.Log ("Petting Cat.")
    }
}

[regal olive]

`

foreach (KeyCode kcode in Enum.GetValues(typeof(KeyCode)))
        {
            if (Input.GetKey(kcode))
                Debug.Log("KeyCode down: " + kcode);
        }

i've got this code here cuz i need to detect what key the player presses (i want custom keybinds but i don't want to have an extremely long scroll menu). tho whenever i press a button from the gamepad it records smth like this. i'd like it to record only JoystickButton9, not Joystick1Button9

idk if it belongs to #🖱️┃input-system

[unreal charm]

Need help with making the enemy ai use the same bullet controller the player is using

[bleak citrus]

don't crosspost in every channel at once.

[unreal charm]

i am not sure in wich channel this question falls in

[real blaze]

why don't you just ignore the one with 1 ?

[regal olive]

i fixed another way, but that was what i as tryna achieve

[ebon walrus]

Just saw there there is a networking channel under simulation - moving the question there.

[white seal]

Hello all! Was curious if anyone knew if Scene.GetRootGameObjects is guaranteed to be in the same order as the hierarchy?

The docs unfortunately don't show this to be the case but wasn't sure if it wasn't documented or if it's just not guaranteed.

[vale zenith]

I am a bit confused how to properly 'instantiate' a prefab via Resources.Load(). Right now I'm essentially doing the below, and it returns a value, however I don't see the prefab appear in my hierarchy in the editor. Do I have to call Instantiate() on myObject or do anything else?
GameObject myobject = Resources.Load(path) as GameObject;

[austere jewel]

Yes, you have to instantiate the object. Also, you should probably use the generic version of Resources.Load Resources.Load<GameObject>(path) and not cast.

[vale zenith]

OK, will do! Thank you

[kindred tusk]

If I have a class like this:

[Serializable]
class Foo { public int X; }
class Bar : MonoBehaviour {
  public Foo foo = new();
}

will there be an extra allocation of Foo before deseriaization when the constructor is called internally? Or will the instance created with new() be deserialized into by Unity?

e.g. if serialization is:

_foo: 5

Will the instantiation logic be?

_foo = new();
_foo = new() { X = 5 };

or:

_foo = new();
_foo.X = 5;

[hushed fable]

Afaik initializing serialized values will create extra work, so yes. I believe you can observe this by checking the value in constructor (before Unity's deserialization) and then in Awake.

[kindred tusk]

Thanks, that makes sense. It's a shame because I'd like to use nullable references, but even things that are guaranteed by unity deserialization need to be marked nullable

[potent shoal]

Greetings, I created a diagram to hopefully help explain my perspective.
I am trying to create an ability in my game that causes my velocity driven player controller to latch onto the closest object, and then maintain it's distance while moving around that object with a line renderer attached from the player to the object.
While moving around the target object, the player will be moving at it's original speed, but for each rotation around the object, the player speed will increase, to eventually create a sling shot effect.

Here are two representations. I want the distance the player is to the object to remain the same for the entire duration of the ability, but consistantly move the player around, rotating in a forwards direction.

Hoping for any ideas. Thank you!

[tiny pewter]

https://en.m.wikipedia.org/wiki/Gravity_assist
Similar to this but much complicated
Since you require maintain the same radius while rotation and increasing the velocity that means the acceleration needed to be increased during the trip, and the problem is when to stop the increasing of acceleration (gravity) so the object can be launched

[granite finch]

Can't really contribute much other than ask question. But... does the prefab look correct if you just drag it into the scene by itself?

[oblique field]

https://assetstore.unity.com/packages/templates/tutorials/google-cast-remote-display-sample-50169

anyone have any solution for casting a unity game? as googles cast doesn't work anymore, any one have any alternative?

I want to cast a game, but There is nothing in asset store.

[pulsar isle]

What are some use cases for Unity Native (c++ dlls) ?

Could a project benefit in speed or other metrics if someparts of the code are used from a native plugin ?

[dusty wigeon]

Integration of libraries that has already been written in C++/C.

[pulsar isle]

Yeah you are right ! But could be there performance improvements when translating a C# piece of code in C++. I know the l2cpp thing makes it almost C++ like in performance but I'm just wondering and trying to learn.

[dusty wigeon]

C++ is one of the faster to execute.

https://greenlab.di.uminho.pt/wp-content/uploads/2017/10/sleFinal.pdf?utm_source=thenewstack&utm_medium=website&utm_campaign=platform

[pulsar isle]

Even in the context of C# code being compiled to C++ via l2cpp ?

Interesting ! Thanks for the paper.

[dusty wigeon]

In theory. However in most case you will lose performance because you will always have a lot of Native-to-managed invocations.

Whenever you want to optimize something, you need to profile before hand.

[novel plinth]

there are cases where Mono is faster than il2cpp

one simple example such as virtual calls

[dusty wigeon]

Also, note that IL2CPP is not an optimization pass. It is mandatory for some platform.

[pulsar isle]

Yeah I'm so clueless about compilers and all these, just curious so if you have any helpful regarding articles, sources etc I could learn a lot without bugging you.

[pulsar isle]

for sure.

[dusty wigeon]

If you want to learn more about Il2CPP, Unity has some blog post on it works internally. https://blog.unity.com/engine-platform/an-introduction-to-ilcpp-internals

If you want to learn more about compiler in general, you should try to learn assembly.

[pulsar isle]

https://github.com/jacksondunstan/UnityNativeScripting

I found this too, @dusty wigeon Thanks for the suggestions !

[dusty wigeon]

There is also University Course

Also, you should definitely look into Burst as it is the standard for optimized code in Unity. @pulsar isle

[sand wraith]

I want create an empty gameobject during editor play mode. Is there any way to make that gamobject stay in scene even after i exit playmode?

anyone?

[tiny pewter]

the only way i know is copy the GO in play mode and paste it in edit(?) mode

[novel plinth]

the usual/ideal workflow is to author your game in edit-mode, like crafting or authoring your game world etc

thus we have edit and play modes

[hybrid gale]

hi ! i am using Input system to read input values of a controler.
so i did it with a PS controler, and it does nice.
and i wanted to do it with a friend Xbox controler, so Unity has recnized the controler, but Input system wont listen to any input of the Xbox controler.
So, is it possible to use 2 differents controler with Input action ?

[regal olive]

yes it is

dm please

[hybrid gale]

it doesn't work even if i use 2 files

[tawny sand]

i have a rigidbody2d which i'd like to smoothly move to a very specific location
i would like to use addforce for this (to avoid the physics engine disembowling me) but am not sure on how to find the right amount of force to add over a certain period of time, as I would like to smoothly deaccelerate the rigidbody as it gets closer to its stopping point
Any help would be very appreciated. Thanks!

[sly grove]

To do this the right way you need what's known as a PID controller.

[tawny sand]

a... what?

PraetorBlue is typing....

oh no

ok so i looked it u

*up

[sly grove]

It's basically how things like cruise control in your car work, and industrial robot arm controllers, drones, etc... You plug in your desired variables (position, speed) and the thing is constantly checking it scurrent state vs the desired state, the error in the variables, and adjusting etc to achieve what it wants
https://en.wikipedia.org/wiki/PID_controller

[tawny sand]

are you a mechanical engineer in real life

[sly grove]

software engineer

[tawny sand]

is game dev a side gig?

it shows lmao

in a good way

because jeesus christ

[patent bear]

Is there any way to set up a SpriteRenderer to use submeshes and different materials? I am familiar with Sprite.OverrideGeometry but I'd like to create two meshes where one is opaque and one is transparent without needing to move everything to MeshRenderer, if possible.

[raw lily]

I have a script I got from Github that works fine, but only when the GameObject is selected in inspector. For that reason the script does not work in a build, any idea what could cause a weird behaviour like this?

[austere jewel]

Seeing as nobody knows what this script is or even what it does, no

[raw lily]

Yeah I was wondering if there wasn't something that happened in Unity when a script was selected in editor, turns out I was looking too far, thanks!

[thin citrus]

Hi. I need to do cutscene sequence in my turn base game. For example, move the camera, play shoot animation on character, play hit on character target, display hit damage, etc. I think Timeline is a good candidate but my first impression is it look like it is better for predefined cutscenes then runtime one.

