#archived-code-advanced

1 messages · Page 78 of 1

copper venture
#

Anyone here actually understand projection matrices? The net only seems to have info about the "default" values.
Tweaking with the variables by hand I can get translation, aspect ratio and a bunch of other random deformations, but what I need is to be able to change perspective/scaling to make it look like the scene changed scale - and I can't find what formula I need to apply to the matrix for that effect.

dusty wigeon
primal bear
#

Hello all. Any method to execute Update method when my iPhone or Android turn to background mode?

dusty wigeon
primal bear
dusty wigeon
#

Both

primal bear
dusty wigeon
#

Also, the library probably have some example inside it.

#

Never developped on mobile, but Switch, Xbox and PS works that way.

#

So, I expect the same procedure.

primal bear
dusty wigeon
#

Also, I'm not exactly sure or not it is possible to execute code in background. However in both platform there is event that are being trigger whenever the console is suspended.

primal bear
#

what does represent console you said? Unity?

dusty wigeon
#

Both are widly different

primal bear
copper venture
dusty wigeon
#

Also, I expect that you would need to scale colliders and other things as well. Not just the mesh

copper venture
#

It turn a 3D point into a 2D point depending on distance. My guess is that there must some maths that allow to display the results as if the scene has been scaled up/down.
And no, I specificly don't want to scale the actual scene - that's what I'm doing right now and it's an absolute mess, if only because it prevent any mesh from being static, break all light probes and anything distance-based (light source, among other things).
That's why I'm trying to keep the scene as is, but deform what's rendered instead.

dusty wigeon
#

In truth, what you want is add an other matrices to define the position of the vertices for that you would need to write a shader.

It might be possible to concat both matrices in one, however I wouldnt do that as other shader might needs the perspective matrix as is.

frosty patrol
#

weird question
but is it possible to open a native linux terminal because I wanna debug log a huge array, doing so in unity just halts the application

dusty wigeon
#

I'm not sure what is link between a linux terminal and debuging unity

sly grove
#

how about writing the data to a file?

frosty patrol
#

unity halts and I am not able to scroll up

#

in the debug panel

sly grove
#

or use the debugger in your IDE

frosty patrol
sly grove
#

no

#

because Unity will be busy writing to the command line

frosty patrol
sly grove
#

VSCode is trash for Unity and its debugger is not supported

#

Rider works

#

but yeah I'd just write the array to a file

#

i don't see how printing it to stdout would be better than that

frosty patrol
sly grove
#

who said anything about windows?

#

I didn't say windows once

frosty patrol
frosty patrol
sly grove
#

DId I say Visual Studio?

#

I recommended Rider

sly grove
#

you can even read the file in a terminal if you so pleased

frosty patrol
steel snow
#

for this bit of code im getting an error that i am not sure how to fix:

                if (info.Offsets(out var offset))
                {
                    using (var data = info.MeshData)
                    {
                        Mesh.MeshData d = data[0];
                        var verts = d.GetVertexData<float3>();

                        //job
                        verts.Translate(offset);
                        verts.Dispose();

                        Mesh mesh = new();
                        Mesh.ApplyAndDisposeWritableMeshData(data, mesh);
                        combineInstances.Add(new() { mesh = mesh });
                    }
                }

Error: ObjectDisposedException: The UNKNOWN_OBJECT_TYPE has been deallocated, it is not allowed to access it

#

it doesn't take me to a specific line either

#

just to the find curly brace

#

i put dipose last and it still gives the same error

upbeat path
steel snow
#

it seems im not meant to dispose at all unity's example doesn't

#

i hate this new mesh api

misty ibex
#

In netcode for game objects in a client authoritative model does the server still have permission to move the client as well? or is it now locked to the client moving themselves

fickle tinsel
tall ferry
#

thats only if its intended to be a game between random people, client authority has its uses otherwise

tall ferry
misty ibex
#

i appreciate the responses btw!

lament dagger
#

Hey yall, for the past month ive been trying to figure out how to make a modular ability system. Every now and then on the asset store I see people using advanced dropdowns in order to add to a list of functions and edit their parameters in the editor. I've been trying to reverse engineer that sort of functionality to no avail so I just wanted to ask if anyone knows what they're doing or if I'm even in the right ballpark. I can send screenshots of what I mean if nessessary.

chilly nymph
lament dagger
#

I do have odin already but it doesnt solve what I wanna do

#

it is really nice tho

untold moth
chilly nymph
lament dagger
#

the ones I see look like this

image.png image.png
chilly nymph
lament dagger
#

can you expand on what parts of odin I can look into to make it? Bc when I went through the documentation and experimented myself I couldnt do it

#

I did only just recently buy it so Im kinda new to it

untold moth
# lament dagger the ones I see look like this

Yep, what I said previously.
The "requirements" and the other addable "components" are probably a subset of classes that is retrieved with reflection, then it's just a matter of using their names in a custom editor.

chilly nymph
#

Odin would serve as the reference hookup and handler of your abilities.

#

While reflection in this case would be okay, I wouldn't reccomend making a complex system, all of these can just follow a basic ability structure that is in text / ID format.

Any project my team and I have done reflection with cost us more time than just hardcoding references.

lament dagger
ancient talon
untold moth
#

I imagine that's not what they want.

lament dagger
#

yea i meant like muliple classes like the screenshot I sent earlier

untold moth
stuck plinth
#

you can use odin and [SerializeReference] to do something like this, odin will handle displaying, instantiating and editing different types

jolly token
#

Or the gross SerializeReference

#

Oh

lament dagger
#

💀

jolly token
#

Well it’s basically saving your qualified typename as string

#

So you break everything if you rename 😄

lament dagger
#

ah

#

also by any chance do any of you know how to use advance dropdowns with odin?

austere jewel
jolly token
#

Yeah but that attribute is horrible 😭

#

I guess you can put that and re-serialize everything then remove it

#

Still horrible workflow

austere jewel
#

I tend to use Rider's find and replace most of the time

jolly token
#

That’s probably better

gritty delta
#

Hello guys, I need you help resolving this error. I updated Unity from 2022.3.7f1 to 2022.3.8f1 today. Now I am receiving this error? Does anyone know how to resolve this?

image.png
sly grove
gritty delta
#

Thanks I'll give it a try.

gritty delta
untold moth
primal bear
#

Can I develop iOS app with writing native code even though I develop Unity app?

raw lily
outer barn
#

I have two cubes one on top of another , i want the cube thats below to change color as the top cube moves over it , which approach should i use

primal bear
primal bear
raw lily
raw lily
#

nvm

primal bear
shadow vigil
#

If FixedUpdate() takes 3000 ms to compute, the engine would still batch FixedUpdate calls in a futile attempt to stay consistent with Time.fixedDeltaTime value? In my tests Unity produces around 16 FixedUpdates per 1 Update at this load. And I don't understand why it's not 1:1 since the computational time exceed both deltas.

#

How Unity handles FixedUpdate() once it exceeds Time.fixedDeltaTime to execute?

sly grove
shadow vigil
sly grove
#

Unity promises Time.maximumDeltaTime / Time.fixedDeltaTime number of FixedUpdate() calls before calling Update() even once and showing anything to the user
There can be anywhere from 0 to Time.maximumDeltaTime / Time.fixedDeltaTime number of FixedUpdate calls in any given frame (per Update call)

shadow vigil
#

Sorry, it was within the context of an absurd computational load within the FixedUpdate(). In my tests a number of FixedUpdate calls exactly matches this ratio per Update. But with no load it sure would be different.

sly grove
#

yeah that's a maximum

#

most of the time the framerate is much higher than the physics framerate

#

so most frames have 0 FixedUpdates

shadow vigil
# sly grove FixedUpdate is guaranteed to run at the cadence of the physics simulation, there...

My understanding is that calling a query like Physics.Raycast during Update() would reduce the load on FixedUpdate(). If a query doesn't result in a manipulation with physics objects, it will result in "better stuttering".
I often see a recommendations to do all physics-related calculations in FixedUpdate(). But as I've just experienced, with a heavy load on FixedUpdate() a user will see complete freezes between Updates(). With a heavy load on Update() there is just stuttering due to fast cycle through FixedUpdate() calls.

strange bane
#

Hello 👋

We are working on a VR game and we are trying to see how we can customize the behaviour of IL2CPP compiler (clang as far as I know) to gain anything positive (frame rate, build size reduction etc.) on our debug/development builds.

In our build pipeline, we are passing the additional arguments by using this API: https://docs.unity3d.com/ScriptReference/PlayerSettings.SetAdditionalIl2CppArgs.html

I wonder if there is a list that tells every single argument that we can pass to the IL2CPP compiler during our builds. I've been looking around for a while but I didn't find anything.

Any helps is greatly appreciated.

merry rose
#

Hey I see you were talking about SerializeReference.
So, basically, I just overused it in my project cause it makes a lot of stuff easy and fast to do.
BUT apparently, when you create an instance of one of those "reference" in an array, then remove it from the array, the reference still exist in the scene even though nothing access/use it. I check in the scene file, and there is, indeed, all my previous references scattered everywhere. Wtf should I do, is there a way to cleanup those instances?

sage radish
merry rose
#

2022.3.4f1

dusty wigeon
shadow vigil
dusty wigeon
# shadow vigil How?

I'm not sure how to explain.

If you do a raycast base on a position that is being update every frame, then you might miss your target.

#

Also, if your FixedUpdate happens more than your Update, you might want to reduce it to 30 frame per seconds instead of 50 frame.

#

And you can also manually simulate the update instead of using the fixedupdate.

#

With Physics.Simulate.

#

I wouldnt recommand doing that though.

#

Because if your FPS is too low, the whole game is going to be unstable.

#

Instead, you should focus effort in reducing the load of your game. Do not try to find work around.

shadow vigil
#

You might be referring to an overpenetration that occurs once an object moves an unexpected amount of space during Update. Which implies that the movement isn't well calculated and framerate-dependent.

dusty wigeon
shadow vigil
#

Now I'm curious.

dusty wigeon
#

Frame independent will just increase the distance base on the time that the frame took.

