archived-code-advanced

[limpid prairie]

The desktop version doesn't let you even reduce the size too much. Like, when you hold-drag the corner of the browser to reduce its width, at some point, it doesn't let you reduce any further.

But I think the UniWebView version doesn't even go this thin. It starts having problems even when bigger.

[plush crane]

Not sure if this is something that anyone here has experience with....

But does Unity 2020.3.48f1 (LTS) support Asynchronous GPU readback? I thought it should - but when using Android and OpenGLES3 I recieve an error stating that it does not support it. But I do not get this error when building with Unity 2021 (LTS)

Specifically I'm trying to make a call to AsyncGPUReadback.Request

For context - I am hoping to not to have to migrate to Unity 2021 and would prefer staying on 2020.

[stuck plinth]

it definitely supports it in general, are you getting false from SystemInfo.supportsAsyncGPUReadback or SystemInfo.IsFormatSupported(format, FormatUsage.ReadPixels)?

[radiant pasture]

hey, I found a bug in openXR packet. I managed to fix it by forking the packet and fixing it locally. How can I submit the fix to Unity so they can fix it for the next update?
I really don't want local forks of packets because updating them is annoying.

[inland knoll]

https://unity.com/releases/editor/qa/bug-reporting

[round jay]

Hello i have question does someone knows how to access a Hololens2 filebrowther in Unity ?

[radiant pasture]

thanks

they could have put all packages on git and we could do prs

[marble forge]

Hi there people, I'm experiencing a crash on startup in my standalone build on some machines. I couldn't grasp the log, so I would like your help to point me on the right direction here. The attached .rar has 3 Player.log files, from 3 different machines, 2 where the crash happened and one where it didn't for comparison. I also posted a thread on Unity forums with additional details:

https://forum.unity.com/threads/standalone-build-error-on-start-some-machines.1494164/

[plush crane]

Thanks for your reply! I'll double check, but yes I believe I am getting false from these. I'm not sure but I think that OpenGL is not supported in versions older than 2021. However Vulkan and Metal appear to be supported.

[plush crane]

@frosty mountain I actually think I saw your thread post about this question a few years ago. Did you have any input on this? Am I correct in saying that OpenGL in Unity 2020 does not support Asyncronous GPU Readback?

[stuck plinth]

i guess you're stuck targeting GLES3 and can't switch to vulkan? otherwise if it wasn't implemented in 2020 you might have no choice but to upgrade 😔

guess maybe you could put in a different code path using ReadPixels for those builds...

[plush crane]

ReadPixels was actually the reason I swapped to using AsyncGPUReadback - it was soooooo slow. :(

Yes that's correct - I can't support Vulkan unfortunately. So might be forced to upgrade. I did manage to find https://github.com/yangrc1234/UnityOpenGLAsyncReadback this, but during my testing I will need to use Cmake to build a native android plugin.

Ty for all the help this has been very insightful!!

[misty glade]

Using unity's new input manager, how should I structure/dispatch events? I'm used to doing it the UGUI way, which is to raycast-capture IPointerX events on the component I care about (buttons, typically).

Should I be making some sort of input dispatch component that routes the input off to the correct code based on game state or something? This seems.. bad? It requires this dispatch component to be tightly coupled to like.. everything in the game.. and also manage state and so on.

What's your approach?

[sage radish]

It was added recently, 2021 sounds about right. But I've seen it count towards a significant portion of GPU time, even with a pretty small texture readback.

[plush crane]

Thanks! Yeah it looks like it was added in 2021 and someone on the forms confirmed that for me.

Interesting that you've seen that it's slower. I had a render texture process taking 15ms on my thread and after updating to Async GPU it was only taking about 2ms.

[sage radish]

It wasn't slower than ReadPixels, I never used that. It was just much slower than what I was seeing on desktop, so it wasn't really worth using in my case.

But it'll probably depend on the specific device hardware and drivers.

[plush crane]

Interesting! Thanks for sharing! I'll definitely keep that in mind. We're running on VIVE Focus 3's - so really trying to get as much performance as we can get. I'll have to play around with it a bit more and see what works and what doesn't.

[flint oxide]

how do I completely disable frustum/occlusion culling?

seems like unticking it on the URP Camera does nothing

[sly grove]

The input system and the UI syustem are orthogonal. You can and still should use IPointerX events, etc. That stuff is part of the UI system, not the input system

[misty glade]

Yeah.. I was more wondering about the "send this event to this component" workflow since what I've come across so far doesn't seem to indicate this is a Thing - ie: https://stackoverflow.com/questions/70636241/unity3d-new-input-system-is-it-really-so-hard-to-stop-ui-clickthroughs-or-figu

I like making my UI components more or less self sufficient, and it .. didn't seem reasonable to write a UI dispatcher that would need to be aware of UI vs "world" input and send it (or not!) to the right place.. the new input system is probably not what I need.. although it does make "gameplay" input nice - using IPointerX handlers is clunky when you want to do things like double touch, pinch and zoom, gestures, finger/touch tracking etc

https://docs.unity3d.com/Packages/com.unity.inputsystem@1.2/manual/UISupport.html#handling-ambiguities-for-pointer-type-input

Re: the bit about ambiguities under "UI and game input" - my games tend to have world and UI interaction at the same time.. ie, something like this (this is just a proof of concept but the idea should be clear): there's world interaction (click and drag to pan the camera) and UI interaction (click a button to deploy a unit)

[misty glade]

Anyone have any brilliant ideas how to get a list of coordinates that a world space bounding box contains on a GridLayout? (for hex shaped tiles as shown in the video above)

I can get a grid's coordinates given a world position, but I can't think of a way to get all the hexes covered by a "box"

At first I was thinking I'd just grab the cell at each of the four corners of the bounding box (and everything in between), but that misses this case:

[misty glade]

Fuck it, I'm too dumb to try to figure out the math for whether or not a line intersects any part of the hex.

I'm just gonna grab lower left and upper right cartesian coordinates and assume everything inside is contained by the box. It works 95% of well.

[crimson pumice]

Is there a way for me to control the values that come out of Random?

For example, if I do InitState, I can set a seed. That is all fine and dandy.

However, the actual values that come out of the static properties (like Range(), or insideUnityCircle) only repeat based on times ran.

This seems completely arbitrary. I have zero control. Say I want to supply my own enumerator value, I simply cannot. It is literally just based on times ran.

Is there a solution to where I can take advantage of Random but mutate the result? So say I have a seed of 5230458 but I want use the value x = 24, I want to be able to get that same exact value if I input x = 24.

Think Minecraft seeds.

I want to use Random without having to come up with my own hashing method.

[regal lava]

Isnt that just hashing

[sly grove]

Random state works exactly the same way as Minecraft seeds

[crimson pumice]

Sure, but all of the properties and methods on Random are dependent on how many times they are called

[sly grove]

That's how Minecrafts prng works too

[crimson pumice]

So if I want to recall the values from the 1st time a property is called, I can't

[sly grove]

That's how all prngs work

[sly grove]

Of course you can. you call InitState

I'm not sure what the use case of that would be though

[crimson pumice]

So if I want to recall a value that is 100 times in, I need to recall InitState and then throw away 99 calls of the property?