For example, I want to apply the previous sequence on any character, the shoot animation is different per character, the camera should be on the side with clear view, etc. I know you can make custom clip but I wanted to know if it is worth going into this rabbit hole or not. I want to be able to easily modify the sequence without the need of code. I am open to any alternative if you have one

[brittle dove]

I have this error that wont tell me what script it is from. All I know is everytime I swap data in a field in my inspector I get this bug. It didn't happen until I upgraded from Unity 2021.3 to Unity 2022.3

ArgumentNullException: Value cannot be null.
Parameter name: _unity_self
UnityEditor.SerializedObject.FindProperty (System.String propertyPath) (at <ac87b54e58ee4be198261fd5c8030d52>:0)
UnityEditor.UIElements.Bindings.SerializedObjectBindingContext.BindPropertyRelative (UnityEngine.UIElements.IBindable field, UnityEditor.SerializedProperty parentProperty) (at <9f0f853070524c139e10a85a1cfcedb6>:0)
UnityEditor.UIElements.Bindings.SerializedObjectBindingContext.BindTree (UnityEngine.UIElements.VisualElement element, UnityEditor.SerializedProperty parentProperty) (at <9f0f853070524c139e10a85a1cfcedb6>:0)
UnityEditor.UIElements.Bindings.SerializedObjectBindingContext.ContinueBinding (UnityEngine.UIElements.VisualElement element, UnityEditor.SerializedProperty parentProperty) (at <9f0f853070524c139e10a85a1cfcedb6>:0)
UnityEditor.UIElements.Bindings.DefaultSerializedObjectBindingImplementation+BindingRequest.Bind (UnityEngine.UIElements.VisualElement element) (at <9f0f853070524c139e10a85a1cfcedb6>:0)
UnityEngine.UIElements.VisualTreeBindingsUpdater.Update () (at <735c071c072642ad8d077fa38c650a28>:0)
UnityEngine.UIElements.VisualTreeUpdater.UpdateVisualTreePhase (UnityEngine.UIElements.VisualTreeUpdatePhase phase) (at <735c071c072642ad8d077fa38c650a28>:0)
UnityEngine.UIElements.Panel.UpdateBindings () (at <735c071c072642ad8d077fa38c650a28>:0)
UnityEngine.UIElements.UIElementsUtility.UnityEngine.UIElements.IUIElementsUtility.UpdateSchedulers () (at <735c071c072642ad8d077fa38c650a28>:0)
UnityEngine.UIElements.UIEventRegistration.UpdateSchedulers () (at <735c071c072642ad8d077fa38c650a28>:0)
UnityEditor.RetainedMode.UpdateSchedulers () (at <9f0f853070524c139e10a85a1cfcedb6>:0)

[brittle dove]

Reverted back to my backup gonna test in 2021.3 tomorrow and see if that error keeps happening.

[random dust]

Are they both LTS versions?

When you don't see where code comes fromon the stacktrace, it is compiled and no real trace exists. This either happends because it's not your code, or you are playing in a build

In this case, it appears to be a Unity bug

So changing Unity versions could fix it

[wide loom]

I'm generating a world layout on a 2D grid. I'm using a L-Systems to generate a road layout (green tiles) And then putting buildings around the roads. Now I want to divide the grid up into "Towns" based on the clustering of populated cells. I think I need to feed the Vector2Int[] array of populated (roads and buildings) cells into a Cluster analysis function; but this is where I'm lost.

Does anyone know of a library that would be appropriate for creating the cluster boxes?

[upbeat path]

you could start looking here
https://github.com/topics/clustering-algorithm?l=c%23

[wide loom]

I found this which seems to work quite well for what I want: https://github.com/ogxd/kmeans-clustering-unity

[dusty wigeon]

That should do the job. There is a ton of Clustering Algorithm out there if it does not do exactly what you want.

[pliant hawk]

Hello everyone,

Im currently working on a windows exe. where 1 user can controll inputs with touch or mouse, while a second user can conect via a XR HMD. You can think of it as a local multiplayer (its not but you get the idea).
I have done this before, last time with:
Unity 2021.3.11f1 / OpenXR 1.4.2 / Interaction Toolkit 2.0.4

since its a HDRP project i chose to update for this one to a newer Unity version because of all the Raytracing Performance increases.

The current project is with:
Unity 2022.2.15f1 / OpenXR 1.7.0 / Toolkit 2.3.1

The setup im running is one complete XR rig, with handtracking and controller other than very similiar in the general structure to the "Complete XR Origin Hands Set Up" that is part of the interaction toolkit. So only one "MainCamera" rendering to screen 1, with depth 0

The second user, the 2D user has an individual player manager with a seperate camera, also rendering to screen 1

Now as soon as i disconnect the headset my FPS drop from around 89 to 3, and the main change i can see is that the "XR.Display.PopulateNextFrameDesc250.18ms" increases by this amount.

I tryed detecting the disconnection from the HMD but i can only detect its connection not the disconnection. I think that there is something still trying to access the HMD even after it beeing disconnected.
I tryed this the following 2 ways:

    void OnDeviceChanged(UnityEngine.InputSystem.InputDevice device, InputDeviceChange change)

    if (device is UnityEngine.InputSystem.XR.XRHMD && change == InputDeviceChange.Added)

Aswell as with the new input system with a HMD action as a Button with "isTracked[OpenXr HMD] binding properties

Either way i cant get e message when the HMD disconnects.
As far as testing goes, there seems to be a problem with me having 2 cameras rendering to screen 1. As soon as i dont use a second camera the fps dont drop. But i do get the same result if i add a Canvas ScreenSpace that renders to screen 1.

Maybe someone has a idea what could cause this behaviour or some recomandations on how to properly use a multi camera setup for VR and a windows touch/mouse user.
Thanks in advance!

[brittle dove]

Yes

[brittle dove]

I should have said this last night. I decided to run the project in Unity 2021.3 and it seems that issue is gone. So the I'm only having the issue in Unity 2022.3

And also this is Editor only atm since I haven't done any build with this new project.

[storm tartan]

can someone throw me some document and tutorials about setting a runtime database? It has to be a safe one, the player should not be able to send&change data without the permission of the database's controller.

currently I am using google runtime database, but the player is able to change the data anytime he want to

so there is supposed to be a controller/confirmer in the middle of runtime - client connection

how can I do that in unity and google runtime database concepts?

any recommendations are welcome

[sly grove]

generally your client app should not be connecting directly to your database

it should be connecting to some web application you control (which enforces authentication/authorization of requests) and the web app is the thing that talks to the DB

though I do think something like Firebase might have document-level auth built in such that a player can only modify his own document.

[storm tartan]

yes I am actually searching for something like that

what is it called?

[sly grove]

Well there are about 39840912312 different ways to make a web application

pretty much every programming language has dozens of frameworks for it

so... it kinda depends on what you're comfortable with

or want to learn

(note that a web application is not the same as a web_site_)

[delicate perch]

Web application made using emoji programming language

[limpid pagoda]

i know i can insert functions as part of my animation
however can i do that as part of an animation's (a state's) transition to another?

i have some interruptions set up
is there a way for me to call a function once the mid point of the transition has taken place?

at the moment when i press LMB down a transition, which happens to take a whole second, takes it from state A to state B.
if i release LMB at any moment, though, the transition inverts, and the animation walks back to A.
how can i call a function (only) if more than 50% of the transition has happened?

[long ivy]

sounds like a job for StateMachineBehaviour

assuming you don't want the more direct route of just checking the animator state on mouse release

[limpid pagoda]

i'll take anything
i know such fine control would require diving deeper
i just don't know where to start
i read the reference but it feels like some terms aren't sufficiently explained

is StateMachineBehaviour the guy responsible for blending the animations along the transition?

[long ivy]

it's a scriptable object that receives messages when animator state changes

you could do fancy things with it sure, but you probably just want to create one that checks the animator state and if it meets your condition, looks at the Animator's GO and calls a method on some script on that GO

[dusty wigeon]

If you really want fine tuning, you can use Playable instead of an Animator.

Howerver, i doubt your issue cannot be solve by an other approach.

[limpid pagoda]

holding LMB starts opening a "mouth" (transition from closed to open which are animations with 1 single frame)
release LMB before 50% transition = transition back to closed
release LMB after 50% transition = transition to bite animation instead

[dusty wigeon]

Can you not add an extra state inbetween ?

Also, is it skinned ?

[limpid pagoda]

that's actually a valid idea

what's skinned in this context?

[dusty wigeon]

Skinned Mesh

[limpid pagoda]

simple 2d image with no mesh

[dusty wigeon]

If there is a small amount of animation data, you could do it by code.

[long ivy]

I would examine the animator state directly for that case and decide which parameter to set on the state machine based on your condition instead of trying to do something fancy

[dusty wigeon]

If this is trully a major part of your game, you can read the animation manually.

It won't be performant, but if you have nothing else going it can do the job.

[limpid pagoda]

it's actually probably the most relevant mechanic lol

[dusty wigeon]

Also, how are you animating the 2D image ?

Is it a sequence of image ?

Like a spreatsheet ?

[limpid pagoda]

it's all rotations of the image, on the animation clip's curve

[dusty wigeon]

It is only rotation ?

[limpid pagoda]

i have one image
and an "animation clip" with one frame, which has it rotated to a certain value

open = 0 rotation
closed = 90 rotation for example

[dusty wigeon]

100% do it through code.

[limpid pagoda]

the animation is actually the transition between those two "frozen" states

[dusty wigeon]

Your use case is not made for the Animator.

[limpid pagoda]

yeah i guess i more or less suspeced that

[dusty wigeon]

You could even use things like DoTween.

[limpid pagoda]

hold LMB, trigger on, scythesclosed -> scythesopen
i have some "interrupts" on (decided to check it when i saw it and it turns out it does exactly what i thought it'd do - allows transition to quit midway)
release LMB goes back.
but is it actually possible to make it "go back" to a different animation (cleave instead of close - say if i turn a trigger on) with the default statemachinebehaviour?

the following question would be how to bind the release of a button to the action map in order to check for how much of the transition has taken place

[dusty wigeon]

Are you trying to make a sort of charge mechanic ?