#

Meaning, that the object travel more distance (In 1 frame).

shadow vigil
#

Isn't this also includes a precomputed distance during FixedUpdate(), so the system have to only interpolate object between predefined points?

dusty wigeon
#

If you want reliably detect collision, you need to do the raycast at every distance.

dusty wigeon
shadow vigil
#

What I've read about frame-independent movement so fare, is that it still kind frame-dependent due to the necessity of Update() calls. However, if a vector is calculated during FixedUpdate(), it then can be reliably lerped each Update() even if there are lag spikes.

dusty wigeon
#

If we take an absurd example
Given an object with a speed of 10m/s that is a position X

Frame 1 (Process the frame for 1 second): Obj at position X
-- Raycast from position X--
Frame 2: Obj at position X + 10m
--Raycast from position X + 10m--

#

If your target object is less than 10m in size, you gonna have a chance to miss it.

#

With FixedUpdate you have:

#

A raycast by FixedUpdate. If your FixedUpdate is 0.5s, then you gonna have a raycast from position Obj at position X + 5m and one from Obj at position X + 10m

dusty wigeon
shadow vigil
#

In this function dt is still Time.deltaTime. If Update() takes 1500 ms (an extreme example), then on the next frame an object will leap forward to account for the lag, right?

dusty wigeon
#

In the case of raycasting, you want to raycast precisely at each given time to reduce the change of missing your target due to the object being move a large distance.

steady tree
#

Hey guys, for example I need to set a animator transition time, due to reconciling on a tick perfect, side predicted client. Have someone an idea how to do that?

I'm already syncing animation states and their normalized time from server to client. But when reconciling while a transition happens the transitions "restarts", so I need to set the normalized time in the transition itself.

chilly nymph
dusty wigeon
#

Also, pretty sure you do not want to sync animation between client and server. The animation should be pretty much only client side and the server should not even know there is an animator.

#

At least, as far as I know.

#

I'm not an expert in networking

steady tree
dusty wigeon
#

In fact, you should reconcile that instead of the animation.

chilly nymph
steel snow
#

you can eliminate hardware by way of deterministic setup and you can also do lockstep to not worry about timing which RTS typically do

#

if you got run time procedural animations you would need to sync them

dusty wigeon
steel snow
#

if its procedural you do because you will get drift issues

#

even more so if its related to user input tracking

dusty wigeon
#

You are talking about long running animation

steel snow
#

im talking about procedural animations

dusty wigeon
#

That are procudural

steel snow
#

duration not really relevant

dusty wigeon
#

You can have short procedural animation

#

Which is the norm

#

It matters

steel snow
#

no it doesnt if you want things in sync you will have to sync the animations

dusty wigeon
#

Short animation wont drift so far that it will start to matter

steel snow
#

it will

dusty wigeon
#

You do not have to sync things that are only aesthetic.

steel snow
#

thats like saying a small bit of drift in physics wont matter

#

not all animation is aesthetic

#

animation matters in fps if you want accurate hit boxes for example

dusty wigeon
steel snow
#

lol

dusty wigeon
steel snow
#

not true at all but i cba to go into the complexity and nuances of it all

dusty wigeon
#

Why would you want hitbox client side.

steel snow
#

you still need them sync

#

so players see your arm or w.e where it needs to be

dusty wigeon
#

Why you need to have hitbox client side.

steel snow
#

🤦‍♂️

dusty wigeon
#

Answer

#

Why you need to have hitbox client side.

steel snow
#

if they arent in sync your not seeing the player where they really are

#

which really matters in an fps

dusty wigeon
#

Hit box are not visible

steel snow
#

the character's animation im talking about

#

you still need to sync the animatiosn to the hit boxes

#

thus animations need to be in sync

#

also some times the player hosts the fps game not all fps games need to be server side

dusty wigeon
#

First, you do not need hit box in client side in a server authorative.
Second, animation will not drift so far that it won't be reliable for the player.

dusty wigeon
steel snow
#

the server still needs to follow the animations from the client

dusty wigeon
#

Sync the Animator is correct.

#

But that animation is way overkill

steel snow
#

the animator is for predefined animations

#

not run time procedural stuff made in code

dusty wigeon
#

The animator define the transition between animation

#

What animation you are doing procuderaly

#

That needs to be sync.

steel snow
#

it depends on the game

dusty wigeon
#

It does not.

steel snow
#

its not hard to understand it really depends on the game

dusty wigeon
#

Procudural Animation is not really popular.

steel snow
#

that doesnt make my point any less valid

#

how popular something is . does not have any point in this topic

dusty wigeon
#

Fine, if you want to include useless information.

steel snow
#

its not useless information they were asking about animation syncing

#

stop making an argument for no reason i have better things to do

dusty wigeon
#

Yes, it really matters. Because if you were doing long procedural animation I would agree.

#

That it can matter. But for short animation, either procedural or not.

#

It does not matter because the server is the autorithy and animation is purely aesthetic.

steel snow
#

procedural animation needs to be in sync if you want two players to see the same thing

#

or they might see different results due to float differences on their pcs

#

this is also the case for physics

#

physics drifts are even more noticeable

dusty wigeon
steel snow
#

no i said procedural

dusty wigeon
steel snow
#

christ almighty

dusty wigeon
#

I already said I agree on that

steel snow
#

so why did you argue it then ?

dusty wigeon
#

Why are you still talking about long duration procedural animation

steel snow
#

i never mentioned anything about long or short

#

i said procedural

#

i made no mention of duration

dusty wigeon
steel snow
#

where in this sentence do i mention anything about how long they are?

image.png
#

anything run time you will need to sync they are not using the animator

dusty wigeon
#

Funny, I got the context and the meaning of the first sentence by the rest

steel snow
#

animator is for pre-made animations

#

pre-made will have the data available for the clients

#

so its easier to sync

dusty wigeon
#

Yes, and this is what people are going to use for almost any situation.

steel snow
#

procedural is done at run time so two clients might not be in sync. thus procedural animation needs to be sync

dusty wigeon
#

And in this situation, that even the original post was about, sync is not something you want to do.

steel snow
#

ok but i gave example of when you would need to sync the animation

dusty wigeon
#

Where

steel snow
#

i said procedural animations liek 5 times

#

bro stop trolling

#

go away seriously

dusty wigeon
#

That is not an example...

steel snow
#

im not here to educate you on procedural animation youre derailing the room

dusty wigeon
#

Are you seriously saying go away...

steel snow
#

if you dont know about run time procedrual stuff go read about it

#

you cant use them in the animator

dusty wigeon
#

You know, you can at anytime drop out.

steel snow
#

im not here to give endless examples

dusty wigeon
#

1 example that matter

#

Is only what I ask

steel snow
#

this room is for answering people's issues on code - i corrected some one saying you dont need to sync animations - and that isn't always true by mentioning procedural animation

#

go google it if you want to know

dusty wigeon
#

Alright, lets keep it there.

steel snow
#

good

chilly nymph
#

Regardless if it's procedural or not, you shouldn't be moving your character with procedural animations for a networked game, this is basic networking 101.

steel snow
#

not all animation relates to a character

chilly nymph
#

In this case it was.

#

I don't know why you're trying to argue.

steel snow
#

the case was me stateing you never need to sync animations

chilly nymph
#

I said you shouldn't

steel snow
#

because someone said that which isn't true

chilly nymph
#

You need to learn to read.

#

You seem to have a issue both reading and injecting your own arguments.

steel snow
#

You also don't sync animations,

#

was what you said

#

there are situations where you do

chilly nymph
#

You should not be controlling your character through procedural animation in a multiplayer game.

steel snow
#

so ? i still made the statement of when there is times to sync animation so i can read but thanks for your snarky comment all the same

steel snow
chilly nymph
steel snow
#

VR games need to track player inputs, facial mapping stuff, eye trackers

chilly nymph
#

You're just backpeddling now.

steel snow
#

non of this can be done in the animator because its run time animations

chilly nymph
steel snow
#

im not back tracking i gave an example of where you do need to sync animations

#

thats still syncing the animation

chilly nymph
steel snow
#

🤦‍♂️

chilly nymph
#

It's transforms.

steel snow
#

if it moves its animation

chilly nymph
chilly nymph
steel snow
#

youre confusing animation clips with the general idea of animation

chilly nymph
steel snow
#

i had no confusion with it when i had to do it for a VR project

#

so

#

¯_(ツ)_/¯

#

you do you

chilly nymph
#

Your entire argument was about drift in player controls, you solve this by Not using animations to control synced clients

#

This is not a difficult thing to understand.

steel snow
#

no my argument was there are situations where you need to sync animations i made no statement on player character

chilly nymph
#

Now you're trying to argue about VR

steel snow
#

i merely made a tidbit of info

chilly nymph
steel snow
#

drift is accumulation of differences between clients

#

physics is a good example of drift and you need to keep it in sync all the time

chilly nymph
#

No one here ever said don't sync animations, 2 people have said don't control your character or VR or anything else with animations.

chilly nymph
steel snow
#

you said "you dont sync animations" i merely replied saying sometimes you do

chilly nymph
#

NOT each client like OP was mentioning.

steel snow
#

he wanted 100% prediction so the client would have to do sync via determistic architecture

#

so the OP isn't wrong

chilly nymph
#

You call the animation and then update the transforms,

steel snow
#

for animation clips sure just send when the animation was triggered etc

#

but im not talking about premade animations

chilly nymph
chilly nymph
steel snow
#

but he wanted a 100% prediction so he would need to use a setup where the clients keep themselves in sync

chilly nymph
steel snow
#

irrelevant

#

you can get 100% prediction if you know anything about DOTS

chilly nymph
#

It is, just because he wants it doesn't mean he's going to oget it.

steel snow
#

he can get it

#

if he uses DOTs correctly

chilly nymph
steel snow
#

unity guarenteed deterministic outcomes across architecture

#

there for he can

#

if you're not aware of this you should read up on it

chilly nymph
#

You should read up on latency, I hear it's a bitch.

steel snow
#

it was one of the hallmarks of why unity wants to offer "rewind" feature in networking

#

then you should hear how its dealt with since the days of age of empires

#

RTS games require deterministic networking or its impossible to sync so many units

#

it would be painful

#

the old school way was a fixed point numbers

#

but they hurt performance

chilly nymph
#

That's not using your main argument of Procedrual animations that's using a set of events that are synced between clients. Our entire argument with you has been about that.

steel snow
#

oh were going back to the old argument now?