[sly grove]

You call InitState, then you run your level generator

[sly grove]

What's the use case for this

[crimson pumice]

That's the thing, I want to generate these on-the fly for an infinitely scaling world

I don't want to have to store a pre-gened massive array

[sly grove]

You can save the current state of the prng after 99 calls

Then use that later

https://docs.unity3d.com/ScriptReference/Random-state.html

[crimson pumice]

Okay, so I have to pre-gen everything

I have no option to generate on-the-fly

[sly grove]

I just gave you said option

[crimson pumice]

I am confirming the option

[sly grove]

No

[sly grove]

^

[crimson pumice]

But that doesn't make much sense given how Minecraft works. Minecraft can chunk-gen in one direction

Unless you're saying that Minecraft is generating a ton of empty chunks in a radius or something when you go in one direction?

The only option I have then, if I want to achieve my specific goal, is to create a custom Hash function

which I was trying to avoid because it's a bit complicated

[sly grove]

Minecraft world generation is based on noise functions. The noise function scale and offsets are randomly generated when the world is generated

After that it's just perlin or other noise

[crimson pumice]

I see

[sly grove]

And noise functions are NOT random

[crimson pumice]

I don't care too much about it being prefectly random, I care about it seeming random enough

[sly grove]

Yes that's what noise functions are for

[crimson pumice]

but that perlin noise example sounds a bit more like what I need so I might do that, idk. It kind of goes along with what I was trying to originally do which was find a function that gives me enough randomness to distribute objects in the world

Thanks for the insight

Also for context: It's part of a PS1 gamejam challenge for VRChat where we need to stay within super tight resource limits, and I had already created a grid system to move existing game objects.

So I was trying to cross reference the grid positions with a function to determine what objects to place in that grid tile, and I knew about minecraft seeds.

[ancient talon]

I'm still quite puzzle as to how exactly AndroidJavaClass and AndroidJavaObject works in general... When I created an object after referencing the class in .jar library, it somehow doesn't seem to recognize the code from the object, even though looking in the .jar file show method availability. I would run into NoSuchMethodError errors. (The method in question is not static, and can verify that the class constructor works...)

[long ivy]

naturally, that's going to be hard to help you with without seeing the code you're trying to use. Since you're seeing NoSuchMethod and not ClassNotFound, it sounds like something is wrong with the way you are calling the method

[ionic ivy]

If someone worked with video encoded with chapters, do you know if there are any additional steps you need to take to have faster seeking times between these timestamps in the video in Unity? (using AVPro, but that shouldn't make a difference)

[jolly token]

Not possible to add state directly. But look into https://docs.unity3d.com/ScriptReference/AnimatorOverrideController.html

And sure legacy animation can do it too

[rugged radish]

does unity expose the number of its internal workers anywhere ?

at runtime

[sly grove]

oh the worker threads? I thought you meant employees lol

[rugged radish]

yeah, thread workers

[sly grove]

You can inspect them with this: https://superuser.com/questions/462969/how-can-i-view-the-active-threads-of-a-running-program

It's probably based on the current hardware or something

(assuming you're on windows)

[rugged radish]

thanks, so they don't expose it through their apis

i just need to get the number at runtime to use it in my own code

but looks like I need to deduce that number through other means

[sly grove]

as far as I know, no.

what are you wanting to do with it?

[rugged radish]

generating data in chunks in background in my worker instances
workers allocate a bunch of persistent buffers, so there's no point in having more of those worker instances than unity's workers, it will just waste ram

[sly grove]

Typically for this kind of thing I just use https://docs.unity3d.com/ScriptReference/SystemInfo-processorCount.html

[rugged radish]

yep, I used that before, just not sure how unity treats cpus with low perf cores, it can ignore them and only use high performance cores
so wanted to check if I can just ask the engine how much worker threads it spawns

cpus like 12 and 13 gen intel

[sly grove]

not as far as I know

[primal knoll]

rider trial expiring today, need to find my credit card

[misty glade]

I want to highlight a tile on a TileMap. How can I do this? I have a tilemap, but only see API for setting a single tile at a position with SetTile(). Is there a way to draw two tiles on top of each other? Or should I maybe make two grids, one that's empty most of the time and contains the "highlight" tile?

(the tile on the right is mostly transparent, I'd like to use that as a highlight outline thing)

https://docs.unity3d.com/2023.3/Documentation/ScriptReference/Tilemaps.Tilemap.html

[sly grove]

You could tint it with https://docs.unity3d.com/2023.3/Documentation/ScriptReference/Tilemaps.Tilemap.SetColor.html or use a second tilemap for an overlay or just a SpriteRenderer or Graphics.DrawTexture

Lots of options

[dusty wigeon]

Second Tilemap is probably the easiest to implement.

[misty glade]

Yeah, second tilemap was the approach I went, will probably use multiples for various UI needs

Now I have another problem... IPointerMove doesn't seem to fire near the edges of the window... I'm 99% sure I don't have anything blocking the raycasts but.. this one's stumping me

Oh, I'm dumb. The ui capture GO wasn't stretched to the canvas 🤦‍♂️

a zillion things to do, but the UI is "functional" - select a unit, give it commands, interact with the map, etc.. It's always fun to start a new game

[spiral fulcrum]

G night all, its an planet, and de red lines is rays from center of planet to floor of planet... idont know why this rays dont casting to de down side, anyone can help?
The reason to cast on all sides is to spawn trees random on planet floor.

[sly grove]

You would have to show the code

[spiral fulcrum]

    {
        if (treeList.Count > 1)
        {
            DestruirObjetosNaCamada(treeList[0].gameObject.layer);
        }

        for (int x = 0; x < gridSize; x++)
        {
            for (int y = 0; y < gridSize; y++)
            {

                Vector3 samplePoint = GetPointOnSphere(x, y, gridSize);
                Ray ray = new Ray(samplePoint + Vector3.up * (planetRadius - 360f), Vector3.down);
                RaycastHit hit;

                if (Physics.Raycast(ray, out hit, planetRadius * 2))
                {
                    Vector3 rayDirection = ray.direction;
                    Vector3 hitNormal = hit.normal;

                    if (hit.transform.name.ToString() == "Starter")
                    {
                        Vector3 spawnPosition = hit.point;
                        origin = hitNormal;
                        target = hit.point;

                        Vector3Pair pair1 = new Vector3Pair(origin, target);
                        listaDePares.Add(pair1);

                        float elevation = Vector3.Distance(spawnPosition, transform.position) - planetRadius;

                        if (elevation >= minHeight && elevation <= maxHeight)
                        {
                            Quaternion spawnRotation = Quaternion.Euler(0, Random.Range(0, 360), 0);
                            GameObject tree = Instantiate(treePrefab, spawnPosition, spawnRotation);
                            tree.transform.up = hit.normal;
                            //tree.transform.parent = transform;

                            treeList.Add(tree);
                        }
                    }
                }
            }
        }
    }```

[spiral fulcrum]

there is

some of noob and ugly code hehe, but is only things im testing

[sly grove]

absolutely ll of your rays seem to be directed directly down: Ray ray = new Ray(samplePoint + Vector3.up * (planetRadius - 360f), Vector3.down);

also I don't see the code here that ctually draws the red lines

[spiral fulcrum]

i will put all class code minute

[sly grove]

BTW if you're a beginner you should be sking in #💻┃code-beginner generally

[spiral fulcrum]

´´´private void OnDrawGizmos()
{
foreach (var coord in listaDePares)
{
Gizmos.color = Color.red;
Gizmos.DrawLine(coord.origin, coord.target);
}
}´´´

ops

lol

[sly grove]

origin = hitNormal;
You're also treating the normal s if it's a position? Normals are direction vectors, not positions

there's a lot of issues with this code.