[limpid pagoda]

kind of

it's basically, open less than 50% and release, back to closed state
open more than 50% and release = cleave

dude chatgpt is nuts lol
tells me stuff about how transitions and triggers work better than unity reference

[dusty wigeon]

Because, the tradionnal way of doing animation is (In, Strike, Out) or (In, Loop, Strike/Out).

[limpid pagoda]

yeah i'll need to study some of the elements i'm already using (without fully understanding) before proceedign

i just learned a trigger is NOT reset if the transition is interrupted

[dusty wigeon]

I have done some charge abilities and it works well with looping animation.

[limpid pagoda]

i'll probably need some finer control regardless because the outcome of the "cleave" will depend on the timing (how much it opened, how long it stayed 100% open before the cleave happened etc)

[dusty wigeon]

You do this by code.

Just use the animation lenght

[limpid pagoda]

thanks for the tips

i definitely need to read more

[dusty wigeon]

It seem that you try to make your logic inside the Animator while it jobs is to animate.

[dire viper]

Hello guys, I am facing some big performances issues when launching my game. The start menu is starting fine but when clicking the play button, the game takes 30s to start :/ Any idea ? I start to look at the profiler but couldn't understand it so far.
You can find the profiler screen shot attached.

I see APplication.Preload.Asset, is it the Resources.Load that I am doing to generate my object in my scenes that takes so many times ?

Which is weird is that the first scene is really slow and the one after are direct :/

[sly grove]

you have to actually show the call stack section of the profiler

[dire viper]

That ?

It seems that it loads all the music 😮

[sly grove]

no not that

Switch from Timeline view to Hierarchy view

[sly grove]

although yes this does seem to show you are reading a bunch of files

whatever this preload assets thing is

[dire viper]

It would totally make sense if it was that, it would explain why it's slow for the load of scene 1

Looks like we have something there 😄

[limpid pagoda]

in regards to double tap:

if i have one action mapped to single click
and another, different action mapped to double click (0.75s max time between clicks is the default)

will the first action (single click) only execute 0.75s after the first click (in order to make sure it's not going to be a double click)?

[sly grove]

No, it will execute immediately. The two actions are independent

If you want to do that kind of exclusive handling you may need to write some code yourself

[limpid pagoda]

mind taking a look at #💻┃code-beginner ?
about preventing double execution of Play routine?

[sly grove]

Already responded

[austere garnet]

Hi, I'm looking for any better solution/workaround to this issue I experience constantly with Visual Studio on Mac:

When I create or rename a script file in the VS editor, I'll often see an error similar to this when I try to compile:

CSC: Error CS2001: Source file '/path/to/project/Assets/Scripts/SomeFolder/NewClass.cs' could not be found. (CS2001) (Assembly-CSharp)
The only way I've found to resolve this is to open unity and select Unity -> Preferences -> Regenerate project files. Is there any way avoid this or to fix it from within VS instead?

this seems to happen because VS creates a new empty file with the name NewClass.cs; the error happens after I've renamed it (on VS Windows, creating a new file opens a dialog, which gets saved with the name I give it)

[upbeat path]

normally you would only get this whilst modifying a file and in play mode in the editor

[austere garnet]

weird, this happens even when I don't have the editor open

[placid shell]

is there any easy way to copy read only MeshData into the writeable MeshData? (guessing using unsafe utilities perhaps?)

[obtuse arch]

Incase there is a build target mismatch the shaders will result render in error mode, For eg if you built for android and ran them on windows, The shaders wont work since they are built for android. Make sure the execution mode is set to 'Use AssetDatabase' instead of 'Simulate Groups'

[finite swan]

Texture2D.GetRawTextureData() is returning a byte length of 0, even though i can happily pass the texture to a shader and it can read it just fine.

I use Texture2D.CreateExternalTexture(...) to create the texture, calling a Native Rendering Plugin I wrote that reads BMP files:

// Creates BMP
extern "C" UNITY_INTERFACE_EXPORT void* CreateTexture2DFromBmpFile(const char* file, int* width, int* height, int* status) {
    
    // Read the BMP
    bmpread_t bmp;
    auto bmp_res = bmpread(file, BMPREAD_ALPHA | BMPREAD_BYTE_ALIGN | BMPREAD_ANY_SIZE, &bmp);
    if (bmp_res == 0) { // Failed to load the image
        *status = bmp_res;
        return nullptr;
    }

    *width = bmp.width;
    *height = bmp.height;

    // Generate the DX13 texture
    ID3D11Texture2D* mPtr;
    D3D11_TEXTURE2D_DESC textureDesc;
    ZeroMemory(&textureDesc, sizeof(textureDesc));
    textureDesc.Width = bmp.width;
    textureDesc.Height = bmp.height;
    textureDesc.MipLevels = 1;
    textureDesc.ArraySize = 1;
    textureDesc.Format = DXGI_FORMAT_R8G8B8A8_UNORM;
    textureDesc.SampleDesc.Count = 1;
    textureDesc.Usage = D3D11_USAGE_DEFAULT;
    textureDesc.BindFlags = D3D11_BIND_SHADER_RESOURCE;
    textureDesc.CPUAccessFlags = D3D11_CPU_ACCESS_READ;
    textureDesc.MiscFlags = 0;

    D3D11_SUBRESOURCE_DATA subres;
    subres.pSysMem = bmp.data;
    subres.SysMemPitch = bmp.width * 4;
    subres.SysMemSlicePitch = 0; // Not needed since this is a 2d texture

    auto tex_res = g_D3D11Device->CreateTexture2D(&textureDesc, &subres, &mPtr);
    if (tex_res != S_OK) { // Failed to create the texture
        *status = tex_res;
        return nullptr;
    }

    // Clear the memory and return
    bmpread_free(&bmp);
    return mPtr;
}

what am i doing wrong here? I've tried setting the CPUAccessFlags to CPU_ACCESS_READ but that didnt change anything.

[stuck onyx]

Hello people!
I want to do something very simple, when i collide with an object i want to take it's material and fade it.
I know i just have to get the component MeshRenderer and do what i want with the material but I remember there was a tricky thing when doing this, like the material was duplicated or something, could someone point me where this 'tricky' thing was and how to solve it please?

materialPropertyBlock was the answer, although not sure if its worth it for just 1 material?

[tall ferry]

an instance of the material is just created, the original material in your assets stays the same because of this. This is likely what u want

[stuck onyx]

yeah but if we do that lots of times dozens or hundreds of materials will be dupli-triplicated

so i guess the only thing to do was to kill the duplicated material once im done with it?

(destroy) ?

[tall ferry]

if you need to destroy the actual object afterwards then do so, i dont know what functionality u are going for

u shouldnt worry about pre-optimizing this unless you've already tried and noticed some performance issue with the simple solution. Also, MaterialPropertyBlock says

Note that this is not compatible with SRP Batcher. Using this in the Universal Render Pipeline (URP), High Definition Render Pipeline (HDRP) or a custom render pipeline based on the Scriptable Render Pipeline (SRP) will likely result in a drop in performance.
https://docs.unity3d.com/ScriptReference/MaterialPropertyBlock.html

[stuck onyx]

no i dont have to destroy the object, the only thing i want is fade to transparent an object im crossing

this is how i did it:

[tall ferry]

how does this even run?

am i blind or is this not 2 infinite loops

[flint sage]

It is

[stuck onyx]

yes

2 infinite loops

🤣

[flint sage]

You're not actually modifying alpha

And you probably want a delay each loop

[stuck onyx]

true, should use a coroutine

[tall ferry]

either way though, is directly modifying the material even going to be an issue. Are u really going to have thousands of these objects just chilling around invisible?

[stuck onyx]

no, actually not, im going to do a tween with the material and f*it

[scenic gulch]

hey, does anybody know of a good resource to learn about how to do

whoops!

how to USE the jobs system with animations? I am looking to optimize our management game that has a few hundred NPCs

https://docs.unity3d.com/ScriptReference/Animations.AnimationScriptPlayable.html

I think that is what I need to be doing, but can't find much in the way of guidance of how to actually implement it

[brave cairn]

Create d3d11 texture with staging usage (not default): textureDesc.Usage = D3D11_USAGE_STAGING;

[finite swan]

Oh ok thanks. Any recommended staging ? I'm new to this low level stuff.

[brave cairn]

What do you mean by "Any recommended staging ?" Staging resources allowed to be read by CPU side.

[finite swan]

Ooh sorry I misinterpreted your message. Set the usage to staging gotchya. Not on my computer to read docs sorry.

[vestal condor]

hmm, just updated unity from 2021.3 to 2022.3 and i'm getting this error, no clue why
I've rebuilt my streaming assets as well, no clue what's throwing this error

DamageFactory is a scriptable object and i see that my scriptable objects are still intact in my project browser

[fresh salmon]

It's complaining about the scriptable object definition itself (the class), make sure you didn't change its name, and that class name and file name match: class DamageFactory : ScriptableObject in a file DamageFactory.cs

[vestal condor]

Yup, it's certainly matching

It only broke after updating to 2022

I'm not sure why it's throwing the warning, especially since existing scriptable objects are certainly working

[fresh salmon]

I'd say ignore these, if it didn't break anything

[vestal condor]

Well there's like 500 of those

[upbeat path]

if you clear the console do the messages persist?