#

thought we were on about his wish for 100% prediction

#

which argument are you trying to have with me here?

chilly nymph
#

No, I need to get back to work, best of luck.

steel snow
#

me too

white pulsar
#

a bit related to something i found on #1064581837055348857, but how many intergers can i fit inside 4096 bytes?

#

mainly cuz of this bit, this fixed list which has a memory allocation of 4096 bytes

36010e06c423a31a41859205df48cfe9.png
steel snow
#

i presume 1024 ints then ?

dusty wigeon
#

So, that would be 1024 if I am right.

white pulsar
#

wow

#

thats a shit ton lol

#

maybe 4096 is a bit overkill...? idk

#

with the naming you can probably deduce that this is for a navigation system with A* (not using unity's navmesh solution due to its limitations)

#

still, thanks, i'll keep the Interger == 4 Bytes bit in mind for the future

dusty wigeon
#

Always get me.

white pulsar
#

likely

#

its rare that i have to deal with such low level stuff like this

# 
if(nodeA.serializedPathNodeLinkIndices.Count >= 1024)
{
    Debug.LogWarning($"PathNode of index {nodeAIndex} has reached it's max capacity of 1024 nodes, cannot add more links.");
    continue;
}

i think this should do the trick unless i got the logic wrong, thanks guys

sly grove
white pulsar
#

thats... neat

scenic forge
#

@white pulsar You can't store 1024 ints in FixedList4096Bytes<int>, because 4096 bytes is the size of the entire struct, which 2 of those bytes are already used by the struct presumably for its count.

white pulsar
#

Oh

scenic forge
white pulsar
#

I'll take a look into it tomorrow

#

Thanks

scenic forge
#

Depends on what you need, UnsafeList<T> is an option as well.

#

Or make your own fixed size container.

tiny pewter
#

Fixed-size buffers in structs
but not friendly, you cannot have other unmanaged struct type other than primitive type, though there is a way to solve this problem

lapis bobcat
#

Hi, someone knows that its possible to get the OS voice language in Unity? I was researching and I only found Application.systemLanguage, but that doesnt get the speech language. Thanks

real blaze
#

any idea how this ambiguity might be caused?

image.png
# 
Assets\Plugins\UnitySocketIO\WebSocketSharp\Net\WebSockets\TcpListenerWebSocketContext.cs(185,37): error CS0121: The call is ambiguous between the following methods or properties: 'WebSocketSharp.Ext.IsLocal(System.Net.IPAddress)' and 'WebSocketSharp.Ext.IsLocal(System.Net.IPAddress)'
#

there's no duplicate of that definition in my project folder

real blaze
#

nevermind gave up

frosty patrol
#

what does GCHandle.Alloc actually do in the background ?