[spiral fulcrum]

this hitNormal is hit.normal (i think its where from ray start, right ?)

[sly grove]

the normal is not where the ray starts, no

that would be ray.origin

The normal is the normal vector of the surface where the ray hit

[spiral fulcrum]

its explain why im get errors hehe

ive changed to ray.origin

and here is the problem hehe

really thanks man

[sly grove]

yep that looks like what your code is doing

all the rays are going straight down

not sure what you are intending to do

[spiral fulcrum]

im trying to send rays from center of planet to a floor of planet... in point ray hit iwill spawn an tree

[sly grove]

wouldn't you need to send rays from outside the planet towards the center?

colliders generally don't work from inside. Is this a mesh collider?

[regal lava]

Could just get vertex info of those of a certain radius from the middle

[sly grove]

I'd basically pick random points on a sphere larger than the planet (with the same center) and raycast towards the center of the planet

[spiral fulcrum]

yes

[sly grove]

yeah you'd have to raycast from outside, not from the center

[spiral fulcrum]

hummmmm... lemme try

[spiral fulcrum]

and calculate the ocean height to dont spawn inside ocean, right?

[sly grove]

I don't know anything about any oceans

if you have an ocean height, sure just ignore any that end up at an altitude less than the ocean height

[umbral trail]

is there any way to get the number of materials on a MeshRenderer without allocating an array of materials?

[sly grove]

You can use https://docs.unity3d.com/ScriptReference/Renderer.GetMaterials.html to avoid allocations

[umbral trail]

I don't want the materials at all

[sly grove]

there is no materialCount property, no

[umbral trail]

I guess using a list would keep the garbage to a minimum, thanks

[jolly token]

Your mesh’s submesh count is material count

[stoic peak]

How do I convert a normal c# async method to use UniTask? Do I just change the return type from Task<T> to UniTask<T>? My method calls an async method from another library that returns a Task

[untold moth]

Unitask just provides some convenient wrappers around the regular task afaik, so you'd use it just like any other task.

[mellow sail]

does anyone ever use notion API in Unity?

[dusk plaza]

Do anyone knows if this works for Built in Render Pipeline?
https://docs.unity3d.com/Packages/com.unity.rendering.hybrid@0.51/manual/material-overrides-code.html

[stuck onyx]

anybody knows why such a straightforward command like" Screen.SetResolution(x,y, true) is not changing my resolution at all?

[compact ingot]

your monitor might not support the given resolution

[stuck onyx]

its an android device

and im using:

Screen.SetResolution(Screen.width/x ,Screen.height/x ,true);

to be sure the ratio is the same

[compact ingot]

you cant set exclusive fullscreen on mobile

Exclusive full-screen mode is only supported on Windows standalone player.

[stuck onyx]

ow, so i shall do the same but with false?

[compact ingot]

idk, i would expect mobile to be somewhat particular about what is allowed though

[stuck onyx]

I've read also in forums the fullscreen toggle had to be disabled in playersettings but the alternative is this:

and doest make too much sense to set windowed for a mobile device 😅

[compact ingot]

all mobile apps run in an equivalent of "fullscreen window"

but ofc thats defined by the OS

[stuck onyx]

gonna try the command without the fullscren flag

thanks a lot @compact ingot

[sweet cedar]

so I locked my mind into a particular save system but now I'm having some issues and need help figuring those out

I suppose I was too strongly influenced by untyped languages and now it's conflicting with c#

the idea is that classes that have persistent data pass a copy of their values to a central savedata object that then gets serialized to disk

easy enough to do with untyped, just need to declare each variable once in the main class then assign it to a unique key in the savedata object

but when typed, I now realize the problematic of having to declare each save entry twice, while sometimes even having to create doubles of some classes just to make them serializable...

is there some way I can do this in c#? to just toss whatever type into an object that gets serialized by json in a neat way?

[sly grove]

basically whatever serialization framework you use will have some rules about how it decides which things to serialize or not

as long as you follow those rules, your struct can be serialized neatly

usually you can shape your object in a way that it's usable both at runtime and for serialization without too much fuss

occasionally you might need a little "glue", especially when it comes to having performant runtime data structures that don't translate well into efficient storage structures

[sweet cedar]

so if I get you right, instead of separating all game logic and persistent data logic, just make all game logic classes serializable directly?

[languid spire]

i wrote a shader a few months ago to occlude object behind it. works great on gameobjects . . . but, ignores terrain. any ideas of what I should google to figure out why.

[untold moth]

Occlude? Why do you need a special shader to occlude things. The default shader does that very well. Unless I'm misunderstanding the meaning of "occlude" here.

[real blaze]

a bit more complicated than that. what you seem to be referring to is commonly known as snapshot of objects, and that's both expensive and pretty unstable. I can tell you why if you needed to.

for your problem, you can just make a serializable data for your object and use that as the saving data. with some initialisation method you can give logic classes a chance to boot themselves up based on these data. There are tons of ways to achieve this. also feel free to ask for more details

[languid spire]

@untold moth i left out an important detail. it invisibly occludes, so with pass through ar, it hides virtual objects behind it, while still showing the real world

i put it on walls in ar passthrough, so you see the real walls, and it hides virtual objects behind it

i got it working last night 🙂 thanks for the follow up question

this may not seem like it, but its the followup question:
I'm working with Meta scene data. I can get a collection of verts that make up the outside edge of a floor. Anyone have a function laying around, that will build the mesh with them, and have correct normals. I've had trouble drawing the tri's when i've tried more complex dynamic mesh generation before. They use a plane by default for the ground, that overextends past the walls, and that breaks what i'm going for

[mellow sail]

Excuse me, does anyone ever use the Notion API with unity?

[spiral fulcrum]

all works !! thanks for ur idea !

[true coyote]

I am generating my terrain procedurally but at the moment it just looks like a slice / layer of terrain, and what I am looking for is to have the terrain look more like a cheese slice instead (basically I want it to be volumetric and to have thickness).

Let's tackle this like I am 5, and call this terrain chunk a cubical room, and this cubical room is let's say for the sake of this scenario 4x4. It has air above and ground bellow, represented by 1s and 0s. Where we have 0s, that's where the air is, where we have 1s, that's where the ground it. Somewhere in the middle of these two values (ISO level) is where these values meet and that is where the "surface" of the terrain is created, but ideally bellow this surface I want to have ground. Currently in my case, bellow this surface level there is nothing.

So, back to the explain like I'm 5 scenario, this terrain chunk is a room, and I want this room (the terrain chunk) to have:

• A Ceiling (the surface), and the level at which this ceiling is mounted is decided by the ISO value and the ground level.
• Walls, that start from bellow the ceiling/surface and go down all the way to the base of the room (the edges of the chunk).

Currently__ I did manage to have a surface (ceiling) for the "room"__, but I don't know how to make the room have walls and ground bellow the surface. I am using marching cubes.

Marching Cubes Shader Code: https://pastebin.com/VWVLuki5
Noise Compute Shader Code: https://pastebin.com/EHYQj1Vr
Noise Generator Script: https://pastebin.com/ZsZTA1Hu
Chunk Mesh Script: https://pastebin.com/7CG3EwyF

What I'm looking for (but ignore the blocky-look): https://ibb.co/GnhVd2V

I am attaching a photo bellow with how my terrain looks right now (first photo) and how it's supposed to look under the surface (the other 2 photos).