[vestal condor]

happens whenever i enter play mode

i’m reimporting the entire project to see if it does anything

[upbeat path]

yep, probably need to delete the Library folder and let Unity rebuild it

[vestal condor]

https://forum.unity.com/threads/scene-couldnt-be-loaded-because-it-has-not-been-added-to-the-build-settings.1400674/

is anyone familiar with this problem?

im running into it again and i really have no clue how to fix it without reimporting the entire project, which i just did...

[plucky laurel]

Scenes/Splash Screen [Title]

Splash Screen [Title] is not in the build settings

[spare sentinel]

Hi. I have a mfps package and the mobile control addons fire button just fires. And it is really anoying that you need to aim with your other finger. Most fps games gave the fire button functionality, that if you hold it, it shoots and also if you move your finger, it aims. So I tried to modify the script for mobile control addon, but made it so if you with EVERY BUTTON. And also the buttons do not work. Just aim. Can anyone please look into my script, specifically onto InFireClicked function and tell me what could be wrong?
https://gdl.space/amapahuxaf.cs

Thank you so much 🙂

[plucky trellis]

A good way to do this for games is to some some kind of second order equations. This video is how I learned about it. It's a bit heavy on math, but he does give example code. https://youtu.be/KPoeNZZ6H4s

[bleak citrus]

Good ol’ diff eq

I learned that back in my freshman year of college. Been a hot minute

[onyx blade]

Is it redundant to specify OnMouseEnter, OnMouseExit etc on an abstract class given they are event functions provided on MB?
I'm thinking I could specify it since most children will define those functions, but im not sure if all of them have to.
I hesitate because I don't currently have functionality for those functions in the abstract base class

the question rly is, is there a point in defineing them? does it matter at all given the compiler will probably strip them out if they are empty?

[sly grove]

there is no point in defining them on the base class if it doesn't have a default implementation for them

it will actually just hurt performance, forcing an empty method to run

[onyx blade]

I figured, I will remove the implementation in the base class for now.
if I end up with a situation where I do have functionality in that base class,
I can add it there and override the function, then call its base

[plucky laurel]

can also be solved compositionally, component that catches OnMouse and relays to target component through interface

no idea why

[onyx blade]

the OnMouse functionality is part of the monobehaviour though.
interfaces cannot inherit as far i am aware

unless I am mistaken

[plucky laurel]

components are monos for all intents i dont think you can directly inherit component without unity throwing, point is you dont always have to shove everything in mega sandbox base

[plucky laurel]

thats not what i meant

[onyx blade]

ah, then I misunderstood

you meant different components on the object relaying information between each other.
I read interface and assumed you meant the datatype

[plucky laurel]

yes, this is an idiotic example because all monobs will catch Mouse events, but for example IOnMouse.. you yourself declare, target types implement and relay invokes when needed, it is very useful in various places but its mostly between game objects

anyway, dont use OnMouse

[onyx blade]

the logic is solid, altho it would indeed be a weird usecase talking about the mouse events in MB's

[plucky laurel]

use event system IPointerEnter etc

[onyx blade]

I have seen that API, never have gotten it to work sadly.
I have it set up on my project right now and it does not seem to behave as expected

[plucky laurel]

what happens?

Physics Raycaster present on camera?

[onyx blade]

public abstract class Interactable : SerializedMonoBehaviour, IPointerEnterHandler
{
    public void OnPointerEnter(PointerEventData eventData)
    {
        Debug.Log("HI");
    }
}```

[plucky laurel]

and all layers setup properly?

[plucky laurel]

event system present in the scene?

[onyx blade]

that's the one

I dont have one in my scene, cuz im usign the other input system and didnt have UI yet

[plucky laurel]

it works in tandem with ui, you wont click through it like with OnMouse

[onyx blade]

I plan to do all the input from input actions myself

[onyx blade]

the way I was doing it would cause multiple events to be fired, if multiple overlapping colliders were triggered?

[plucky laurel]

parameter on raycaster

OnMouse you mean? i dont know i used it a long time ago

you have zero control over it

with event system you have its full source

[onyx blade]

I was about to say

[plucky laurel]

you can do whatever you want

[onyx blade]

that is quite a bit more convenient if im honest

more control is always better

and since the pointerevents are interfaces I can implement them in the base or its children depending on where I need them

[plucky laurel]

yes

ill send you some code

[onyx blade]

that leaves me with th equestion why you would want to use the system I was using 🤔

[plucky laurel]

what do you mean

[onyx blade]

well. the method you described seems much more elegant and easier to use cuz of the flexibility

[plucky laurel]

OnMouse is legacy

[onyx blade]

which makes me question what the purpose is in having both systems exist

ah

that'll do it

[plucky laurel]

shouldnt have been there in the first place

like a lot of other baggage

[onyx blade]

there are quite a few things like that in the engine either way

some legacy, some unfinished

personally, I am referring to the many itterations of node stuff in the editor

[plucky laurel]

my favorite is - magic methods, properties on Component, BroadcastMessage, blackbox animator, general naming of things like Random/Debug

everyone is stuck with magic methods because at some point unity supported 3 languages

[onyx blade]

I tihnk I found 4 different node code-bases in the editor while I was crafting our state machine

[plucky laurel]

which ones?

[onyx blade]

and all of them had their own annoying bugs

[plucky laurel]

animator sm i know

[onyx blade]

I used XNode (tho thats a 3rd party one), the one from Bolt,
then there is the shadergraph api and amother one I forgot the name of

the last one was highly experimental and only exists on a git repo

cant even find it anymore, been a while

[plucky laurel]

xnode and bolt use imgui afaik

i dont know how people work with it

[onyx blade]

its weird, I worked with it for a day or 2 and just ew

the shadergraph api is quite nice, but it has a lot of setup

[plucky laurel]

it lags that the biggest downside, its made so badly compared to something like dearimgui its laughable

shadergraph at least is retained

so you can have thousand nodes and its fine, but the feel of it, the design is bad

[onyx blade]

ya it doesnt feel great

I remember one of the api's for nodes was super laggy when you wanted to drag

either a node or panning the screen

it was shitting itself

[plucky laurel]

yeah unity imgui is not made for that, you can squeeze everything out of it with your own optimizations, like doing manual bvh, swithing to Repaint event, not using layout at all, hiding any property fields , culling

i just hope dearimgui will be ported to editor by some genius

some day i dont want to use uitk and manage retained objects

[onyx blade]

uitk is quite nice to work, it's much better compared to the other systems once I got the hang of it

some things work, others are lacking. but most things work well enough to build on top of

[finite swan]

Hey, i have tried it and unity giving me an error now.
d3d11: failed to create 2D texture shader resource view id=1007 [D3D error was 80070057]

From my research, it seems this is a invalid argument. All i did was change the Usage to USAGE_STAGING.

I tried:

• change CPUAccessFlags to D3D11_CPU_ACCESS_READ
• BindFlags to 0, D3D11_BIND_SHADER_RESOURCE, D3D11_BIND_CONSTANT_BUFFER
• Format to sRGB (DXGI_FORMAT_R8G8B8A8_UNORM_SRGB)

None of these worked and all threw the same error.

// CreateTexture2DFromBmpFile loads a bitmap from a file and stores it in a new Texture2D, returning the pointer of the texture.
extern "C" UNITY_INTERFACE_EXPORT void* CreateTexture2DFromBmpFile(const char* file, int* width, int* height, int* status) {
    
    // Read the BMP
    bmpread_t bmp;
    auto bmp_res = bmpread(file, BMPREAD_ALPHA | BMPREAD_BYTE_ALIGN | BMPREAD_ANY_SIZE, &bmp);
    if (bmp_res == 0) { // Failed to load the image
        *status = bmp_res;
        return nullptr;
    }

    *width = bmp.width;
    *height = bmp.height;

    // Generate the DX13 texture
    ID3D11Texture2D* mPtr;
    D3D11_TEXTURE2D_DESC textureDesc;
    ZeroMemory(&textureDesc, sizeof(textureDesc));
    textureDesc.Width = bmp.width;
    textureDesc.Height = bmp.height;
    textureDesc.MipLevels = 1;
    textureDesc.ArraySize = 1;
    textureDesc.Format = DXGI_FORMAT_R8G8B8A8_UNORM_SRGB;
    textureDesc.SampleDesc.Count = 1;
    textureDesc.Usage = D3D11_USAGE_STAGING;
    textureDesc.CPUAccessFlags = D3D11_CPU_ACCESS_READ;
    textureDesc.BindFlags = 0;
    textureDesc.MiscFlags = 0;

    D3D11_SUBRESOURCE_DATA subres;
    subres.pSysMem = bmp.data;
    subres.SysMemPitch = bmp.width * 4;
    subres.SysMemSlicePitch = 0; // Not needed since this is a 2d texture

    auto tex_res = g_D3D11Device->CreateTexture2D(&textureDesc, &subres, &mPtr);
    if (tex_res != S_OK) { // Failed to create the texture
        *status = tex_res;
        return nullptr;
    }

    // Clear the memory and return
    bmpread_free(&bmp);
    return mPtr;
}

[raw lily]

Hi, I'm duplicating a terrain at runtime, to a new GameObject, but it doesn't seem like I'm able to apply deformations to that new terrain at runtime, any idea what's wrong and how to work around it?