Allocates a handle for the specified object.

``` from the docs
<https://learn.microsoft.com/en-us/dotnet/api/system.runtime.interopservices.gchandle.alloc?view=net-7.0>
#

what does allocates a handle mean here ?

flint sage
#

This handle type represents an opaque handle, meaning you cannot resolve the address of the pinned object through the handle. You can use this type to track an object and prevent its collection by the garbage collector. This enumeration member is useful when an unmanaged client holds the only reference, which is undetectable from the garbage collector, to a managed object.

#

It's a way to tell the GC to not collect a specific object until you tell it to

frosty patrol
#

ahh the word alloc is missleading then
I thought I was doing unnecessary allocations

languid scarab
#

I'm attempting to use Timeline in a multi-scene context, where the Cinemachine Brain is on the main cam in a different scene than the Playable Director / timeline object.

This is challenging as Timeline doesn't seem to support multi-scene workflows.

Searchin the web for solutions I've come across this, but not being a pro-coder, I'm having a hard time actually getting the variables stored across scenes. I could really need some help.

https://github.com/starikcetin/Eflatun.SceneReference

hardy jacinth
#

how do I add source generators to existing unity project on linux?

#

ideally using roslyn

scenic forge
dusty wigeon
#

It does seem to be closer to what you need.

regal olive
#

i think ive might have found a bug, i have this button that i use for interacting when im on mobile but of some reason it doesnt want to interact with one of the two gameobjects, so ive messed around for about 4 hours now and i found out that if i change the buttons transform in any way and then CTRL-Z (the important part i think), it makes it works but why? And how do i fix this so i dont have to press ctrl-z everytime i play the game?

hardy jacinth
# scenic forge <https://docs.unity3d.com/Manual/roslyn-analyzers.html>

no, yeah, I read that, but I am completely oblivious to how csprojs, solutions, nuget packages etc wok in unity. I tried creating project with dotnet new sln and dotnet new classlib and installing the Microsoft.CodeAnalysis in the Rider window, but it doens't do much. ISourceGenerator is not found, Generator attribute as well, I don't know how to do that

scenic forge
#

Are you writing your own source generator, or using one?

hardy jacinth
#

I want to write my own which finds classes annotated with my custom attribute and generate appropriate partials

scenic forge
#

I don't use Rider so not sure about them, but I've heard their UI is practically the same as VS's.

#

And I assume by adding the package Rider should do restore for you automatically.

#

What does your csproj look like?

hardy jacinth
# 
<Project Sdk="Microsoft.NET.Sdk">

  <PropertyGroup>
    <!--    https://stackoverflow.com/questions/46621164/how-do-i-target-net-standard-2-0-in-a-freshly-created-net-core-2-0-web-app -->
    <TargetFramework>netstandard2.0</TargetFramework> 
    <ImplicitUsings>enable</ImplicitUsings>
    <Nullable>enable</Nullable>
  </PropertyGroup>

</Project>
#

also this is inside my <unity project root>/Packages/<my package root>/SourceGen

#

because I am writing a package

scenic forge
#

Well you haven't added the code analysis package.

hardy jacinth
#

oh, I assumed they are added somewhere in the obj directory in one of the *.nuget.* files

stuck plinth
#

make sure your analyzer's source files aren't being included in your unity project, i think you just want to compile the dll separately in rider and drop that into your project?

cerulean wasp
#

to make it work, you need to give the dll the RoslynAnalyzer tag in unity, so it knows to use it

#

asset label I think it is called

hardy jacinth
#

actually what I want is add the support for that in the output package, so that someone using the package can extend with his own classes, annotate them, and the source generation will kick off

#

but that may come later, for now I need source code generation for myself

cerulean wasp
#

i got it going last week. it is suck

#

i basically apply the asset label whenever I want to turn it on/off

#

and I need to manually drag the DLL every time I edit the source generator

hardy jacinth
#

does csproj detect transitive dependencies or something? for example if Microsoft.CodeAnalysis.CSharp depends on Microsoft.CodeAnalysis.Common will the CSharp only definition in csproj work?

cerulean wasp
#

because mac VS lacks the batch build feature

scenic forge
#

Why do you need to turn off SG though?

cerulean wasp
#

because I don’t want it running all the time because it generates files

stuck plinth
scenic forge
#

It doesn't generate files, they should all be in memory.

cerulean wasp
#

it makes a filename based on attribute, so if I type [Make(“myclass”)], it generates a file for m, my, myc, mycl, mycla, myclas, and myclass. All because I’m typing lol

#

i’m having it make a file.

#

anyway, it’s a minor inconvenience, and my SG isn’t worth the time I used to make it tbh

scenic forge
#

If you want to generate file rather than in memory source, imo you are using the wrong tool for the job.

cerulean wasp
#

probably

scenic forge
#

That sounds more like a build step.

cerulean wasp
#

i do have a big issue that will take time to explain, that I desperately need help with

#

do you have time?

#

i think you might be the only one who can help

scenic forge
#

I highly doubt that, but you should write it out anyways so everyone in the chat can potentially help.

cerulean wasp
#

I’m trying to figure out my electrical grid, and I’m really stuck. I need help.
-General Graph: I have a directed graph of vertices on a 2D grid, where wires/electrical sources/etc define the connectivity. Vertices are like mini directed subgraphs, where a vertex can (for example) pass electricity from up to down, and left to right (i call this subgraph a pass-profile). Edges occur at adjacent vertices that allow connections (ie if V1 allows elec out going up, and above it is V2 which allows elec in from down, then V1->V2 is an edge).

-Flow of electricity: Vertices are either sources (send out signal, and depend on nothing), conductor (move signal from one vertex to next, or receiver (a conductor whose pass-profile depends on how it is electrified). Every source vertex has an ID, and every vertex keeps tabs of which sourceIDs are activating it (and from which direction) (eg a wire knows it is electrified from above by source1&3, and from left by source 1&5). To electrify, I do a DFS from a source, find all vertices connected and directions they were reached, and update their contents.

-My issue: I am struggling with receivers (pass profile changes depending on how they are electrified). What I have been trying to do is: when a receiver is going to change its pass-profile, send out a negative signal to turn off everything it can reach, change the pass profile, then send out a new positive signal. The issue is that I run into loops and vertices where you reach them from a different direction (ie direction from source ≠ direction from the changing receiver). This makes it hard to update everything in a way that is correct. Can someone help me?

scenic forge
#

I'm guessing your problem involves loops in the graph?

cerulean wasp
#

it’s a couple of things, really

#

i can fix loops, but fixing loops means I unfix branches upstream of the receiver

#

and the presence of diodes in the graph makes some things screwey when I have loops

scenic forge
#

Oh wait, it was you who I gave the semilattice explanation last time.

cerulean wasp
#

yes

hardy jacinth
#

what I could probably recommend you is take a bite at this problem from a more physically-correct standpoint, where energy flow is an implicit function of voltage potential. You would flood-fill potential for vertices, consume the potential for receivers and then see if electricity flows (and in which direction).

scenic forge
#

I thought it was someone else and was just about to refer you to that conversation 😅

cerulean wasp
#

lol

#

i figured out how to block infinite loops

#

i think

#

i refactored dramatically, but I’m still stuggling

cerulean wasp
#

if I put something with logic, I want that logic to be preserved. I do not want it to also depend on how far you are

#

it’s a maker style game, and I want to minimize the headaches for someone trying to use the thing which is my core mechanic

#

which is already pretty complex

hardy jacinth
#

wdym by "not going far enough"

cerulean wasp
#

if signal intensity drops off with distance, then the logic of the circuit depends on how far the connections are

#

like minecraft

#

unless you use repeaters, which I always hated as a system

#

levels in my game should be finite, and evaluating the circuit on one frame should be doable

#

if i’m understanding you correctly

#

but maybe i should be storing distance to each vertex direction from the source?

hardy jacinth
#

... you don't have to add the drop-off distance then, no? :|

cerulean wasp
#

it wouldn’t be a drop-off, but to measure distance via BFS

#

which is a more complex data structure. But I could do it if it would fix my problem

hardy jacinth
#

BFS and DFS are both very simple, DFS keeps it's data implicitly on the function execution stack, BFS explicitly on the manually created stack in the function body. In terms of performance and reliability you could even say BFS is safer because it doesn't get impacted by the maximal call stack depth that DFS does. What I mean by flood fill is: you have (probably) some sources (they are like batteries), connectors (liek diodes) and receivers (any component converting electricity into work). Image below is a mental model of that. I suggest you thinking about this model because It may simplify some of your work.

maxresdefault.png
stoic otter
#

Trying to cache my Transform component why doesnt this work tho?

image.png image.png image.png
hardy jacinth
#

because you update the cached transform's position

sly grove
#

you have two references to the same Transform

tiny pewter
#

what is not work? its behavior same as my expectation...

#

you should restrict the T parameter, it should be unmanaged

hardy jacinth
cerulean wasp
#

i think i have an idea

#
  1. give all sources & receivers a serial #.
  2. When source turns on, DFS from source (stopping at any receivers), and give them its source ID from those directions reached.
  3. For each receiver reached, receiver reads its current source state, and decides if it should send an activating/deactivating signal for its own id and in which directions.
#

this way a vertex knows if it is only being turned on by a given receiver and/or a given source?

stoic otter
cerulean wasp
#

i think this might work, but i need to think thru counterexamples before coding more lol

stoic otter
#

well if i do it with a int excact same thing

#

it will work

hardy jacinth
#

maybe you didn't quite cach the difference between value and reference types

stoic otter
#

?

tiny pewter
#

you just create a new reference to an existing instance

stoic otter
#

how to unpointer it than?

hardy jacinth
stoic otter
#

is there a way tho?

cerulean wasp
#

is there a straightforward way to make a ulong 128 bits? or related

tiny pewter
#

fixed size buffer in struct

hardy jacinth
cerulean minnow
#

what file do I download for an updated gradle? binary or complete?

Screenshot_258.png
cerulean wasp
#

like a struct for a 128bit integer?

stoic otter
#

aa yeah i noticed that is something working

#

how stupid

hardy jacinth
#

Vector3 is a struct (i.e. value) type so you should be able to just write val cachedPos = transform.pos and it will create new vector with the old values

hardy jacinth
stoic otter
#

yeah thats indeed a way but stupid tho thought i was really smart

tiny pewter
#

what you want is not caching, it is deep coping an instance

cerulean minnow
hardy jacinth
#

(so I never bothered with installing it manually)

cerulean minnow
#

I only have to due to lightship ardk

scenic forge
#

@cerulean wasp What are some example circuits that you have trouble with?

jolly token
#

.NET 7 has UInt128

hardy jacinth
stuck plinth
hardy jacinth
#

because ideally I would want this to be somehow available in the same repository

stuck plinth
#

you would probably have a problem with this step:

Create and assign a new label called RoslynAnalyzer. To do this, enter “RoslynAnalyzer” into the text input window in the Asset Labels sub-menu. This label must match exactly and is case sensitive. After you create the label for the first analyzer, The label appears in the Asset Labels sub-menu. You can click on the name of the label in the menu to assign it to other analyzers.
because you wouldn't have a plugin dll to do this with

hardy jacinth
#

is there a way then to add a subproject to my package in the /Packages/, make unity build the dll and assign the tag then? (automatically maybe, or if the subproject changes, the dll is regenerated but tag remains)

stuck plinth