If anyone can have a look at my scripts and tell me what I am doing wrong or what I need to do please let me know.

[misty glade]

I'm getting erratic behaviour when I try to destroy a game object from OnComplete() in DOTween. Docs don't say anything about this. Anyone know anything about this?
Edit: Nevermind, error elsewhere.

[dusty wigeon]

Pretty sure that Cube Marching should already be able to generate cave. If it is what you want. If we ignore 3rd dimension, the following should work.

My advice would be to isolate your issue and try to generate exactly that in 3D. If you arent able, take a step at a time an try to generate 1 cube, then 2 cubes. At some point, you are going to be able to generate what you want. From there, implementing it in the actually game will be easier.

[true coyote]

I've made a diagram with my current situation. So the issue at hand right now isn't that I don't have caves, what I'm trying to figure out is why I don't have solid ground beneath the terrain and been trying to add more noise but it messes up the mesh. I don't get it, it should be as simply as saying anything bellow surface (iso) level is 1, therefore fill it up with ground, everything above the surface is 0, therefore air, therefore don't generate anything there.

So before getting to caves, I'm just trying to understand how I can have it like in the 2nd version where I have terrain and ground and the rest of the chunk is air. Currently I'm in the first version, where there is air above and bellow surface level and I don't understand how to fix that first. I've played with the ISO level, with the noise values and the only thing I managed is to have the terrain spawn lower or upper across the verticality of the chunk, but can't seem to fill it bellow surface level. Basically all I've managed to accomplish was lowering the surface level or setting it higher, which is not what I want. Again, 2nd version in the diagram should explain things more clearly.

Or is this only possible if you use cubical voxels like minecraft ? Is there an issue with the Marching Cubes algorithm where when you smooth out the voxels you can't have solid terrain underneath anymore ?

For example in minecraft, the chunk is actually solid. Caves are not even generated initially, they are gerated later down in the octane tree. As in, they don't only generate the surface, they generate the ground as well, which is solid, no caves. Caves are carved later.

I can't even generate a solid chunk so I'm not even thinking about caves at the moment until I solve this issue first.

[dusty wigeon]

Minecraft is not made with voxel, all individual cubes are cube. They arent using Marching Cubes.

[dusty wigeon]

Marching Cubes does not prohibit having underground.

[true coyote]

Then do you know what do I need to change in my alogrithm to have underground cause at the moment it refuses to do so, it just moves the surface up and down across the y axis ?

And because there is no underground, when I dig, the ground ends.

[dusty wigeon]

Your issue is probably because you are using a 2D noise generator for the Height and not incorporating other type of generation such as Perlin Worm.

https://www.youtube.com/watch?v=aQ-bomm3lEc

[true coyote]

This is what happens when I try to dig.

[dusty wigeon]

Cube Marching does not have the notion of Up and Down.

[dusty wigeon]

Either have a hard limit or have a way to dynamically generate new "chunk" as you dig lower and lower.

[true coyote]

I did watch the video you linked, it looks like that's for cave generation but I'm trying to give volume to the terrain.

[dusty wigeon]

This is what you have at the moment

Here are 3 solutions

1- Add a hard ground under the level. (Not really what I would do)
2- Add dynamic chunk generation for digging. (Definitely more challenging and demanding - Ask yourself how much deep digging is important in your game)
3 - Limit the digging (Definitely the most scalable solution and easiest solution)

And, if you want what Minecraft does, then Cube Marching is not the way.

[true coyote]

What I have in mind is this.

[dusty wigeon]

That would be solution 2.

[true coyote]

As you dig down more chunks generate.

Yes but because I don't have my chunks solid I cannot do that, which is what I'm trying to solve here.

At the moment if I layer chunks vertically it would make no sense cause they would look like paper slices.

[dusty wigeon]

What do you mean ? You have a hole because you do not have any data below it.

[true coyote]

Wait, do you mean that if I add vertical chunks they would automatically connect with the other chunks therefore making the terrain solid ?

[austere jewel]

Below the ground is considered solid by the data, there doesn't need to be any geo to show that.
For example with minecraft there wouldn't be vertical walls if you did a cutaway of a chunk like your above pics. The vertical walls only exist because they're putting them there, presumably because there's empty space next to the chunk

[dusty wigeon]

I mean, you might need to do some seeming between the chunk, but you would be basically doing the same as increasing the height of your chunk.

[true coyote]

Wouldn't this happen if I tried layering the chunks vertically as I have thme right now ?

Ignore the chunk 1, it's supposed to be a slice as well.

[dusty wigeon]

Why ? Just fill your chunk with filled data and not empty

[austere jewel]

It's far below the y height of the ground, so why would you generate the ground surface there

[true coyote]

Well yes, that's what I want as well. Of course I do not want internal walls as long as there's solid stuff connecting to the chunk (like other chunks) but when isolating a chunk or slicing it, it should appear as if it's a solid piece of ground (but only then, otherwise the engine would have to render stuff the player wouldn't even be able to see anyway, so there's no point).

[true coyote]

I was just trying to show the fact that my terrain only generates the ground level and there is nothing solid underneath, so I thought that if I generate chunks vertically, it would just do the same for the other chunks, therefore creating a bunch of ground slices on top of each other, separated by air.

[dusty wigeon]

Then it shouldnt.

[austere jewel]

If the ground's y height isn't in the chunk then they're shouldn't be a ground generated there. The same goes for the chunks that exist in the sky, if something is totally air or totally ground then there should be no geo

[true coyote]

But in my case currently it generates air bellow the iso level even though it should generate ground. It only seems to generate terrain data between the 0 and 1 at iso level, but nothing bellow it, therefore air. That's what it doesn't make sense to me.

It does exactly this @austere jewel

See, nothing bellow the terrain layer.

[dusty wigeon]

There is someting, it just not generating it because it isnt needed.

If you fill the whole things with white dots, it wouldnt generate anything till you make a dot black.

[true coyote]

Then how comes the black dots are empty as well in the example, (above ground) ?

There something I don't understand here, which you guys obviously do.

[dusty wigeon]

Cube Marching generate a mesh for the transition between black and white dots.

[true coyote]

Good. And what if I want to generate stuff bellow the black dots/white dots. I just like to say black dots are 1 therefore ground, white dots are 0, therefore air. Iso level = 0.5, therefore generate terrain in the middle, but make sure to fill underneath as well, because everything bellow iso level is considered ground/black dots. Like isn't that how it's supposed to work ?

[dusty wigeon]

If there is no transition between black and white, there is no mesh. If you want something below, then generate a new chunk filled with white dot (In your case).

[true coyote]

Let me actually do that and see what happens.

I'll be right back with the results.

[dusty wigeon]

It is hard to be more explicit then this.

[true coyote]

Unfortunately it looks like I was right, the result is exactly what i thought it will happen, told you guys.

[dusty wigeon]

Visualize the points. Make a snapshot where you show the points as reference

[true coyote]

Give me a moment.

[dusty wigeon]

This is one chunk

Do it with two

[true coyote]

M1 melting 😆 hold on

[dusty wigeon]

So, from what you see, which version you have from the picture I sent ?#archived-code-advanced message

[true coyote]

The first one (left)

[dusty wigeon]

Exactly, then it is expected to have what you have.

You want the right one

It just means that you have to change every points under to a red one

[true coyote]

I apologise, updated the visualisation to be a bit more readable.