[slim crater]

Hi, I have a Unity game where I am using adressables (loacally), when I build for Android all work fine and adressables are loaded, but on IOS they are not loaded. Online I found a lot of people with this same problem but nobody found a solution to overcome the problem, is there someone that know how to solve this issue?

[thorn narwhal]

Try #📦┃addressables

[icy smelt]

You guys know if IL2CPP is compatible with the Roslyn API?

[sly grove]

what features of Roslyn are you wanting to use exactly?

All of this stuff works fine:
https://docs.unity3d.com/Manual/roslyn-analyzers.html

[midnight mural]

Hi, this code works perfectly fine in the unity editor environment but doesn't work on a windows build, any ideas why? Thanks: https://hastebin.com/share/gebohihife.csharp

[icy smelt]

I'm planning to use Roslyn to compile C# code on runtime to Expression Trees (as JIT compilation, or runtime library loading won't work with IL2CPP). I mean runtime stuff, not editor time

[sly grove]

what doesn't work about it? What debugging steps have you taken?

[midnight mural]

The build crashes when the function is called but in the editor it works fine

The error Im getting is a null reference exception, a gameobject is missing, by some reason, when value B is a null gameobject it is assigning value A for some reason, this happens just in the build in the editor it works fine

[sly grove]

possibly just a script execution order issue

for example some variable you are depending on is not yet initialized due to script execution happening in a different order

[regal olive]

Should I use System.Net.Http.HttpClient from inside C# Jobs?

MS docs (https://learn.microsoft.com/en-us/dotnet/api/system.net.http.httpclient?view=net-7.0#instancing) are clear about HttpClient is meant to be a long-lived object, and it is managed; a job on the other hand must work with unmanaged types if I want to work with the Burst compiler, meaning I am already dropping Burst out.

Moreover, my understanding is a job cannot work with a long-lived mutable object outside of it as it's not inherently thread-safe. Should I just create and dispose multiple clients inside the job in this case?

[umbral trail]

You can't use managed objects in jobs, but you wouldn't want to make a web request in a job anyway. You can just use UnityWebRequest and check if it's completed every frame until it's done

[scenic forge]

And lots of async options like UniTask or good old coroutines.

[vestal condor]

strange question but any idea why clicking "use null comparison" doesn't do anything? this warning is quite annoying to fix one by one

pressing "fix all occurences in document" doesnt actually do anything

is there some plugin i'm missing in my visual studio or something?

i'm pretty sure i have visual studio tools for unity installed

[fresh salmon]

Pretty sure it's the code fix that comes with the analyzer that's broken. Indeed you shouldn't use ?. because that ignores the "destroyed but not null yet" state of unity objects (it'll end up calling the method even if the script was destroyed, throwing an exception).
It'll most likely fix the code into

if (pressActivateActions)
  pressActivateActions.SetOwner(owner);

[vestal condor]

Yup, I know that the code fix is broken

I'm wondering how to make the code fix not broken, because this particular thing is quite annoying to fix manually

[fresh salmon]

See if the Visual Studio Editor package in the the Package Manager can be updated

[vestal condor]

basically I'm doing some code cleanup and noticed a lot of these and it would be preferable if I didn't have to fix them individually

[fresh salmon]

Other than that, you could use VS's search & replace that works with the full regex specification (yes, even capture groups and replacement values) to fix that in one go

Got one for ya https://regex101.com/r/hYlGJH/1

[vestal condor]

hm yea, I'm familiar with regex but I was hoping to avoid doing it lol

[dull kestrel]

Coroutines aren't async though?

[upbeat path]

no they are not

[regal olive]

smartest thing anyone in here has ever said

[scenic forge]

Not the narrow definition of C# async, but asynchronous programming in general.

[dull kestrel]

Async allows for blocking code to run without blocking the main thread. Coroutine doesn't allow that.

If I tried to download a 1GB file with a HttpRequest it'll freeze the game until it's finished.

[scenic forge]

That’s not true, UWR can run in a coroutine without blocking the main game loop.

[dull kestrel]

Then it's doing something internally.

Perhaps in a Task/Thread.

[scenic forge]

Stuffs like IO whether it’s disk or network are like that.

[dull kestrel]

If I did

do 
{
  i++;
} while (i < int.MaxValue);

It would freeze the application until it's finished. It's nothing to do with the IO.

Coroutines run on the main thread, as per the documentation.

However, it’s important to remember that coroutines aren’t threads. Synchronous operations that run within a coroutine still execute on the main thread

https://docs.unity3d.com/Manual/Coroutines.html

[dusty wigeon]

However, you could add a yield return which would make the things run on multiple frame. Making it asynchronously.

It will run on the same thread, we agree on that.

[scenic forge]

This blocks because it’s entirely CPU

[dull kestrel]

It's as much asynchronous than having another component to do stuff in an update or fixed update method.

[dusty wigeon]

False. Spreading the work on multiple frame is a valid optimization technique.

[dull kestrel]

If that's your stance though, you can recommend IEnumators/IEnumerables.

[scenic forge]

Disk and network isn’t, they are usually just CPU telling disk “ok go grab this file for me, and call me back when you are done” and CPU is free to do other things in the meantime, such as continuing the game loop.

[drifting vessel]

Does anyone know if Array.FindAll is fast, or should I implement my own algo? Iterating through hundreds of values.

[austere jewel]

Generally if you find yourself using those methods there's a better collection for you

[drifting vessel]

I'm listening

[small latch]

Makes a new list and array 😔

[austere jewel]

Dictionary, Hashset, whatever, it's unclear what you're trying to do

[drifting vessel]

Right, so as it stands, I'm pinging off of an API to get a list of values beginning with a search value. But I want to do this every time the player changes the value, so ideally I'd want something local and fast, and it can later use that name as a parameter for the API to get the rest of the data.

[scenic forge]

Hundreds is tiny number for modern computers unless you are doing this enough for performance to matter, and at that point the allocation is probably what’s hurting more.

And as usual with performance, always benchmark first before stabbing at code blind.

[fresh verge]

How safe is it to do this inside a coroutine?

Time.deltaTime;
yield return new WaitForEndOfFrame();

Aka, will delta time return the expect values?

[sly grove]

The first line of code is an error as it's not valid C#

[fresh verge]

I am aware

[sly grove]

the second line - you want yield return null if you just want to wait one frame

[fresh verge]

Thought return null would quit the coroutine entirely

[sly grove]

why would you think that

yield break; would quit the coroutine

[fresh verge]

ah, sorry, missread

So, yield return null is what I would use to go to the next frame instead of wait for end of frame?

"go"

[austere jewel]

There is no 'go', there is only waiting

WaitForEndOfFrame is almost never what you want to use

yielding null will make execution return to that point on the next frame

[fresh verge]

Huh, I wonder why it's null instead of something like "WaitForUpdate"

[austere jewel]

So it doesn't allocate anything

[fresh verge]

Make sense

[mellow sail]

Do you guys know anyways to disable drag and drop feature on list using Odin Inspector, so if I drag a game object into the list in inspector, it wont be added?

[ember temple]

Would marking it as read only work? https://odininspector.com/attributes/read-only-attribute

[mellow sail]

Well i could but the thing is, i need the object in those list to be costumizeable (Inline Editor) but i just wanted for the list to be not able to drag and drop

So when using the tool i build, people would not accidentaly added an object in the list

but could still added an object that are already in the list for the object that is intended using a button

so read only is not an option

never mind, found it

found another way around

[regal olive]

hey yall how do i lerp a color to transparent? ive been stumped on how to use lerp for 2 days including today now

i have a scene variable set for color, from there i grabbed the material color for the asset i want to use for the variable, i called that objecta so thatd go to lerp a<-, b, t, right? so whats the line to set the alpha for objecta to 0 then apply it to b

do i need to create a second variable and basically set alpha to 0 for the second color variable? if so how do i go about doing that, and if not, what do I do for b on lerp?

thanks

[tall ferry]

this isnt an advanced issue, but to answer all you need to do is set the material.color to your new color.
Storing the color from a material, then adjusting it wont change the color on the material

your alpha value is between 0 and 1, not 255 incase you were using that

[regal olive]

yea ure right this should be in general now that i think about it, but lets say the object is an instiante

im think i should create a variable for the instiante and set the material for that, does that sound right

[tall ferry]

doesnt matter where the object comes from, as long as you have a reference to it. Then you just grab the renderer's material, and update the material.color with whatever you want

[regal olive]

ahhh ok ok, ill need to adjust the renderer of the asset the instiante grabs to the new color is what ure saying ok got it

thanks

[tall ferry]

im not sure what you mean for instantiate, that is for spawning objects. Once the object is spawned you just need GetComponent for the renderer. The renderer has a .material, and the material has a .color

[regal olive]

yeah, i didnt mention that. i want to spawn an object then fade it out from the up position of a button. i have everrything set so that the instantiate is destroyed, just cant figure out how to fade it. i got the general gist about it now thank you

[tall ferry]

hopefully u already have a shader for this, because changing the alpha on the regular opaque shaders do nothing

[regal olive]

idk what that means, im more of a hands on learner so once i get to that bridge and unity stops returning errors then ill try to look into that before i ask wtf does that mean lmao

[tall ferry]

default URP shader selected, i just made a material for it to edit it.
I have the alpha on 0, but the cube is still fully visible

unsure if this is different in other render pipelines, i know almost nothing about graphics.
But theres a few workarounds, like using surface type - transparent or writing a custom shader.
Transparent is kinda weird if your object has any unique shape because it doesnt write to depth and can render inside of itself then

now that its on transparent, its gone

[regal olive]

ah ok i just need to change the surface type to transparent. sounds good, ill remember to do that

[quartz gulch]

Hey all - so I have a webgl project (2022.2) where I have several scenes bundled as asset bundles and I'm loading each one depending on which is needed - this is working great with one issue.