#

i don't think there's a built-in way, but it's probably possible to set it up with enough scripts - you could keep the source in a hidden folder of the package and write your own script to invoke the build system

hardy jacinth
#

hmm, I am starting to wonder if I didn't fall into an XY-problem trap

stuck plinth
#

rather than needing to recompile it all the time can you feed it a config file or something simpler?

hardy jacinth
#

because what I ideally want is:

  • add atribute to my, or possibly user-defined files
  • generate partial classes for those annotated files (with appropriate setters and constructors)
scenic forge
#

Do you need to modify the generation logic?

hardy jacinth
#

it's just weird for me to think about what is essentially a part of this package (generators) as separate project

scenic forge
#

Oh that's what you are hung up on.

#

I'd just ship the dll (and optionally the source of the SG project) and call it a day.

cerulean wasp
scenic forge
#

Can you show the circuit?

cerulean wasp
#

these are the two main test circuits i am up against

IMG_1495.jpg
#

in a square with an arrow means transistor, where connectivity (perpendicular) is turned off if it’s arrow port is active.

#

number in square = source

#

arrowheads outside boxes are diodes

#

top circuit tests that the algorithm doesn’t get stuck in infinite loops. Bottom circuit tests that what something sends out is properly updating things

hardy jacinth
#

Did you take into accounts oscilators? Is your divagation about things not looping essentialy a way of preventing oscilations?

cerulean wasp
#

i need it to halt

#

player will be fucking around with it. you can’t have the game hit an infinite loop, because if I try to evaluate it all in one frame, then the game will freeze.

#

because there is no out

hardy jacinth
#

how does the source marked with I work in your case, in which direction does it give the power

cerulean wasp
#

it gives source outward in any direction it is connected to

#

it does all one DFS. So if 1 is connected to up and left, then it sends a signal of 1-ON to anything it can reach from up or left

hardy jacinth
#

so in the top circuit you basically have the oscilator if both left arrows (I guess inputs?) are off, right?

#

tfu

cerulean wasp
#

if both left arrows are off, then it should just let electricity thru

hardy jacinth
#

when someone powers any of the left transistors, then it oscillates

cerulean wasp
#

no oscilation

#

it must halt that frame

#

halt into a stable config

hardy jacinth
#

stable in that tick

cerulean wasp
#

yes

#

next tick, it will not change without something externally changing it

#

you’d need to bring in another source

#

and if that source is clicked, it need to do its thing and finish.

#

into a new config

#

and halt

hardy jacinth
#

you know that the DFS approach makes it nondeterministic when you want to simlaute electricity flow with the DFS, right? Depending on which branch is taken first the result will differ

cerulean wasp
#

well, I’m checking for that

#

rarher, on paper, making sure that the algorithm ends on same spot regardless

#

i might be able to block off infinite loops at the start of the level, before anything evaluates

hardy jacinth
#

what is your intent for a player experience when he encounters infinite loop?

cerulean wasp
#

the general rule is; if there is any path that can forseeably cause an infinite loop, that transistor allows electricity in, and never spits anything out

#

it becomes a dud

#

this also blocks circuits where what it does makes no logical sense

hardy jacinth
#

hmm, maybe start a thread for that, because it seems like it's becoming a lengthy discussion

cerulean wasp
#

i forsee players wanting to primarily use circuits with more sensical inputs. like turn 1 on to open a door, then turn 2 on to block 1’s signal so the door closes…
But it’s basically expected that players will do stupid shit

#

i think I have a lead for now. i just hope this works.

#

my lead is something closer to BFS

hardy jacinth
#

so, let me get this straight, you want players to be able to play with it, but don't have too much freedom and let loose their creativity? doesn't it sound restrictive to you?

#

seems like a bunch of unnecessary restricions

cerulean wasp
#

creativity, but if the circuit does something that no one can predict, that isn’t helpful to anyone

scenic forge
#

(Use semilattice 😄)

cerulean wasp
#

there will be other parts that send signals at given times

#

like one piece that sets a delay before turning on. and another that sets a delay before turning off…

#

the logic just needs to clearly flow without infinite loops

hardy jacinth
# cerulean wasp creativity, but if the circuit does something that no one can predict, that isn’...

that depends on how you create your rules. If your rules are atomic and predictable, then everytihng else is predictable as well. For example I think about is makeing 2 phases of calculations: one is propagating potential, another is making components react to the potential changes. This sounds deterministic and possibly simple to do, because in phase 1 only potential propagates acoording to already set component state, while in phase to only components state changes without affecting potential flow, so everything can happen simultanously.

#

becasue your infinite loops checking, DFS etc. seem like a complicated set of hidden rules that massively add to programming complexity

#

and add a bunch of edge cases

cerulean wasp
#

my current plan is:

  1. DFS from source to all receivers.
  2. Add each receiver to a queue (unless it fails a dependency check).
  3. while queue not empty:
    —DFS from frontmost receiver to send out its receiver ID (depending on its previous and current electrical state). Stop at any receivers.
    —For each receiver found, if receiver is not in queue, and receiver passes dependency check, then add it to the end of the queue.

Done

#

if I do a dependency check, then I can also telegraph it by drawing something on screen to show that a given tile is not functional

#

you should first discover it by doing the logical thing of seeing “i wonder what happens if I try to make an infinite loop”

regal olive
#

has anyone experienced this error with astar? NullReferenceException: Object reference not set to an instance of an object
Pathfinding.EditorBase.OnInspectorGUI () (at Assets/AstarPathfindingProject/PackageTools/Editor/EditorBase.cs:104)
UnityEditor.UIElements.InspectorElement+<>c__DisplayClass72_0.<CreateInspectorElementUsingIMGUI>b__0 () (at <da771086bc2e4cfc9ad0a72e083a7f98>:0)
UnityEngine.GUIUtility:ProcessEvent(Int32, IntPtr, Boolean&)

hardy jacinth
#

hmm, I created new C# library project and it can't find ISyntaxContextReceiver

#

the ISourceGenerator is available though

hardy jacinth
#

hmm, the Microsoft.CodeAnalysis version 3.8.0 doesn't have it and I don't know what it has instead of that. However installing Microsoft.CodeAnalysis.Common in 3.9.0 makes it available

cerulean wasp
#

i needed to grab multiple nuget packages to make a source generator

#

i saw one tutorial on youtube that helped tremendously

#

and so many which were 100% useless.

#

the good tutorial was like an hour long

#

This was critical. Also I needed unity’s little tutorial to know how to enable a source generator in unity

hardy jacinth
#

the amount of shitty in quiality tutorials and docs is unbelivable

hardy jacinth
#

from the amount of problems I encountered already I could have written hundreds of Medium posts already...

cerulean wasp
#

there’s just a lot that goes into it

hardy jacinth
#

yeah, the whole ecosystem is very complicated. For example compared to Java's gradle this it's like rocket science vs highschool math

cerulean wasp
#

you need to: set up packages, set up connection to unity, make new project, learn how to make the source generator AND syntax receiver, and learn how the syntax tree works to be able to pull info from it.

#

it’s not complicated. there’s just a bunch of shit you need to know before you can do anything

hardy jacinth
#

what is it about the first two steps, "Set up packages and set up connection to unity"?

cerulean wasp
#

and you get no validation if you’re doing it right until it’s almost entirely done

#

you need a bunch of packages to make a source generator. Like code analysis and a few others. and they need to be the right versions for unity compatibility

#

and you need to know how to make unity actually use the source generator

#

again: you have no idea if you fucked up until you are very deep in. takes a while

scenic forge
hardy jacinth
scenic forge
#

They are using 3.8 though

#

Well, for reference here's what one of my SG's .csproj looks like:

<Project Sdk="Microsoft.NET.Sdk">

  <PropertyGroup>
    <TargetFramework>netstandard2.0</TargetFramework>
  </PropertyGroup>

  <ItemGroup>
    <PackageReference Include="Microsoft.CodeAnalysis.CSharp" Version="3.8.0" PrivateAssets="all" />
    <PackageReference Include="Microsoft.CodeAnalysis.Analyzers" Version="3.3.3">
      <PrivateAssets>all</PrivateAssets>
      <IncludeAssets>runtime; build; native; contentfiles; analyzers; buildtransitive</IncludeAssets>
    </PackageReference>
  </ItemGroup>

</Project>
dreamy marlin
#

any idea how to create multiple objects containing a material with a decal shader graph where each object has it's own texture ?

#

i don't really want to create multiple materials just to change the texture

#

my current solution: script that creates a mat and assigns a different texture per object

hardy jacinth
#

sounds simlpe — expose a texture via property in a shader and pass it in the script for each object / assign in material overrides or something

#

take a look at material property blocks

#

and check official documentation

#

it may be required to do something like that:

  Properties{
    [PerRendererData] _Color ("Color", Color) = (0, 0, 0, 1)
  }
#

and if you use shadergraph, they should have this as a toggle on property

hardy jacinth
dreamy marlin
#

but my renderer is a Decal projector

#

does that still work ?

hardy jacinth
#

Hmm, I don't know, you will have to check. Sorrym I missed that part in your question.

dreamy marlin
#

yeah without a renderer i can not set a property block i guess

hardy jacinth
#

what about the DecalRenderer component?

dreamy marlin
#

`using UnityEditor;

namespace UnityEngine.Rendering.Universal
{
[AddComponentMenu("Rendering/URP Decal Projector")]
[CanEditMultipleObjects]
[ExecuteAlways]
public class DecalProjector : MonoBehaviour
{
public DecalProjector();

    public Material material { get; set; }
    public float drawDistance { get; set; }
    public float fadeScale { get; set; }
    public float startAngleFade { get; set; }
    public float endAngleFade { get; set; }
    public Vector2 uvScale { get; set; }
    public Vector2 uvBias { get; set; }
    public DecalScaleMode scaleMode { get; set; }
    public Vector3 pivot { get; set; }
    public Vector3 size { get; set; }
    public float fadeFactor { get; set; }

    public bool IsValid();
}

}`

#

docs don't mention a DecalRenderer

mighty dagger
#

Is Zenject deprecated (last commit 3 years ago)? Is there any other DI framework to use that is maintained? Or do you recommend against using it in Unity?

#

Ah sorry, I see there is Extenjecy fork. But last commit also 6 months ago

hushed fable
# mighty dagger Is Zenject deprecated (last commit 3 years ago)? Is there any other DI framework...

VContainer: https://github.com/hadashiA/VContainer
Reflex: https://github.com/gustavopsantos/reflex/
I'm injecting pretty much all dependencies, but the amount of times I needed to setup something complex and couldn't use Unity's DI (inspector serialized values) ended up being very few in most projects, so I felt having some glue scripts around for those cases didn't seem like a heavy burden. Having code that is walking through larger hierarchies just to pass in few references didn't feel super reasonable.

jolly token
cerulean wasp
#

my method works. it finally works

#

ty for all the help @hardy jacinth and @scenic forge

#

even though I didn’t use a lot of those ideas, it was something tooster said at the end that gave me the idea that worked. And burrito helped a lot earlier on. big thx

wet burrow
#

oh hey @hushed fable didn't realise you were on this server

#

So I'm trying to work with Unity's IAP package. Is there a reason to not generate products individually? It seems you're supposed to upload multiple at a time in batches.
I have to deal with some code someone else made that involves adding individual products, and I'm figuring out if I should try making it work or if I need to scrap it and figure out a system that uploads those individual products in a batch instead

#

Also, what happens if you try to generate a product that's already been generated? Does it override it or create a duplicate?

hardy jacinth
#

oh god what have I found...

#

if you use C# records, and by chance you have a cyclical reference, then everything crashes with stack overflow... because compiler generated toString and printMembers are recursively calling each other...