[dusty wigeon]

The other one is better

It really encapsulate what Cube Marching is.

However, in both case you see the issue. You need red, not blue under

[true coyote]

Do you have any idea why is it like that ?

As in, in which one of the scripts do I have to look for the issue ?

[dusty wigeon]

I mean, you are definitely able to answer that.

Just look at the correspondence between the pattern.

[true coyote]

Well it's not the ground percent. The issue is somewhere else, cause even when the dots are red, there is no ground bellow.

Ah, I think I get it now. The blue dots represent the ground, not the red dots.

[true coyote]

Ah, nope. Actually, red is indeed ground. So basically back to square 1, ground isn't solid where it should be.

[true coyote]

If anyone knows what exactly causes this to happen please let me know. You can have a look at the scripts I am using here.

[true coyote]

💡 I think I just got an insight. I was looking at this from the wrong perspective, because I was missunderstanding what I am seeing. The blue dots do not represent "empty space" and the red dots do not represent "ground space", and the level of blue / red dots (the horizontal line which splits the two across the chunk) is actually the ISO level. If I increase the ISO level for example, I see more red dots than blue, but that doesn't mean now there should be more gound, it just means the weights have different values, either tending more towards negative either towards positive. That is what I think it's actually represented here.

So, in conclusion, this means that there is no issue with the way the terrain is generated, it generates like the code tells it to. The issue is me, I do not understand how the code or the algorithm works exactly, and I am just starting to get a bit of grasp. I guess unless someone takes their time to look a bit into my scripts I will not be able to solve this one until I figure out the rest of the puzzle as well. It's a slow and painful process but at least I'm making progress, even if it's very slow.

[zenith osprey]

Hey! I'm reaching out to the community here after spending hours despairing over Unity Physics. Maybe there is an expert on mesh colliders here.

The following setup: I want to move a cube (box collider) through a dynamically generated voxel game world. I move the rigidbody of the cube physics-based, causing it to rub across the collision surface. In the first version, I had created mesh colliders for my voxel world based on the visual meshes. Accordingly, there were chunks and different mesh colliders in the world. The problem was that the cube got stuck in the edges between different mesh colliders, apprubtly changing the velocity.

To solve the problem, a single mesh collider is now generated around the cube and updated regularly (see video). As you can see in the video, with a small probability it still gets stuck between the triangles of the meshcollider and changes direction. I am walking straight to the left in the video, but bounce of the wall. The video is x0.25 for better understanding. The movement should be as deterministic as possible, which is why such behavior is unacceptable. Does anyone have an idea how I could solve this problem?

Here is also another video looking more from the top.

The Cube RigidBody is already set to: Interpolate, Continous Dynamic.
Physics Settings: Bounce Threshold 2, Sleep Threshold 0.005, Default Contact Offset 0.05, Solver Iterations 8
MeshCollider: No Rigidbody, but the generated mesh has Mesh Cleaning and Weld Colocated Vertices enabled

[clever urchin]

@zenith osprey Tried changing your physics precision?

In project settings somewhere

[zenith osprey]

Probably yes, I played around and increased all physics settings based on the docs

[clever urchin]

Default contact offset

Set to like 0.0001

[zenith osprey]

I don't understand why I didn't stay on these settings earlier. At first it seems to do the trick. I think for now this solves my problem until other issues occure.
One more aspect I can think of in the future is, that I want performance sparing values for a later mobile port.

I just got another idea as a last resort in case of problems. Do I understand it correctly that I can also filter the occurring velocity changes caused by overlapping? Since I have access to the voxel data, it's relatively easy to see where I'm bouncing off a smooth surface where that shouldn't be happening.
https://docs.unity3d.com/Packages/com.unity.physics@0.3/manual/collision_queries.html

Damn, I remember why this is not an option 😄
A low value improves the collision with the walls, but breaks the collisions of several cubes with each other. As the docs also say, the collision becomes jittery. I can probably find a sweet spot by minimizing both problems. But a really reliable working solution would be better. (Video at 0.5x)

[mellow sail]

does anybody know how to request API post with unity?

[zenith osprey]

I previously did use the programm postman to test my requests. It can also generate you C# Code for the requests that also works in Unity.
https://learning.postman.com/docs/sending-requests/generate-code-snippets/

Otherwise eg. this post looks promising
https://stackoverflow.com/questions/51926762/how-to-write-http-post-request-in-unity-by-using-unitywebrequest

[mellow sail]

okay thank you

its driving me crazy with this Notion API, the syntax is correct but it keeps resulting bad request 400

[sly grove]

Usually it will also give a message explaining what's wrong with your request

400 just means the request is invalid it doesn't necessarily mean a syntax error

[mellow sail]

Yeah, i already fixed it. The learning postman web is a live saver. I just found out there's a loop mistake in my code when requesting

But it's still c#, i think im going to migrate it to use unitywebrequest

[random shell]

Hello,
I'm facing a problem and I'm hoping someone can help me out. Here's the situation:

• Blue Arrow: Represents the transform's local forward vector.
• Red Arrow: Represents the camera's look direction with y = 0.
• Red dot: Represents the point towards which the transform should rotate.

I want the transform to rotate towards the red dot based on the movement input. For example, if the transform wants to move forward, I want the blue arrow to rotate towards the red arrow.To achieve this, I'm looking for guidance on how to calculate the rotation angle between the blue and red arrows. Additionally, if the transform wants to move downwards, I would like to know how to rotate the blue arrow in the opposite direction of the red arrow.Any help or suggestions would be greatly appreciated. Thank you!Please note that I'm seeking advice on the implementation details, so if you have any specific code examples or resources, that would be fantastic.Thank you in advance for your assistance!

[devout coyote]

Set the world rotation of the blue to the rotation value from your camera, if you need to inverse it, you can multiply it by a quaternion created from a rotation of 180 on the by you specified axis, although there is probably a utility method for this

[sly grove]

Here's one way.

Vector2 input = HoweverYouGetInput();
Vector3 input3D = new Vector3(input.x, 0, input.y);
Vector3 camForward = cam.transform.forward;
camForward.y = 0;
Quaternion camRotation = Quaternion.LookRotation(camForward);
input3D = camRotation * input3D;

// Now you move/rotate your character in the direction of input3D

[oblique lake]

Cheers, worked perfectly.

While my old way worked, it was giving me wacky values when I zeroed out the stick which I didn't want.

[tiny pewter]

to who delete his question
i think this is nearly impossible to solve unless you raycast from all points to all points in another Mesh (btw float is discrete so you can do this unlike real world) since every point has its own "sphere space" (idk how to explain it but you can imagine Celestial sphere), and it means no any universal mapping (to 2D space) can apply to the target mesh
i simply the 3d case a lot to 2d case (here is 2d to 2d point mapping)

i maybe wrong there maybe some transformations

[rapid osprey]

Sorry I thought my question was just too low quality (didn't have enough info). As you said, the number of raycast needed is crazy and I can't really reduce the amount because of edge case like these where partial part of the model is obfuscate.

Someone proposed checking the pixel inside a stencil buffer instead of ray casting. I would check on ~5 points in different angles and it should work on most case. Since the game is turn based even if it is expensive I can spread the work on multiple frames and recalculate once in a while.