It seems that baked NavMeshes aren't exported with my scene in the asset bundle which means that when I load the scene there's no NavMesh. I've been thinking up solutions and I could either generate the NavMesh at runtime (not a fan of this solution as WebGL is extremely CPU limited and I'd rather do it offline). Instead I want to serialize the NavMesh and store the data in a gameobject, then deserialize this data at runtime to recreate the NavMesh. Is this something which is doable? Should I just bite the bullet and generate at runtime?

I can't seem to find a good way to do this from my brief experiment, if anyone could point me in the right direction or tell me if this solution is not the way to go I would appreciate it

[tropic stag]

I'm struggling with "active ragdolls" for a while now. Anyone have some pointer on how to loop through character joints and create configurable joints from them to match the 2 axis and limits?

The ragdoll builder works, but uses character joints, and I'd like to have the configurable ones to apply "drive".

[misty glade]

Does https://docs.unity3d.com/ScriptReference/Object.FindAnyObjectByType.html restrict the search to the current scene? I'm unclear what "loaded object" means in this doc.

[flint sage]

If you're using it for game code, you don't really have to worry about it

[real hazel]

I am calling this every fixed frame to detect collisions at really high speeds and I seemingly have to use raycast all to check collisions since the raycast hits the projectile. I do not know much about performance but I am pretty sure sorting an array every physics frame is bad. Is there another way around this?

        float distance = direction.magnitude + 0.2f;

        RaycastHit[] hits;
        hits = Physics.RaycastAll(lastPos, transform.forward, distance);
        System.Array.Sort(hits,(a, b) =>(a.distance.CompareTo(b.distance)));

        foreach (RaycastHit hit in hits)
        {
            GameObject hitObj = hit.collider.gameObject;
            Vector3 hitPos = hit.point;
            Vector3 hitRot = hit.normal;

            if (transform.GetComponent<Collider>() != null)
                if (hit.collider == transform.GetComponent<Collider>()) continue;

            activeProjectile = false;
            OnHit(hitObj, hitPos, hitRot);
            transform.position = hitPos;
                
            break;
            }
        }
        lastPos = tip.position;

[dull kestrel]

• Does your projectile require a collider?
• could your projectile be on a separate layer which you exclude?

[real hazel]

some projectiles will need a collider and some projectiles will need to collide with eachother

[dusty wigeon]

Object.FindAnyObjectOfType doesn't return Assets (for example meshes, textures, or prefabs), or inactive objects. It also doesn't return objects that have HideFlags.DontSave set.

[misty glade]

The problem with that sentence is the API has an override that takes FindObjectsInactive.Include

[dusty wigeon]

It still says that it does not return Asset.

I do not know what could be an UnityEngine.Object if it is not an Asset or an instance in a scene.

Every time I used it I had no issue. I used: FindObjectOfType

[misty glade]

the autodocs indicate it does find inactive objects:

FindObjectOfType is slow - even FindFirstObjectByType is documented as slow

[dusty wigeon]

Yeah, not an issue on an Awake.

If you use it with care*

So the issue, is that it find disabled object ?

I though you were asking about if it was only object in the scene.

[misty glade]

It's all good - I've got a workaround solution for now, just saving a reference to the object. I'm using it to find a "fader" when transitioning between scenes, and the component is disabled until it's needed, but I needed a way to find it. It's also fine since I'm only calling this once between scene transitions

[dusty wigeon]

Seem to me that you have truly a better solution than using FindAnyObjectOfType.

[scenic forge]

In 2D, let's say I have a massive amount of static objects and their bounding boxes, and I also have a dynamic moving view box. What's the best way to store these objects' bounding boxes, so that I can efficiently query which ones overlap with the view box?

Oh I'd like to implement this on my own, not using Unity stuffs.

[sly grove]

A quadtree

[scenic forge]

Quadtree stores points though, no? How would I adapt that for boxes?

[tall ferry]

You should just go through the process of deleting the joints manually and making configurable joints. You can select all the body parts you need to do it at once.
It's really not that long of work compared to making an entire script to do it for you.

[sly grove]

bounding boxes are made up of 4 points

[plucky laurel]

quadtree can work with bounds just as well

[scenic forge]

Good insights thanks, I'll think about it.

[tropic stag]

But the ragdoll helper sets up the character joints with sensible axis limits, I was hoping I can copy those limits over to a corresponding configurable joint... I also saw someone on the internet say they didn't replace the joints but added configurable ones, leaving the character ones to limit motion, and adding freemoving configurable ones to use the drives... but that seems wrong to me somehow?

[scenic forge]

Are you also suggesting the same approach as Praetor's, by adding all four corners of each box to the quadtree?

[plucky laurel]

i suggest looking at github for any quadtree implementation using bounds

[scenic forge]

Quadtree itself only stores points and provides a way to query those points, no?

[plucky laurel]

https://github.com/dolejska-daniel/unity-quadtree/blob/76db238c5546fe9124a4d26356f0dc3e5c512b30/Scripts/INode.cs#L17

[tall ferry]

That is indeed wrong, your configurable joint should have complete control. With how fragile these are to setup,I wouldnt use any weird lifehack to save a few minutes.
An easy way to set it up, maybe just copy your ragdoll setup character so you have 2. On one, delete the character joints, make configurable joints, then copy from character to configurable.

[plucky laurel]

@scenic forge https://github.com/Nition/UnityOctree/tree/master/Scripts

its octree but its a good example of both approaches

[scenic forge]

Yeah, reading both repos atm 👍

[plucky laurel]

there is a simple approach with spatial hashing grid

easier to write, faster inserts

but unless you make it hierarchical you will have fixed size buckets which can be undesired depending on scenario

[scenic forge]

Yeah I don't know how close or far apart object boxes are.

[plucky laurel]

you have zooming?

[scenic forge]

Hmm unsure but unlikely, what are you going to suggest?

[plucky laurel]

problem with fixed buckets is that if you zoom out you get massive amount of buckets to query and you cant predict or find the golden balance of bucket size

but if amount of buckets at any point queried is invariant, no zoom, then its easy to estimate

for example it can be 1/4 of camera frustum

so you end up with max of 4-6-8 buckets on screen at any point

can go lower/higher to find best performance, since biggest impact is not from bucket iterations but overlap/distance tests per object

[scenic forge]

I see, I'll read up on spatial hashing grid as well.

[tropic stag]

copy what? I've been fiddling with this for a while, and I don't like setting the axis and limits with the gizmos, I'd like to copy those, but I'm not sure how the attributes map.. Is the configurable joint a "union" of properties of the other joints? (obviously not in the case of fixed or spring...)

[tall ferry]

copy the axis, but overall this is just something you need to take the time to setup. I was just suggesting a way to setup the configurable ones easier since you said the character joints have limits already.
and yea the configurable joint can be used to make every other joint. Even the spring or fixed joint

[dusty wigeon]

Just be cautious to not preemptively optimize. I would really be surprise that a custom solution is required in a 2D game. I'm expecting that only a small number of objects to actually be loaded at the same time.

[tall ferry]

i also didnt really like how setting up the axis were, but its just something you'll be fiddling with for awhile. Even with gizmos i found it extremely hard to visualize where the limits truly were. Theres a reason not too many games use real active ragdolls as their main character. Even on discussion forms, one commonly linked video goes into active ragdolls but the guy uses animations to drive the positions, so its not even an active ragdoll basically lol.

[tropic stag]

I'll give it another shot. I was also reading the famous ConfigurableJoint extension code everyone links and it specifically "reprojects" the bone rotation to a coordsystem built from the configurable joint axis, so in theory the axis don't HAVE to match, but I can't get it to work so I'm thinking it would probably be better if they do match...

[tall ferry]

do you mean this or something else? i dont really remember seeing extensions other than the following
https://gist.github.com/mstevenson/7b85893e8caf5ca034e6

[tropic stag]

yes, I thought for a hot second that worldToJointSpace guarantees that the rotations will match regardless of the joint configuration, and they will, but of course the "paths" towards the rotations depend on the axis so they should match the "standard" ones.

also.. slerpDrive.mode = JointDriveMode.Position; this makes me question why position and not rotation, but trying to look, the docs say that the jointdrivemode is obsolete, so perhaps there's some new line needed to properly configure the joints

[tall ferry]

Yea i dont remember slerp drive having a mode option

I dont really know if you'll be able to use those extensions. Theres really no reason you cant just set the motion to all locked, angular motion to limited in the inspector. id only really take parts of that code if you were dynamically creating them, but thatd be a real challenge

Honestly all I did was setup a copy of the character to run via animation, then my character copies the animators rotations only and it works quite well. You just want to make sure you calculate the rotation as an offset from the original animator rotation. The extension code somewhat has this formula already, but for the hips (or whatever you use as the center of mass), you'll have to adjust it to not copy the Y rotation

[tropic stag]

thanks, before manually replacing the joints you used unity's ragdoll helper right? do you remember what values you used for the "drives"? for spring and damping? (and if you used slerp or x + yz ?