#

but the worst parts is that it crashes everything — any debugging session, whole editor etc...

#

and it's inherently a C#'s problem

jolly token
#

Honestly this is understandable, what else would you expect from cyclic reference.
Crashing debugger part is not desirable though, that should be fixed.

hardy jacinth
#

I am fine with syntesizing methods that are recursive (although I find it weird), but the side effect of crashing debugger and crashing the editor is grossly unexpected. I assume some kind of exception should be thrown and error shown in the debugger.

trail cedar
#

Hi, is there any way or tool to analyze reserved native memory?

image.png
limber snow
#

on addressables percentcompleted is only for downlaod content, not for load scene really?

fallow eagle
#

Hi ! Since a SerializedProperty have hasMultipleDifferentValues, is there a way to access these multiple different values ? floatValue, intValue, etc will return a single value, so I'm assuming this is just a flag and I'd have to parse SerializedObject.targets to get individual values

sage radish
fallow eagle
sage radish
fallow eagle
coral latch
#

how do i combine meshes correctly?

# 
public void CheckAndUpdateMesh(Block block)
    {
        float x = block.transform.position.x;
        float z = block.transform.position.z;

        MeshFilter meshFilter = block.GetComponent<MeshFilter>();
        CombineInstance[] combine = new CombineInstance[4];

        Vector2[][] cornerOffsets = {
            new Vector2[] { new Vector2(1, 0), new Vector2(0, 1) },  // Top-Right
            new Vector2[] { new Vector2(-1, 0), new Vector2(0, 1) }, // Top-Left
            new Vector2[] { new Vector2(-1, 0), new Vector2(0, -1) }, // Bottom-Left
            new Vector2[] { new Vector2(1, 0), new Vector2(0, -1) } // Bottom-Right
        };

        float[] baseRotationAngles = { 0.0f, 90.0f, 180.0f, 270.0f };  // One for each corner

        for (int i = 0; i < 4; i++)
        {
            Vector2[] offsets = cornerOffsets[i];
            Vector2 diagonalOffset = offsets[0] + offsets[1];

            bool side1 = blockMap.ContainsKey(new Vector2(x + offsets[0].x, z + offsets[0].y));
            bool side2 = blockMap.ContainsKey(new Vector2(x + offsets[1].x, z + offsets[1].y));
            bool diag = blockMap.ContainsKey(new Vector2(x + diagonalOffset.x, z + diagonalOffset.y));

            (string meshSource, float rotationAngle) = SelectMesh(block.item, side1, side2, diag);
            Mesh cornerMesh = LoadBlockMesh(meshSource);  // This will load the Mesh
            float finalRotation = rotationAngle + baseRotationAngles[i] + block.rotation;  // Combine with base rotation

            combine[i].mesh = cornerMesh;
            Matrix4x4 scaleMatrix = Matrix4x4.Scale(new Vector3(0.25f, 0.25f, 0.25f));
            combine[i].transform = scaleMatrix * Matrix4x4.Translate(new Vector3(offsets[0].x * 0.25f, 0, offsets[1].y * 0.25f)) * Matrix4x4.Rotate(Quaternion.Euler(0, finalRotation, 0));
        }

        meshFilter.mesh = new Mesh();
        meshFilter.mesh.CombineMeshes(combine, true, true);
    }```
#

i have 4 meshes for each corner of a block - i choose them and then i want to combine to form a single mesh to then load the block

dusty wigeon
coral latch
#

that doesnt say anything to me

#

i want to glue them together

#

4 glued meshes in center

dusty wigeon
coral latch
#

i know what they are

dusty wigeon
#

Then, what you do not understand by combining those array ?

coral latch
#

im working with Mesh cornerMesh

#

i dont know how meshes work under the hood wheres vertices and all that stuff

#

i just need to combine thats it

#

meshFilter.mesh.CombineMeshes(combine, true, true);
makes it weird

#

it should be full size

image.png
dusty wigeon
#

What do you mean full size ?

coral latch
#

touching

dusty wigeon
#

Touching what ?

coral latch
#

the corners are 1/4 of a block

#

fucking blocks bruh

dusty wigeon
#

Then, combining mesh will not do anything here.

#

Combine mesh does not modify the result

#

They will still be 3 blocks, exactly how they are after you combine them.

coral latch
#

they should be bigger

#

not more

dusty wigeon
#

Combine mesh simply merge mesh together to batch them for GPU performance.

coral latch
#

it combines as if it squishes them all into one

dusty wigeon
#

It will not do that.

#

If you want to do that, you will need to do it manually.

#

Maybe there is a library for that, however I really doubt it.

#

In fact, you should read on Marching cubes.

coral latch
#

nah i think ill just create 4 mesh filters

#

for each corner

dusty wigeon
#

It will still not merge them ?

coral latch
#

im just doing the corners of walls

image.png
#

like this

#

4 combined makes a wall

dusty wigeon
coral latch
#

why not

dusty wigeon
#

Because, combining mesh does not change the data.

#

It simply combine them in 1 mesh

#

For performance reason

coral latch
#

i said ill just load 4 separate

#

and position them

#

combining didnt work for me

dusty wigeon
#

For performance reason

#

Static Bashing.

coral latch
#

ok so you say the combining is literally just combination of meshes as one where it doesnt change anything

dusty wigeon
coral latch
#

okay so then ill need to position them before combining

dusty wigeon
coral latch
#

scaling was a problem

image.png
#

last thing just fix rotations

spiral zenith
#

Does the image you're using have a transparent background?

fallow eagle
#

It looks like the fake PNG you get on Google

spiral zenith
#

Can you send it?

fallow eagle
#

The background isn't transparent

cerulean minnow
#

Anyone have a png of a white arrow on a transparent background? I know it's weird I'm asking for one , like "Just make one in paint bro" but I updated my pc to windows 11 awhile ago and it's messed up all my pc apps.. like they are there but I cant use them. Trying to get windows 10 downloaded again

spiral zenith
cerulean minnow
fallow eagle
#

you saved it as jpg...

#

Plus, this is definitely not code related issue.

frozen flax
#

TestClientRpc(new ClientRpcParams {Send = new ClientRpcParams{TargetClientIds = new List<ulong> {1}}}); why this error? Assets\NetworkFPSController.cs(37,74): error CS0117: 'ClientRpcParams' does not contain a definition for 'TargetClientIds'
Here is that void [ClientRpc] TestClientRpc(ClientRpcParams params){Debug.Log("testing client rpc");}

sly grove
#

seems clear there is no TargetClientIds field or property in that type

frozen flax
sly grove
#

look it up in the docs or just jump to the definition in your IDE

frozen flax
sly grove
#

double check the package version

#

and look at the definition in your IDE

frozen flax
frozen flax
frozen flax
sly grove
#

You wrote ClientRpcParams instead of ClientRpcSendParams

#

basic error

frozen flax
#

Thanks 🙏

#

Im so blind sometimes

limber snow
#

i think i have to change the arquitecture of my game, im using a loading screen that loads scene addressables like main menu or gameplay, the problem is im loading these scenes are additive to be able to show a loading progress, this is working ok. the problem become when i load additive gameplay and some objects are instanced, and these objects move to root scene and not with the scene loaded additive, so when i move to main menu again, these objects instanced are alive in main menu

#

in summary, should i load gameplay as single or check when i instance any game object to move to correct scene?

sly grove
limber snow
#

Nice!! Thanks for suggestions

upbeat path
timber flame
#

How do you link between collider and the component you want to get in raycaat hit.
I know several ways to handle it but would like to know your approach.
The script on that child with the Collider.

The script on the root and a general script on the child with collider to link it with root gameobject.

#

hit.Collider.GetComponent<ScriptA>()
hit.Collider. GetComponent<RootLink>().Find<ScriptA>()

My way is
First I check if RootLink exists on the Collider gameobject or not.
If it exists, find the script.
If it does not, try to get that component on the Collider gameobject itself

#

Another way is to find that script using GetComponentInParent
But I do not like it and it can cause some problems in specific situations

sly grove
#

let that script deal with the intricacies of its own hierarchy

#

that's not our concern

regal olive
#

i have a script that make the gun rotate at your mouse and the shooting .But i want when i m walking right or left i can only shoot 180 degrees in my moving direction

sly grove
regal olive
#

ok

timber flame
sly grove
timber flame
#

It searches all parents until the scene root

sly grove
#

I'm assuming this is some object like... an enemy or a bullet

timber flame
#

Maybe that. Instanced go does not have it but the parent has.

sly grove
#

why would there be an Enemy or Bullet script on anything up through the scene root except the object we care about

timber flame
#

It returns the script but it is clear it is wrong

sly grove
#

there's no bugginess, just use a component that clearly demarcates the root of the section of the hierarchy you care about.

timber flame
#

Because of that, I prefer to find the root of the instanced go, and search downward on children not parents

sly grove
#

the downward search part is definitely an antipattern

#

that requires your raycasting code to know way too much about the hierarchy of the object it's interacting with

timber flame
#

Think about nested panels Ui

sly grove
#

find the root of the instanced go
I agree with this part. Find that root and make sure it has some specific component that we care about

#

from there, just call a function on it

#

the script can manage its own hierarchy and deal with those details

#

Enforce a separation of concerns 🙏

timber flame
#

RootLink component keeps the root of the go. Assign it to collider go

#

also a boolean field to search on children or not.

hardy jacinth
hardy jacinth
cerulean wasp
#

it sucked, but I eventually got it working

#

was not worth the time investment

hardy jacinth
#

because I made some source generator, put it in my Package directory but it's not working at all (no symbols are detected)

#

a while ago I managed to make it work for a bit (attribute generated by source generator was seen), but I moved something here and there and now it doesn't work (it worked on the stage where it was in the Assets folder)

hardy jacinth
#

omg why is it so hard, I don't friggin understand how to do it

#

both importing dlls to a package and making unity recognize my source generators...

maiden tangle
#

I have a question for anyone with some experience using GeNa. I've spent a bunch of time trying to figure out a strange behavior that is preventing me from accomplishing a game mechanic. I am trying to create 200 x 200 terrain tiles with various configurations of GeNa created road pieces on them. I create the spline and then bake the mesh which adds it as a child to the terrain. The issue is when I reposition the terrain the road doesn't move equally with the terrain. It seems to move with a 50% offset. In other words, if I move the terrain 100px the road only moves 50px. I don't see any attached scripts on the road gameobjects that would apply this change so I am assuming there is still some relationship with the Gena Manager that is doing this? One solution might be to place the terrain and then reposition the road, but that seems like it shouldn't be necessary.

#

In fact, just now I deleted the GeNa spawn manager altogether and it is still doing it so there must be some other reason.

#

Ok I think i've narrowed it down to being the mesh renderer? I disable the mesh renderer and I can see that the mesh collider is still positioned correctly. Is it possible the mesh renderer material has an applied offset or something?

steel snow
#

really hate the submesh setup in unity

#

theres no way to extrude along a spline with submeshes

scenic forge
#

Have you imported the SG dll into your project and given it the appropriate asset label?

honest agate
#

is there a way to get the current frame being rendered to the viewport and project it to a texture

BUT just a single object from the frame is being projected to the texture with nothing but transparancy around it

#

like if an object was there and then suddenly it was a floating png

hardy jacinth
#

Yeah, render textures

untold moth
honest agate
honest agate
untold moth
honest agate
#

or is it just called a layer

#

it is just called a layer, sorry

#

thanks!

untold moth
#

Are you sure it's and advanced issue? Either way, share the code properly.
!code

thorn flintBOT
#
Posting code

📃 Large Code Blocks
Large code blocks should be posted as links to services like:
https://gdl.space/, https://paste.ofcode.org/, https://hatebin.com/
https://paste.myst.rs/, https://hastebin.com/

📃 Inline Code
Surround code with three backquotes. Not quotation marks.
To get C# formatting the first line should only contain cs or csharp.
Add a comment with a line number if there is an error message.
```cs
// Your code here
```
Do not share screenshots of code unless requested.