[hazy bear]

I need to create exact copy of UnityEvent instance and tbh idk what to put in that method, propably some reflections which is always tricky to me

            var newEvent = new UnityEvent();
            for(int i = 0; i<Event.GetPersistentEventCount(); i++)
            {
                var target = Event.GetPersistentTarget(i);
                var method = Event.GetPersistentMethodName(i);
                var mode = Event.GetPersistentListenerState(i);
                newEvent.AddListener(); //It takes UnityAction which is some sort of delegate
                newEvent.SetPersistentListenerState(i, mode);
            }

Any help?

[dusty wigeon]

UnityAction stringFunctionToUnityAction(object target, string functionName)
{
    UnityAction action = (UnityAction)Delegate.CreateDelegate(typeof(UnityAction), target, functionName);
    return action;
}

https://stackoverflow.com/questions/47367429/is-it-possible-to-turn-a-string-of-a-function-name-to-a-unityaction

[hollow plinth]

Compute Shader Multithreading Question.
I tried to do my own research for this, but I can't seem to find the right keywords to use.
I'm still not very knowledgeable with both Compute shader and Multithreading, my current concerns is stuck in my head and preventing me from progressing.
I want to create a simulation of battles millions of times, where every 9 consecutive normal battle wins, the next battle is a boss battle. Is this possible to do in Multithreading?

[untold moth]

It's possible to do without multithreading too.

Multithreading implies running several threads that execute code in parallel, so you can increase the speed of something being processed if it needs to do a lot of the same logic.
That being said, most unity API can only be used on the main thread, so your simulation would need to be implemented without relying on unity.
Also note, that there are many pitfalls with multithreading, like race conditions.

As for compute shaders, they are similar to multithreading in the sense that there is a simple program that runs in parallel on many many small processing units. The difference is that it runs on a different device - GPU, so there are many restrictions imposed.

[tired basin]

i'm facing an issue with UnityWebRequest, i was using Unity 2021.3.15 before and content downloading for me was working fine, but after i upgraded to 2021.3.30, the same batches of WebRequests are getting time out errors. i can reduce the batch size but that makes the initial loading slow. but the catch is why this is causing issue on upgraded Unity, can anyone help me here

[dusky kettle]

hey guys im putting this here cos the math here is messing with my brain a little and i'm a little lost. I'm trying to create a system to roll a ball along a flat plane, but with a controlled outcome.
the part that I'm stuck on is i'm trying to create a function that will take in a starting position, ending position, ending rotation, and a radius, and return a quaternion representing the starting rotation that the ball would need to be at to roll without slipping or skipping and end at the ending rotation.
I was actually using chatgpt to help me with this because i'm pretty hopeless with this kind of math, and I got a pretty usable answer. this is what i've got at the moment:

public static Quaternion CalculateStartingRotation(Vector3 startPosition, Vector3 endPosition, Quaternion endRotation, float ballRadius)
{
    // Step 1: Determine how many revolutions the ball needs to make
    float distance = Vector3.Distance(startPosition, endPosition);
    float circumference = 2 * Mathf.PI * ballRadius;
    float revolutions = distance / circumference;

    Debug.Log(revolutions); //I added this to debug, this wasn't chatgpt's doing

    // Step 2: Determine the direction that the ball will rotate
    Vector3 direction = endPosition - startPosition;
    Vector3 axisOfRotation = Vector3.Cross(Vector3.up, direction).normalized;
    if (Vector3.Dot(Vector3.up, direction) < 0)
    {
        axisOfRotation *= -1; // Reverse the axis if needed
    }

    // Step 3: Determine the starting rotation
    Quaternion startingRotation = Quaternion.AngleAxis(revolutions * 360f, axisOfRotation) * endRotation;

    return startingRotation;
}```
This code will give me a rotation that rotates in the correct direction, but the amount of revolutions seems to be messed up. I tried it with some parameters where the debug.log outputted 0.8 and it somewhat looked okay, but then i tried it with around 2.2 revolutions and it rotated in opposite direction, less than 1 time.
does anybody know how I can fix this? do quaternions store revolution counts at all, or would I need to use eulerangles or something? i'm kinda at a loss

[dusky kettle]

interesting, i fed it parameters with a distance equal to the circumference so that the revolution was equal to 1 and the ball didnt rotate at all

yeah i cant figure out how to get it to return a rotation that doesn't wrap

which sucks cos im using dotween and id imagine that means theres no way to get dotween to rotate more than one full revolution

[upbeat path]

logically you should be tweening on revolutions

or on start and end positions

[dusky kettle]

what

[dusky kettle]

could you elaborate on that please

lmfao

i found something on google taht says its possible to do DORotate that takes eulerangles beyond 360

im wrangling with chatgpt to see if i can get it to understand what I now need but its frustratingly bad at doing things i'm bad at

[upbeat path]

tweening goes from 0 (start) to 1(complete). if you want revolutions to go from 0 to 2,2 for example then that is what you should be tweening on
likewise goes from start to end positions

[dusky kettle]

are you talking like literally with dotween

or just the concept of tweening

for something so seemingly simple im shocked that literally 0 solutions exist for this online from what ive found

[upbeat path]

why would there be a solution online for something simple?

and yes I'm talking about tweening in general

@dusky kettle How are you actually moving your ball?

[dusky kettle]

do you know of dotween

cos im using that

the use case is im trying to create a sorta scripted thing with a ball that has a face on it

i need the face to end facing a certain way

the way the ball moves is by rolling

[upbeat path]

?
If you're tweening the Quaternion how is that converted into movement?

[dusky kettle]

dont worry about that part

im just doing a second tween at the same time

thats not part of it anyways

right now the part i cant figure out is how to figure out that starting rotation and how to get it to rotate accurately

im just trying to script a ball rolling in a way that allows me to control the final position, rotation, and initial position

without accounting for anything that would complicate it like ramps or drops or whatever

[upbeat path]

Look it should be simple.

1. Set the start rotation from your start, end and end rotation
2. Tween between start and end positions
3. for each tween calculate current rotation based on distance traveled

[dusky kettle]

lmfao dude i know

thats the easy part

my problem was never the tween

ignore the tween part

the calculate part is what isnt working

thats the part that isnt simple

[upbeat path]

your problem is you are tweening the quatenion when you dont need to

[dusky kettle]

you're misunderstanding the situation

i need the ball to end at a rotation I need it to be at

the end rotation needs to be predetermined

[upbeat path]

yes

yes

[dusky kettle]

the start rotation needs to be backwards-engineered from that

[upbeat path]

yes

[dusky kettle]

this is not relavent then

[upbeat path]

so what have I misunderstood?