[tall ferry]

i did use the ragdoll helper but that was purely just to get the colliders and rigidbodies premade. For the drives, you'll just need to really experiment. I set all my character weights to 1 so my drives arent even ideal/cant be used on a character where the weights are set to match real world weights. I might adjust it but definitely not rn.
Honestly I dont know if theres a major difference between slerp or x + yz, i dont notice any. X + yz just lets u separate the X out to have a different force, no clue why they dont separate yz as well.
Slerp applies the same drive to all 3 axis

[bleak citrus]

i keep getting really close to implementing a quadtree/octree

[bleak citrus]

but never actually doing it

[rotund tendon]

Hello, I am overriding the GameObject's transform position(because have a script simulating custom physics for it), but I want to be able to drag the gameobject in the editor, but since the position is overridden by my physics, trying to drag the gameobject with the move tool in the editor doesn't work, the object's position is locked.

Is there a way to get the event or value of the editor's move tool if the user drags and moves it by one of the axes? (so then I can add that to my physics object's position)
Thanks

[dusty wigeon]

I do not think there is. You could:

• Check if the position of the currently selected objects has changed since the last time you have seen it.
• Replicate the Physics.SyncTransform operation
• Replace Unity Standard Tool for your own tool.

[rotund tendon]

Thanks a lot, gonna try that 👍

[forest jacinth]

Is there a way to create a "slow motion" sphere that would.. put enemies in slow motion without affecting the entire game and the player? God of War had such an effect when dodging at the right timing with a certain item equipped.

[boreal jasper]

Procedural animation for a crab with Rigging animation & Inverse kinematic

[dusty wigeon]

Simply reduce the animator speed of the enemies when it come close to your Character.

[nova summit]

How can I apply a rotation in world space to a transform of a humanoid character? Is it possible to get a identity values for a transform in humanoid character so that I can apply the rotations easily?
I want to consider default idle position as the reference pose.

[dull kestrel]

I'd try to help, but I'm not understanding what you're trying to achieve

[nova summit]

Well this is what I have.

1. I know which points in object space where I can place each joint of a character
2. I calculated the angles for each bone so that I can apply at the joint transform of humanoid character
3. But the problem is the joint transform may have some offset and setting it to identity doesn't match the idle pose (consider idle pose is something like looking forward and hands, feed downward direction) - Consider this with hands down https://media.sketchfab.com/models/82f155b2abe142df9e0600ce3d006a80/thumbnails/4502258dace64d759f73d83fdb85f99f/757707edab6a44b3859087ae30e98cad.jpeg

[dusty wigeon]

Are you familiar with coordinate system ?

[regal olive]

If you are talking about slowing down the actual time of the game, but only for certain objects, I do not think that is possible. However all you need to do is have a point and distance / radius for the sphere. And the slow motion script loops through every enemy in that radius and changed certain speed and animation variables to make them look slow

[forest jacinth]

i had tried doing that, but all i could find was things that slows down the entire game. I just settled with spawning a sphere with 1 hp, that if killed, spawns a bigger sphere that activates an "EnemyFrozen" animation. just gotta add such an animation to every enemy though 😄

[nova summit]

Yes

[dusty wigeon]

Then you could make a conversion of coordinate system.

[mellow sail]

Use chronos asset

[forest jacinth]

a wut

[mellow sail]

its a plugin, library etc whatever you call. Basically you could control time flow in easier way without hard coding (using Unity,Time)

just look it up

you can make a sphere time zone too just like you wanted

its a built in method in chronos

Well the bad news you couldnt find it in unity asset anymore, but as jack sparrow said "A life of a pirate". You could still search it in web

[forest jacinth]

https://www.youtube.com/watch?v=-zq1OZuGXoc&ab_channel=Ludiq This i assume?

[mellow sail]

yup

[forest jacinth]

noice

[mellow sail]

its OP trust me

just put your head more in the documentation

https://alexiyous.itch.io/chronos-fox

I use it in one of my game, but still basic

[scenic forge]

@plucky laurel Continuing the conversation of quadtree from last time, if I'm understanding correctly, the idea is that for objects whose bounding boxes don't entirely contained in a node's children, they are added to the node itself instead.
That got me thinking about what you said about zooming, since in my case zooming is irrelevant, does that mean instead of constructing the quadtree top down, I could choose "1/2 view box width and height (or 1/4 view box area)" as the size of leaf, and construct the quadtree bottom up?
This has the issue that scene size may not be powers of 2 of the leaf size, which I don't believe is a required property for quadtree to work, right?
And lastly, is there a general rule of thumb for choosing the size of the leaf, assuming that most objects are: 1) way smaller than the view box size and 2) more or less evenly distributed throughout the scene?

[next sluice]

Guys i need help, how can i place a GameObject on a hexagonal grid via script?
Thank you in advance

[sly grove]

Use Unity's Grid component

which can be configured to use a hex grid

you can then get world space positions for grid coordinates with https://docs.unity3d.com/ScriptReference/Grid.GetCellCenterWorld.html

[next sluice]

oh, i was thinking something more like the GameObject brush

Thanks

[sly grove]

you asked for "via script" so I figured you meant in code

if you mean makijng an in-game brush at runtime, then what I said before is how you'd do it

[next sluice]

@sly grove so i have a slight issue, i have a normal hexagonal tile grid, and when i put this normal grid, it doesnt line up

[sly grove]

move the grid so it lines up

[next sluice]

thought it would be more complicated than that

[fickle sapphire]

I have an object with a character controller component that moves to a target position (represented as the red gizmo sphere). This object has a movement vector given in local space using animation curves to apply root motion during an animation, such as a dodge. I have purposefully made the object's position lag behind the target position to an extra extent to debug this.

My question is about rotation. I have a movement vector, represented in local space, but because we already have a target position, we want to move the object toward that target position, using the movement vector. If I didn't have a target position, I would simply just do the following:
movement = transform.rotation * movement;
This would convert the movement to world space and allow me to use it in the CharacterController.Move() method.

However, because I have a target position, I need to change the rotation to ignore the object's current rotation (it can change during the movement), and rotate the movement vector to reach the target position.

[dusty wigeon]

Not sure how the control should behave. You want to move your character as if the target position - character position is the forward ?

[fickle sapphire]

yes

basically that's it

[dusty wigeon]

This is the same principle as if you were using a ViewDirection. Not exactly sure what is the formula from the top of my head.

Something like Quaternion.LookRotation(target position.xz - character position.xz, Vector3.up) * movement.xz.

Maybe this video can help you understand

https://www.youtube.com/watch?v=7kGCrq1cJew

[misty glade]

I'm looking to do a drag and drop arrow thing in UI space, and sorta stumped with some angle/quaternion stuff.

I've implemented IPointerDown, Up, and Drag handlers, and I instantiate my prefab using the eventData.position at the start. While dragging, my intent was to update the arrow with a method called SetDrag(Vector2 from, Vector2 to). When I try to calculate the Quaternion using Quaternion.FromToRotation, I'm getting odd values. Do I need to cast my Vector2s to Vector3s manually?

(without any rotation currently)

[dusty wigeon]

Quaternion.FromToRotation use 3D vector.

Your vector is already been cast

[fresh salmon]

Can you try with Quaternion.LookRotation(to - from) instead? This rotates on Z so you might need to pass a Vector3.forward as argument 2

[misty glade]

Sure, standby

[fickle sapphire]

I've tried this. You can take a look at the video to see what happens

movement = Quaternion.LookRotation(networkTransform.currentPosition - networkTransform.transform.position, Vector3.up) * movement;

[misty glade]

Oh, wait, whose question were you answering?

[fresh salmon]

Yours

[dusty wigeon]

You need to make things flat also.

[misty glade]

two Quaternion questions at the same time :p

[fickle sapphire]

lmao

[fickle sapphire]

kk will try

[dusty wigeon]

Remove the y value

[fickle sapphire]

yeah I gotcha

[fresh salmon]

If that doesn't work, then it depends on how the arrow is set up, you might have more luck setting its .right or .up directly to the direction vector, so it does the rotation job for you

[misty glade]

Hm, closer. I can probably use this but I'm not understanding the output. In terms of "clock positions" - 3 o'clock (oc) is 0 degrees, noon is 90 degrees, 9oc goes back down to 0 degrees, but 6oc is 270

so the only values i get are 0-90 and 270-360

I mean maybe this is fine.. lemme just slap it on the gameobject and yolo-see what happens

yolo yo-came and yo-went without effect

[misty glade]

I'll try this

When you say set the .right or .up directly - do you mean edit the prefab so it's facing the correct way? ie, using the uh.. red arrow (I think) for .right?