long ibex
#

Sorry, and yeah it appears to be. It seems to be related to the networking im doing

untold moth
#

Null reference exception is not advanced regardless of where it is thrown.

long ibex
#

huh. Ok sorry, I'll move it to a better area then

untold moth
#

And upload your code to paste site.

fallow eagle
#

Hi ! A few weeks ago, somebody suggested to encapsulate the use of SerializedObject in a using, but is there a reason to this ? SerializedObject automatically calls Dispose() when destroyed so the risk of bad memory management is limited and, for example, if I create new SerializedObjects from serializedObject.targetObjects and wanted to cache them so that I don't have to make new instances on each OnInspectorGUI call, I wouldn't be able to respect the using scope. Is this suggestion a real/important one, in which case your insight is very welcome, or can I ignore it ?

real blaze
#

hey. structure question.
how would you structure this data?
you have a non-runtime data that consists of general data as well as upgrades and some data related to formula for calculating certain values.
and then there's a runtime data that just has the final values

as you see, there are a lot of data overlap between the two

image.png
#

and as the DRY says, 'don't repeat yourself'

#

I can simplify those scripts in a smaller text format if that makes it easier to read

#

here:

public class NonRuntimeData : ScriptableObject {
    public float health; // or 'healthBase'?
    public float mana;
    public float damage;
    public GameObject prefab;
    //...
    public float level;
    public float healthBonus;
    public float manaBonus;
    // ...
}

public class RuntimeData {
    public float health;
    public float mana;
    public float damage;
    public GameObject prefab;
    // ...
}
tiny pewter
#

declare a serializable struct outside these card?

real blaze
#

that consists of the common variables? well that would just be the RuntimeData class

#

since nonRuntime contains runtime

jolly token
tiny pewter
#

i use struct because it can copy the value types, isolates the run time data from the initial value
from you runtime constructor i think you may prefer it....

fallow eagle
jolly token
austere jewel
#

It's not. I tend to do it because if you make something disposable then that's what you're asking of me. It is using a destructor to do it... Which is generally not advised

real blaze
#

spotify link? did I see right 👀

fallow eagle
real blaze
#

I totally feel you

jolly token
#

Isn’t level/exp runtime data?

fallow eagle
real blaze
# jolly token Isn’t level/exp runtime data?

ignore the names;
but generally, no, in this context, the level refers to the level of the card. in the game we don't show the level (just damage, hp, and things like that for now). if we had to show the level someday, I'll just add it to the runtime data

tiny pewter
#

just keep the data that will different from initial value inside the struct, the prefabs may be the same all the time...

jolly token
real blaze
#

it's not bad practice per se. but you'll be making it hard to use. if you use it, you gotta use it well

#

like unsafe territory

jolly token
#

IDisposable implies that user controls release of the resource. But Unity.. doesn’t care lol

fallow eagle
#

Putting Dispose() in a destroyer is not recommended ?

jolly token
#

Not recommended

#

Using Finalizer is not recommended in the first place

novel plinth
real blaze
#

if it's a UnityEngine.Object then why not (edit: oh I thought u meant OnDestroy. but yeah finalizer is recommended for disposing stuff, if you're in control of the lifespan of your class )

fallow eagle
#

OK. I'm more familiar with C++ when it comes to finalizer, so I wasn't aware it wasn't recommended in C# (mostly used C# in Unity until a few weeks ago, in a very basic way, but I'm trying to go past the Unity's way)

real blaze
jolly token
#

But Finalizer is not same as C++ dtor, just syntax looks similar, but it’s not RAII

real blaze
#

gets called on GC collect I guess?

jolly token
#

Yeah GC has to call it

real blaze
#

definitely bad then...

fallow eagle
#

Yeah, I looked for it after sending my message. GC stuff.

real blaze
#

why don't you just make your class disposable?

#

makes all the sense

fallow eagle
#

Well, I didn't consider it because I was focused on the reason why I should put SerializedObject in a using in the first place haha Now you helped me have a better grasp on it, yeah, I will probably make my class implement IDisposable

hardy jacinth
#

No idea what's going on when I copy my *.dll to my Packages/<packagename> directory:

Assembly 'Packages/me.tooster.sdf/SDF.SourceGen.dll' will not be loaded due to errors:
Unable to resolve reference 'Microsoft.CodeAnalysis'. Is the assembly missing or incompatible with the current platform?
Reference validation can be disabled in the Plugin Inspector.
Unable to resolve reference 'Microsoft.CodeAnalysis.CSharp'. Is the assembly missing or incompatible with the current platform?
Reference validation can be disabled in the Plugin Inspector.
Unable to resolve reference 'System.Collections.Immutable'. Is the assembly missing or incompatible with the current platform?
Reference validation can be disabled in the Plugin Inspector.

why it cannot be loaded?

real blaze
hardy jacinth
#

I've read this manual probably 3 times already, but it still doesn't want to work, or the instructions are too vague

real blaze
hardy jacinth
hardy jacinth
real blaze
#

ah yeah when you move that dll , you'd need to restart the Editor

hardy jacinth
#

I tried that, but there were some weird problems and it didn't work since then

#

for example when I moved the dll from Packages/ to Assets/ and restarted unity now, the unity prints error:

error CS0006: Metadata file 'Packages/SDF/SDF.SourceGen.dll' could not be found
l not contribute to the output and compilation errors may occur as a result. Exception was of type 'NullReferenceException' with message 'Object reference not set to an instance of an object.'
#

but now it at least detected the generated attribute from source generator

#

but still, even though Rider shows autocompletions and redirects me to generated source, unity says things like

Packages/SDF/<directory-with-assembly1>...: error CS0246: The type or namespace name 'SDF' could not be found (are you missing a using directive or an assembly reference?)
real blaze
#

also, for Unity to recognize your local package you need to have a certain file structure in your package folder. example being my package up there, make sure README and packages.json and LICENCE are there, and put everything you need to compile at either /Runtime folder or /Editor

#

otherwise Unity won't compile your local package properly

#

oh and make sure your analyzer is NOT inside any other assembly. or else it would just work with that assembly and not the main one (Assembly-CSharp)

#

(note that your /Runtime doesn't necessary need a separate assembly to work. it's for convenience only. you can, like me, make a sepaarte /Runtime/src/ for your package scripts with a separate assembly (preferably following the naming guidelines), and a /Runtime/sourcegen for your analyzer)

hardy jacinth
# real blaze on a side note, whys your package name SDF? it should follow the giudeline "com....

I think I misnamed it in the beginning and forgot about it, I'll check in a sec. Aside from that, I have things like READMe, package.json, an asmdef (but root namespace is empty, is that wrong?), As for the sourc egenerator in the assembly, I'll check in a second. So I should try putting the dll in the root of my package? Also, where do I put common code (for editor + runtime + tests + samples etc)

#

I'm gonna mention that the primary usecase for my package is the in-editor work, if that changes anything, and I currently don't have any runtime scripts (not even Runtime directory)

real blaze
# hardy jacinth I think I misnamed it in the beginning and forgot about it, I'll check in a sec....

So I should try putting the dll in the root of my package?
no need. just make sure it's not under any folder with a assembly definition
Also, where do I put common code (for editor + runtime + tests + samples etc)
editor scripts go to /Editor folder, runtime scripts go to /Runtime, test scripts go to /Tests and samples will go to /Samples:

├───Runtime
├───Samples
└───Tests```
> I'm gonna mention that the primary usecase for my package is the in-editor work
so, if the analyzer is also an editor-only thing, then I would do it this way:

├───Editor
│ ├───sourcegen (all source gens)
│ └───src (all scripts, and the editor assembly definition)
├───Runtime
├───Samples
└───Tests

but feel free to use any structure you think is best for your project. as long as the source gen DLLs are not in any separate assembly, it'll be fine
#
  • I updated my package to put the editor scripts inside the /Editor properly, so to follow the guidelines. you can check how that's implemented, the referencings and whatnot.
hardy jacinth
#

by common code I meant the code that can be used in both editor and runtime

real blaze
#

that would be runtime code

#

if it goes to build, that is

#

if it doesn't go to build, it's editor

hardy jacinth
#

ok, so editor includes runtime, I thought they are completely separate

#

can I create sub-assemblies inside my Editor directory then?

real blaze
real blaze
#

don't know if Unity even supports them

shadow orchid
#

have you guys ever had a situation where you preferred using an abstract class over an interface? So far to me seems like interface is superior since it can be modular instead of limited to one parent class inheritence

hardy jacinth
#

yes, when you stick to DRY principle and share data in one common class. There are pros and cons of both abstract classes and interfaces. For example a con of an interface is that you have to explicitly implement interface methods and properties in classes implementing it. You can resort to using default interface implementation, but then you can't use them directly on a type without casting to the interface first ((IInterface) obj).implDefaultMethod(), (contrary to how it works in Java, for example). You can't also restrict the construction of such types. But with abstract interfaces you can, for example, restrict constructor to protected, and require derived types to provide factory methods.

orchid marsh
hardy jacinth
#

There are just things that from a design standpoint make more sense for one or the other at times

shadow orchid
#

awesome thanks guys, sounds like one is explicit whereas the other is implicit being the main difference

hardy jacinth
#

well, saying that "the main difference" exists is most probably an understatement

#

imo the most common "main difference" mentioned is that interfaces should mostly be understood in a form of a contract that something fulfills (implementation agnostic), while abstract class includes implementation details

shadow orchid
#

yeah that makes sense

#

cheers

orchid marsh
shadow orchid
#

Yeah I understood that

scenic forge
#

Inheritance models "A is B," composition models "A can B."

brisk pasture
#

generally find composition the better approach as well, things are just easier to follow

scenic forge
#

People do say "composition over inheritance"

#

Unity's "using components to compose behavior" encourages composition as well.

real blaze
#

yes, although that's a bit different than the original definition of composition in software development I believe

#

generally whatever makes you write less code 🤷 have it whatever name it has

brisk pasture
#

yeah composition is not heavily defined, interfaces help alot with it though

#

but unity components are really no differnet then just having a class that has other classes as fields to add features instead of extending

#

just wish, when clicking and dragging into fields it could do it via interface type, like GetComponent does

scenic forge
#

Not just writing less code, maintaining that code as well

brisk pasture
#

it helps more on the maintaining side

scenic forge
#

If later on "A is B" no longer holds true, it's going to be a painful experience refactoring it.

brisk pasture
#

makes it way easier to rip things out when no longer fit for purpose

#

and can rewrite parts while effecting less of the codebase

#

though man wish C# had implicit interfaces

#

it really changes how you write things, since you start doing all interfaces on the consuming side as a well to state the dependencies of a method or object

sly grove
#

Yes! The duck typing in Go for example is amazing

brisk pasture
#

yeah that is what i am referring to

#

its like duck typing but still 100% statically typed

sly grove
#

It would resolve the issue with Behavior, Collider, Renderer all having .enabled without being in the same inheritance chain for example

brisk pasture
#

also makes it easier to mock things in tests

scenic forge
#

Structural typing 😄

brisk pasture
#

and easier when you want to write something that can work with a existing thing or your own new type

scenic forge
#

TS has it as well and it's a joy to use.

brisk pasture