[dusky kettle]

and is not the problem

because atm my issue is that i dont have a functional way of determining what the starting rotation is

[upbeat path]

yes you do, the math given by Gpt looks ok to me

[dusky kettle]

its not

ive been testing it lmao

the axis is correct

but its not giving me a rotation that works

unless im mistaken, quaternions dont store rotations above 360 degrees in magnitude

they just wrap around

[upbeat path]

Ah, so quaternons dont work the way you want them to, my solution negates that problem

[dusky kettle]

you're going to need to give me an example mate cos im really not following what your solution is

[upbeat path]

OK,
Say we have a distance of 1 circumference
So the ball needs to rotate 360 degrees
If you tween between 0 and 1 for each step you calculate the rotation over that distance, then you apply that to your ball

[dusky kettle]

yeah nah but thats not solving the problem

the problem is I don't know if the ball needs to rotate 360 degrees

because the thing isn't calculating that properly

like sure i can get the revolutions

[upbeat path]

that was just an example, it will work for any distance

[dusky kettle]

but it's not returning it in terms of the axis it needs to rotate around

okay this is just going in circles rn

[dusky kettle]

the problem still stands here

does anybody know how I might be able to get a change in rotation based on a starting position, ending position, ending rotation, and ball circumference, given that the ball is rolling without slipping on a flat plane

[upbeat path]

Ah, I see your point, what you need is to a return an array of Rotations where each entry is 1 full turn of the ball

[dusky kettle]

thats an extremely janky solution dude

it would be a lot better to just return one vector3 with numbers that exceed 360 in magnitude

[upbeat path]

no, it's a game dev solution

[dusky kettle]

brother what are you on about

[upbeat path]

it's one that
A) Works
b) is easy to implement

[dusky kettle]

look if you wanna try solve it yourself taht way, be my guest, but I am not doing it your way, and taht also still has absolutely nothing to do with the problem here

ijust need help calculating those numbers in the first place

not figuring out a data type to return the result

[upbeat path]

best of luck with that, considering you had to use GPT in the first place, bye

[dusky kettle]

😐

[dusky kettle]

bump for anybody else who might be able to help out please

[untold moth]

Not sure I understand what you're asking.
You can get the difference in rotation in angles by subtracting the initial rotation from the final one.

[dusky kettle]

yeah im just ttrying to calculate that initial rotaiton

reverse engineer it from the start position, end position, end rotation, and ball circumference, assuming the plane is completely flat

im essentially trying to animate a ball that rolls lmao

[untold moth]

Oh, if you have the final state and the angular velocity, you can calculate it. Assuming the angular velocity is constant. You can also calculate the angular velocity from regular velocity and circumference, but you'll need to google the exact formula.

[dusky kettle]

the thing is im not using rigidbody at all

[untold moth]

Who said anything about rb?

[dusky kettle]

i can if necessary

oh so you just mean like

hmm okay

i am using tweening libraries tho

so the trajectory isnt constant

the point is that im not trying to calculate it based on something that happened, rather trying to make it happen so that it ends exactly as I need it to

[untold moth]

If it isn't, then you'll have to keep track of all the velocity changes.

[dusky kettle]

do I?

if it takes the shortest available path from a to b

the rotation of the ball should never change

rather

the rotation over the path should never vary

assuming the ball never ceased contact with the plane it was rolling on

[untold moth]

Yes, if you don't, you don't have enough info on what happened to the object, at which point you can't calculate it's initial position without guessing.

I'd assume that's why Steve mentioned storing stuff in array.

[dusky kettle]

in fact

even without taking the shortest path

no nevermind

that wouldnt be the case

[dusky kettle]

no you're not calculating the initial posiution

you give the initial position

[untold moth]

In this case the angular velocity is constant, and you should be able to calculate the initial state.

[dusky kettle]

you only are calculating the initial rotation

the intial position, final position, and final rotation are all given

it returns the initial rotation

[untold moth]

So you have initial position and velocity?

[dusky kettle]

velocity isnt really relevent here

time wouldnt affect it

unless the ball is slipping

which is isnt

[untold moth]

I don't understand why you have initial position, but not rotation. Is the ball rolling backwards?

[dusky kettle]

i want to chose where the ball ends

as i said

i am trying to "animate" a ball that rolls realistically

so i need it to end in a certain spot with a certain rotation and position

and I need it to start in a certain spot

it doesnt matter what the initial rotation is

just the ending rotation

and i need to do this frequently enough to justify writing code that will do it for me

[untold moth]

Why can't you just rotate it realistically every frame?

[dusky kettle]

what

wdym

[untold moth]

Just rotate it every frame a little bit.

[dusky kettle]

thats not the problem

time is completely irrelevant to the problem

nor the implementation

[untold moth]

What does it have to do with time?

[dusky kettle]

just ignore all the tweening part

my problem is I need to figure out how to get its initial rotation

or the change in rotation from the intial rotation

that one actually

i think i need that one more

in a formwhere the magnitude is still 360 degrees times the amount of revolutions the ball did while rolling

[untold moth]

Why are you so stuck with the "initial rotation"? If you have a position and rotation in one point and you want to get the rotation and position in point b, you can just tray point a as the initial state, and point b as the final state.

At which point it's the same calculation as if the ball is moving from a to b.

[dusky kettle]

okay sure

you can flip the points around

the problem stands

i need to figure out the rotation delta of a ball from a point to another point only knowing the rotation at one of those points

[untold moth]

Then it's what I mentioned before.

The ball moves the distance of it's circumference in one 360 degree rotation. There's that relation between distance moved and rotation of the ball. For a more detailed formula, you'll need to google.

[dusky kettle]

oh i see what you're saying

i wonder if thats any easier for me though

id still like the number as a representation of a rotation that has the ability to be greater than 360 degrees

idk if quaternion is capable of that

and idk how to do that math with euler angles

[devout hare]

Quaternion doesn't store the amount of rotations. But I don't see why you'd need that.

[dusky kettle]

for this i do need that

cos i need to figure out how much a ball rotates when it rolls from point a to b

but now that i can word it a lot simpler like that

google may have a solution to it already

[devout hare]

You'll have to store that information to something else other than a quaternion

but if you start with the correct rotation and rotate the ball as it moves, there shouldn't be any need to know beforehand how many times it will rotate

[dusky kettle]

i was thinking just as a euler angle that exceeds 360 in magnitude

[dusky kettle]

for my case I need to know the final rotation though

because i'm going to flip it around

and make the final rotation the initial rotation

since what I actually need is to control the end rotation and not the start rotation

[upbeat path]

@dusky kettle Just in case you're interested, I implemented my solution, it's not perfect but it works
https://pastebin.com/LzNanN8m

[wraith storm]

Will this work?

Vector3 positionStart;
Vector3 positionEnd;
float ballRadius;

float rotationAmount = ( positionEnd - positionStart).magnitude / ( 2f * Mathf.PI * ballRadius ) * 360f;

[fast falcon]

Hello, I'm not sure if this question would be better asked here or in #archived-urp, but I'll ask here first:

I am trying to copy the rendered texture of a camera while still letting the camera render to the screen. Rn I'm using a script on the camera object to call Graphics.CopyTexture in OnPostRender() to copy RenderTexture.active to a different render texture.

But for some reason, this has inconsistent results. It works great in play mode in the editor window, but when I build the project, the render texture is blank.

Does anyone know why this might be happening, how I could fix it, or a different way to copy a camera's output to a render texture?

[bleak citrus]

I think this is a more general problem. I remember seeing this exact thing happen before

Something wasn't being done automatically in the build

I do not remember what it was. Maybe you didn't create the render texture properly?

maybe it was related to not calling Create?

I'm looking at the project where I had this issue. it was a game jam that had a total of 5 commits

so, can't exactly pull the answer out of that...

[misty glade]

Documentation is a little light on Tilemap.cellBounds and I can't seem to figure it out (https://docs.unity3d.com/2023.2/Documentation/ScriptReference/Tilemaps.Tilemap-cellBounds.html). I've made what appears to be a sparse (contains a couple empty places) 13x12 grid but bounds is returning 16x15 with an origin of -4, -4. It's making it a bit hard for me to recreate it, but maybe I'm doing it wrong. Screenshots of the tilemap and the same map inside a Grid looking how I want it to. The lower left grass tile should be (0,0) and in the grid, appears to be so..