(I probably ought to flip the asset around as well - it currently points left by default so I have to flip it manually)

[fresh salmon]

Nah that's fine, the arrow is pointing left but no problem, you can still set its transform.right to the reversed direction vector instead

Since transform.left isn't a thing

[fickle sapphire]

Same problem here, I did try this in the past and it didn't seem to work

Quaternion targetRotation = Quaternion.LookRotation(new Vector3(networkTransform.currentPosition.x, 0,                              networkTransform.currentPosition.z) - new Vector3(networkTransform.transform.position.x, 0, networkTransform.transform.position.z), Vector3.up);

horizontalMovement = targetRotation * horizontalMovement;

movement.x = horizontalMovement.x;
movement.z = horizontalMovement.z;```

[misty glade]

close enough for government work

i'll just manually figure out the flippy flippy

quaternions make my brain hurt

[fickle sapphire]

dw they made my brain hurt when I first started too

well actually they still make my brain hurt

[fresh salmon]

Yeah it's inverted, just have to slap a * -1 somewhere

[dull kestrel]

To me you just wanna get the mouses world position and then set the transform.right to that direction and skip all the rotations (if I understand what you're trying)

[misty glade]

almost, a -180 somewhere

i'm just experimenting because i'm too dumb to math it out

[fresh salmon]

The direction you're constructing from to and from, does it work if you reverse them in the subtraction?

(a - b instead of b - a)

Hm wait no, it's not entirely reversed, it's mirrored on the Y axis

Saw the video again

And then mirrored on X at some point? Wtf is going on here

[dusty wigeon]

@fickle sapphire Is it what you want ?

public class Movement : MonoBehaviour
{
    [SerializeField] private Transform target = null;
    [SerializeField] private float speed = 0.0f;

    private void Update()
    {
        float vertical = Input.GetAxisRaw("Vertical");
        float horizontal = Input.GetAxisRaw("Horizontal");

        Vector3 direction = target.position - transform.position;
        direction.y = 0;
        direction.Normalize();

        transform.position += Quaternion.LookRotation(direction) * new Vector3(horizontal, 0, vertical) * Time.deltaTime * speed;
    }
}

[misty glade]

I know, right? 😛 I'll get it, eventually

wait that's wrong, standby

[sand acorn]

Hello there, I assume this is sort of a newbie question, I have a parallel job where I am doing some calculations and right now I am refactoring my code to modify a struct value; The problem right now is that I have some conditionals on my job and in those I want to modify a value inside my struct at the current index, but for some reason the struct does not change at the specified index unless I call the function outside of those conditionals, what am I doing wrong? I assume the problem relies on the index differences, like blockBack can happen at index 0, 20, 24 and so forth

 //back
            if (blockBack == 0)
            {
             //Does not work
                AssignData(index, blockPos, BackFace);
            }

            //front
            if (blockFront == 0)
            {
                 //Does not work
                AssignData(index, blockPos, FrontFace);
            }

//Does work outside of conditions
 AssignData(index, blockPos, BackFace);

  private void AssignData (int index, half3 blockPos, NativeArray<half3> face)
    {
        _VertexData[index] = new VertexData
        {
            Position = AddQuad(blockPos, face),
            index = index,
            //  Normal = _VertexData[index].Normal,
            //  Uv = velocityPosition.Uv,
            // Tangent = velocityPosition.Tangent
        };
    }

I have used debug log on the conditions and they do happen

This is the AddQuad function if it helps

 private half3 AddQuad (half3 blockPosition, NativeArray<half3> vertexPositions)
    {
        half3 outValue = half3.zero;
        for (ushort i = 0; i < vertexPositions.Length; i++)
        {
            half3 vert = vertexPositions[i];
            outValue = new half3(math.half(vert.x + blockPosition.x), math.half(vert.y + blockPosition.y), math.half(vert.z + blockPosition.z));
        }

        return outValue;
    }

I highly doubt that function fails because it does log a result

[dusty wigeon]

Is it what you want ?

public class Rotate : MonoBehaviour
{
    [SerializeField] private Transform mouse;

    private void Update()
    {
        Vector3 direction = mouse.transform.position - this.transform.position;
        direction.y = 0;

        transform.right = direction;
    }
}

I guess we can simply though

[fickle sapphire]

I'm not sure what is happening here

Which cube is the target?

[dusty wigeon]

There is a cube turning around an other

When it does a circle, it is when I press left or right arrow.

[fickle sapphire]

right so, that seems fine. However, my problem is that when I apply the rotation that would usually work (I've done stuff like this before) it throws off all the movements that aren't straight forward

Ok so let's say this:

[dusty wigeon]

?

[fickle sapphire]

it's not what I'm looking for

Let me give an example

Player object's position is 0,0,0. Target position is 10,0,0. Movement vector provided by the animation's root motion is 1,0,0.

When the player's rotation is Quaternion.identity, the local axis lines up with the world axis, so no problems here

however let's say the player is rotated 90 degrees

now the target position will be 0,0,-10 and the movement vector will remain the same

For some reason, rotating by the target position - player's current position isn't working

[misty glade]

Sorry had to step away @fresh salmon , but yeah, that was much simpler and worked without having to touch the quats

i'll flip the text around later.. have to step out

[fresh salmon]

Nice! Looks great

[misty glade]

Yeah, I like the effect.. will do some mouseover-hex-highlighting and minimum drag distance and stuff, but .. pretty much works like i wanted it to

[fickle sapphire]

if I just rotate it using the player's current rotation, that delivers the desired result, however, the player will not reach the target destination properly (see original video)

see on the second dodge, how when I rotate the player at the end, because the target position has already been calculated, the movement vectors go out of sync

this is done using movement = transform.rotation * movement;

[dusty wigeon]

Transform.TransformVector ?

[fickle sapphire]

yes that's the same thing

or TransformDirection I thnk

basically it converts the movement vector from local space to world space

because this is being applied using the character controller's move() method

which takes a world space vector

now if I use movement = Quaternion.LookRotation(targetPosition - transform.position) * movement;

this is the result

That happens regardless of including or excluding Y values

I am kerfuffled

[dusty wigeon]

public class Movement : MonoBehaviour
{
    [SerializeField] private Transform target = null;
    [SerializeField] private float speed = 0.0f;

    private void Update()
    {
        float vertical = Input.GetAxisRaw("Vertical");
        float horizontal = Input.GetAxisRaw("Horizontal");

        Vector3 direction = target.position - transform.position;
        direction.y = 0;
        direction.Normalize();

        Quaternion viewDirection = Quaternion.LookRotation(direction);

        Vector3 movement = viewDirection * new Vector3(horizontal, 0, vertical) * Time.deltaTime * speed;

        if (Input.GetKeyDown(KeyCode.Space))
        {
            movement += viewDirection * Vector3.forward;
        }

        transform.position += movement;
    }
}

I do not understand why it is not working.

From my side, this work.

Vector3.forward would be the animation movement.

[fickle sapphire]

the animation movement has horizontal movement as well

like the clip I keep showing is the player dodging to the right, which maintains looking forward on the world axis

[dusty wigeon]

forward can be right, left, down

Whatever you want.

[fickle sapphire]

yeah ik, idk why it's not working either, I've been working on this since 8 am this morning lol

[dusty wigeon]

public class Movement : MonoBehaviour
{
    [SerializeField] private Transform target = null;
    [SerializeField] private float speed = 0.0f;

    private void Update()
    {
        float vertical = Input.GetAxisRaw("Vertical");
        float horizontal = Input.GetAxisRaw("Horizontal");

        Vector3 direction = target.position - transform.position;
        direction.y = 0;
        direction.Normalize();

        Quaternion viewDirection = Quaternion.LookRotation(direction);

        Vector3 movement = viewDirection * new Vector3(horizontal, 0, vertical) * Time.deltaTime * speed;

        if (Input.GetKeyDown(KeyCode.Space))
        {
            Vector3 animatorMovement = Vector3.right;
            movement += viewDirection * animatorMovement;
        }

        transform.position += movement;
    }
}

[dusty wigeon]

The only answer is that you are not doing something correctly.

[fickle sapphire]

ik

[dusty wigeon]

Try it yourself.

My code.

I took the time to open Unity and Test it. -_-

[fickle sapphire]

I'm working on it rn

[fickle sapphire]

same result

                    direction.y = 0;
                    direction.Normalize();

                    Quaternion viewDirection = Quaternion.LookRotation(direction);

                    movement = viewDirection * movement;```

[dusty wigeon]

Use only my script. Make a new scene if needed

Adapted for Root motion. Could not test it though.

public class Movement : MonoBehaviour
{
    [SerializeField] private Transform target = null;
    [SerializeField] private float speed = 0.0f;
    [SerializeField] private Animator animator = null;

    private Vector3 animatorMovement = Vector3.zero;

    private void Update()
    {
        float vertical = Input.GetAxisRaw("Vertical");
        float horizontal = Input.GetAxisRaw("Horizontal");

        Vector3 direction = target.position - transform.position;
        direction.y = 0;
        direction.Normalize();

        Quaternion viewDirection = Quaternion.LookRotation(direction);

        Vector3 movement = viewDirection * new Vector3(horizontal, 0, vertical) * Time.deltaTime * speed;

        movement += viewDirection * animatorMovement;
        animatorMovement = Vector3.zero;

        transform.position += movement;
    }

    private void OnAnimatorMove()
    {
        animatorMovement += animator.deltaPosition;
    }
}

[fickle sapphire]

you're good, the movement is working properly only for dodges that are going forward, all other directions are having problems

[dusty wigeon]

What is the problem ?

Because it is working on the side too.