#

since you just write a interface that has the method of the existing thing you care about and that is it

#

yeah never got too deep into TS, only used it for like 3 projects

#

most familar with this approach from Go

#

know a lot of people think its boring, but really like Go alot

scenic forge
#

Goroutines avoids the whole colored function problem that plagues languages that use C# style async/await.

#

If only Go had error bubbling like Zig/Rust so you don't need to write if (err != nil) everywhere...

brisk pasture
#

well most people dont understand concurrency

#

or even what async/await is

#

since its async/await is just a way to hide the cocurrency from the user, so when they do hit a problem they dont understand why

#

but yes goroutines and channels are very nice

#

yeah the error handling is very verbose, but i do prefer it to what all the try/catch langauges do

#

not used enough Zig/Rust to give comment on how that

#

thought rust just did it through its enum union thing so Result<T>

scenic forge
#

You just need one experience of "welp this method is async now, I guess I need to refactor everything that calls it to be async, and then everything that calls those to be async, and then..."

scenic forge
brisk pasture
#

so x being the returned error, or x being the thing you are calling?

#

or it being like the returned enum that could have some or error

scenic forge
#

Yeah that, any (result, err).

brisk pasture
#

is there a good short hand for wrapping as well?

#

in Go i often add context to the errors as i return up

scenic forge
#

That I'm not too sure, but obviously you can always do your own manual check.

brisk pasture
#

often ending doing like a return fmt.Errorf("My Own Error Message: %w", err)

#

prolly should just try it, but does zig have defer?

scenic forge
#

Not too sure what that is

brisk pasture
#

can do something like
defer f.Close() for example

#

then that gets executed after the function returns

#

and it does not matter where the function returns

scenic forge
#

Oh yeah I think so, same keyword as well.

brisk pasture
#

cool, yeah its useful when you error handling is more or less bailing early with a certain return type

#

guess used in similar ways to the finally in Java,C# etc

#

but function scoped

scenic forge
#

Or using.

brisk pasture
#

less setup then using which is nice

#

since you just give it a statement to execute

#

but yeah similar idea to that or pythons with etc

cloud crag
#

Hey, is there a way to set offset of currently playing state?

image.png
#

this is get only. Without using Play() function?

#

if I use Play() function, why should I use parameters at all in animator... am a bit confused

#

I have many zombies in scene and everyone are in sync with animation when walking. I just want to randomize offset a bit to be different

#

can I set this somehow before transition?

#

from code to be randomized

thorny whale
#

you can't touch the transition settings at runtime outside of the editor unfortunately. As far as im aware from outside the state you can only really set the normalised time with Play.
Why don't/can't you use Play jsut to get them all started at an offset?

real blaze
brisk pasture
#

CrossFade is also a option

real blaze
#

this is way more modular than Mechanim's Animator

#

(Timeline uses this)

#

there has to be some custom visualization/editor repos out there too, for playables

cloud crag
#

Thank you for your answers!

#

I have really simple animator with just idle and move

#

with IsMoving parameter when zombie moves

#

I wanted to make it somehow randomize offset of move animation when zombies start moving around

#

And Playables are too low level for that

thorny whale
#

sounds like CrossFade would work for you then. its essentually you jsut making a manual transition, can you can specify the duration and offset etc.

cloud crag
#

CrossFade and Play yes but, this means I do not have to use parameters then?

cloud crag
thorny whale
#

yeah thats right

cloud crag
#

Very strange that I cannot use Parameters + Randomize offset at the same time?

thorny whale
#

not sure what your game is exactly. but unless your zombies are all playing red light green light they'll naturally desync themselves? So you might just need to initialise them at a random offset?

cloud crag
#

Ah nope. it is really simple. they just come to player

#

when they spawn they are in idle but when they start moving they are like soldiers

#

same animation, very synced.

#

I want somehow to randomize offset

#

and was interested if I could set _animator.SetBool("IsMoving", true); and after that set offset or something

#

so this means, I have to use just CrossFade with randomized offset... what if I just use CrossFade every time then

#

why do I need parameters, I have really simple fsm

#

what do you suggest?

dusty wigeon
#

You could also have different "Alert" animation

#

Which take a different amount of time to be done

cloud crag
#

Thank you for suggestion very much! I will try that

hardy jacinth
#

you could also take a look at DoTween and see if that would fit your use case

cloud crag
#

DoTween for what

#

to blend 2 animations? Zombie Move and Zombie Idle?

#

I managed to solve it using Crossfade

hardy jacinth
# cloud crag

ah sorry, I just skimmed through chat after seeing this

hardy jacinth
#

how do I add dll to a package?

#

I want it to be available for some assemblies in that package

#

but when I try to assign Assembly References (with Override References for assembly turned ON) there are only the unity dlls

regal olive
#

@cloud crag I'm not sure if this is what you're looking for,
however! I thought I'd let you know that it is still an option.

you could use the https://docs.unity3d.com/ScriptReference/StateMachineBehaviour.OnStateEnter.html
method on an Animator State like so: https://docs.unity3d.com/ScriptReference/StateMachineBehaviour.html

then have a randomized delay before starting the Animation Clip...

essentially something like this: https://hastebin.com/share/asozonimap.csharp
keep in mind that it's not the best way to solve your problem... (it's wacky, but worky...)

slow remnant
#

Hello. I have a little mathematical question, based on that thread:
https://forum.unity.com/threads/what-is-dots-equivalent-of-unityengine-worldtoscreenpoint-which-takes-camera.1362358/
It is a DOTS thread, but there is nothing from DOTS there (except for float3, float4, float4x4 which are basically the analogs of Vector3 and so on - nothing difficult. And also there is a math.mul function which is just an overloaded function of matrix multiplication). In fact there is a code for manual WorldToScreenPoint()
So the question (attention to the screenshot): the first step of translating 3d point to the screen is to translate its global coordinates to coordinates based on the camera (so basis is camera`s direction vectors - right, up and backward. And a zero coordinate is the camera's position). And for translating to that basis we need to use an inverted matrix. But here is used something strange instead of that (transformationMatrix is not an inverted matrix, it is something different...). But at the same time code works good and without a mistake. So what is that mystical transformationMatrix and why does it works?

image.png
Unity Forum
Question - What is DOTS equivalent of UnityEngine WorldToScreenPoin...

Hey guys.

What would be the bursted equivalent of WorldToScreenPoint method, which takes camera properties?
I understand I can not burst camera in the...

grave jackal
#

Hey everyone, I'm here looking for suggestions, I want to integrate AI into my FPS game, I want to create behaviors like patrol, attack, attack force, and alert state behaviors, but especially the behaviors I've been working on are cover behaviors. , and behavior of flanking the player, but I feel that everything has been getting more and more complicated for me, I want to know if you can recommend any asset from the Unity Asset Store that will help me with the basic behaviors but that will allow me to create my own advanced behaviors.

At the moment, the ones that I have seen that are closest to what I want are these two assets "Invector FSM AI Template" and "Behavior Designer - Behavior Trees for Everyone", if someone has used them and can tell me which of the two I want can help me meet my goal of advanced behaviors without using ML, or if you can recommend me a new one please.

hardy jacinth
# slow remnant Hello. I have a little mathematical question, based on that thread: https://foru...

Note, that you usually don't need a full-blown inverse matrix. In cases liek that you often want a change of basis matrix, simply. The inverse matrix hes very strong assumptions, for once that M*M^-1 = M^-1*M = Id. A simpler thing is a pseudo-inverse (if that's what it's called, I don't remember official name) matrix. For example it assures only one part of that equation, i.e. xAB = x, where A is a psuedo-inverse of B. In this case xBA =/= x, i.e. B would be right-hand pseudo-inverse (because only when multiplied on the right with A) it works "like" inverse. This case is similar, it uses a simpler matrix which accomplishes what is wanted without calculating full-on inverse, because it's used only on one side of the equation. Read more on https://en.wikipedia.org/wiki/Change_of_basis

Change of basis

In mathematics, an ordered basis of a vector space of finite dimension n allows representing uniquely any element of the vector space by a coordinate vector, which is a sequence of n scalars called coordinates. If two different bases are considered, the coordinate vector that represents a vector v on one basis is, in general, different from the ...

#

change of basis matrix is often simply created from the basis vectors by using them to construct row/column matrix (depends on definition of your vector, is it column or row)

#

this works because the new basis vectors are all orthogonal (so a new basis is just old basis + translation + rotation, without skew)

#

I believe the matrix in question I want is a half-inverse, because it only applies from one side

#

as for the concrete math behind it, you would have to draw the thing yourself and see whwat is multiplied by what

#

because most probably what the matrix in your screenshot does is calculate for each component in the vector a dot product between the vector in old basis and the column vector of a matrix, which describes the camera basis vector

#

there is a hidden *1 multiplication probably after in that equation, because one of those bases (target?) is already an orthonormal base (1,0,0 etc.)

slow remnant
slow remnant
hardy jacinth
hardy jacinth
#

from the code you linked above, it looks like it's applied to "a point in a reference frame, where a camera is in the (0,0,0) but rotated in any way", this the camera basis vectors to form rows in the matrix

#

I propose you check it in 2D, see with yourself but follow the thinking path of "I have a vector U in world space and I want it in camera space — how do I do that?", where you invent what the matrix must look like. Starting thinking about it from the inverse/half-inverse matrix is like starting a proof from the end and trying to find a way to that solution.

slow remnant
#

Well, I am trying to proof that this matrix should be half inverted, but I have no succeed
So if it is half-inverted the equality from the first picture should be right ("r", "u" and "b" are the - camRight, camUp and -camForward. And so Rx is camRight.x and so on)
So the result of the multiplication is on the second picture - it is just a bunch of dots products between these vectors. (r,r) = (u,u) = (b,b) = 1 is true because camRight/Up/Forward are all normilized. But all other values are not guaranteed to be 0-es if I didn't mistake in my calculations. So because of that the equality from the first picture is not true and that matrix is not half-inversed
Can you please hint me where I am wrong in this proof?

ie6WKreGkfo.png vO_v8i0tGbU.png
#

Just in case - E is a matrix of identity

hardy jacinth
#

on the first picture you have essentially written M*transpose(M)=E (btw identity matrix is most commonly written as I or Id)

#

that's not the same as A*B=Id

#

intuitively it makes sense that this matrix is half-invertible, because it means "B is a change of basis matrix, when I multiply it by P I go to basis coordinates. So when I am in basis B, I just get old coordinate system as new basis and do the same"

#

I don't know how to understand the matrix transform intuitively geometrically

#

sorry for the quality, I am using mouse to draw xd

slow remnant
#

No sorries, you are doing very much for helping me understand this thing. Thanks a lot

#

I am trying now to understand all of this :)

hardy jacinth
#

no problem, the best thing is that I also understood it now for the first time in my life :)

#

I think you should stop thinking about the matrix all together. It is just a tool which lets you compute the final vector.

#

in short:
you translated point using p-camPos, so now your camera is in (0,0) but it is rotated. your point is in non-rotated camera space now.