I was thinking I'd make a prefab of the "map" (the Tilemap) and then recreate it at runtime by doing a Tilemap.GetTilesBlock() and then iterating it and setting tiles of the runtime map based on what's in the prefab map

(also the array returned by GetTilesBlock() has nulls in it, but the docs seem to indicate it shouldn't? compared to GetTilesBlockNonAlloc()

hm.. nevermind.. google to the rescue..

[dusky kettle]

it’ll calculate the the amount of degrees the ball needs to rotate yeah, but the part i’m stuck on is then getting that as a rotation around the correct axis

[dusky kettle]

i’ll give this a shot when i get home, thanks for sending

[untold moth]

You can get an axis perpendicular to it's movement direction with cross product.

[dusky kettle]

will that go over 360 degrees tho

actually ig it doesn’t matter

cos i can just add an extra 360 degrees worth of rotation based on the amount of revolutions right

[untold moth]

Axis is a direction vector, not a rotation angle.

[dusky kettle]

so then how would i turn that into a vector3 representing the same rotation delta

[untold moth]

You don't, but you can use it to rotate an object around that axis .

[dusky kettle]

for the sake of dotween i think it wants me to do it

in theory it’s as simple as multiplying the normalised axis direction vector though

[untold moth]

You don't need to rely on dotween. In fact I'm not sure that you can use it for your use case.

[dusky kettle]

i don’t have to but i’d like to be able to easily chose the ease type

since it’s meant to happen synchronously with the actual movement

actually i suppose i can like

use dotween to get a value from 0 to 1

and then just map the movement along that

that might’ve been what steve was trying to say earlier

[untold moth]

Yeah, I think you're overcomplicating the solution...

Perhaps it would've been simpler without dotween at all.

[dusky kettle]

dotween gives me free ease types tho

[untold moth]

You can have ease without dotween.🤷‍♂️

[dusky kettle]

regardless i don’t see how dotween is complicating it more

[dusky kettle]

yeah but you don’t have those giant selection of choices tho

there’s like a billion ease types

[untold moth]

Then I guess this would be the most appropriate solution.

[dusky kettle]

yeah

idk i’ll have to actually try it

dot and cross product and all that are hella confusing

[dusky kettle]

like

for this is it just

[steel snow]

is it possible to display meshes without making game objects ?

since im constantly adding/removing them wondering if i can just skip all that extra overhead

[dusky kettle]

cross product of vector3.up and the endPoint relative to the startPoint?

[untold moth]

You can probably issue a draw call manually, but I don't think it's gonna be much different from just enabling/disabling a renderer.

[steel snow]

i presume i would have to call it every frame in update for it to remain seen meaning i would still need a gameobject?

i guess at least i would only need one mesh manager

[untold moth]

It takes 2 direction vectors, so endpoint - start point, up.

[dusky kettle]

yup

i can try that

[untold moth]

Yes, but you don't need a separate GameObject for it. You can issue these commands from your game manager or something.

Still isn't worth it imho.

Well, depends on the context I guess.

[steel snow]

it would simplify my code in my case because i have to instantiate child objects and delete them all the next frame

or at least every time i mouse the mouse

[untold moth]

Why not use some kind of pooling system?

[steel snow]

well - more code to write for meshes that only there temporarily

only once i finally click i solidify them in place

[untold moth]

I mean, issuing a draw call is not a one line thing either...

[steel snow]

https://docs.unity3d.com/ScriptReference/Graphics.RenderMesh.html

i thought that was what this does

[untold moth]

Hmm

[steel snow]

since i have a List<Mesh> alongside List<Material> i was thinking of using it

[untold moth]

I guess there's this then. Still need to calculate the world matrix and stuff.

[steel snow]

oh true

i presume unity's mesh renderer does this under the hood

[untold moth]

Yes. This and many other things.

Seems like you can create that matrix with utility methods though.

[steel snow]

i dont even know when these functions became a thing

unity just sneaks them in and tells no one 😄

[untold moth]

I think they've been here for a while. Just never had a need in using them.

[steel snow]

ah i see one downside it wont cull as it has no bounds

[round jay]

Hello i have a question, i'm building an app for Hololens2 in order to use UWP library in my code i need to switch my Scpriting Bakcend from IL2CPP to ".NET" i'm getting in to Project Settings > Player Other Setting> Scrpting Backend but it's grayed and i can't switch. Is there a way to change this ?

[dusty wigeon]

The platform you have selected does not support Mono. (Scripting Backend is between Mono and IL2CPP)

Unity has two scripting backends: Mono and IL2CPP
(Intermediate Language To C++), each of which uses a different compilation technique:

Mono uses just-in-time (JIT) compilation and compiles code on demand at runtime.
IL2CPP uses ahead-of-time (AOT) compilation and compiles your entire application before it is run.

https://docs.unity3d.com/Manual/scripting-backends.html

[round jay]

So yeah it's busted

I'll go by sharepoint

[hazy bear]

I need a little bit of help from some Reflection magician if possible :/
I have Type type variable and Action callback variable. I need to create a delegate of type EventCallback<> of generic type type, that will invoke callback. Im experimenting a bit and getting somwhere but tripping all the time...
I came up with this idea, but I got exception: methods arguments are incompatible :/

            Action callback = () => {
                Debug.Log("lol");
            };
            Type eventType = Type.GetType(eventInfo.EventType); //eventInfo.EventType is a string with AssemblyQualifiedName
            Type eventCallbackType = typeof(EventCallback<>).MakeGenericType(eventType);
            Type genericExecuterType = typeof(GenericEventExecuter<>).MakeGenericType(eventType);
            object executer = Activator.CreateInstance(genericExecuterType, new object[] {callback});
            var callbackDelegate = genericExecuterType.GetMethod("HandleExecution", BindingFlags.NonPublic | BindingFlags.Instance)
                .CreateDelegate(eventCallbackType, executer);

and my dummy class

    internal class GenericEventExecuter<T> where T : EventBase
    {
        private Action callback;

        public GenericEventExecuter(Action callback)
        {
            this.callback = callback;
        }

        private void HandleExecution<T>(T type)
        {
            callback.Invoke();
        }
    }

nahh... sry my bad apparently it works, i have a problem somewhere else

I think some ui event types holds additional info on which depends the execution or something, and with this generic approach they dont work

like ClickEvent (which is left mouse button) work, but ContextualClickEvent (which is right mouse button) doesnt

because what I do is

            Action callback = () => {
                Debug.Log("lol");
                if (enabled) eventInfo.Event?.Invoke();
            };
            Type eventType = Type.GetType(eventInfo.EventType);
            Type eventCallbackType = typeof(EventCallback<>).MakeGenericType(eventType);
            Type genericExecuterType = typeof(GenericEventExecuter<>).MakeGenericType(eventType);
            object executer = Activator.CreateInstance(genericExecuterType, new object[] {callback});
            var callbackDelegate = genericExecuterType.GetMethod("HandleExecution", BindingFlags.Public | BindingFlags.Instance)
                .CreateDelegate(eventCallbackType, executer);


            MethodInfo registerCallbackMethod = typeof(VisualElement).GetMethods()
                .Where(x => x.Name == "RegisterCallback")
                .FirstOrDefault(x => x.IsGenericMethod)
                .MakeGenericMethod(eventType);
            registerCallbackMethod.Invoke(target, new object[] { callbackDelegate, TrickleDown.TrickleDown }); //where target is some visual element

or maybe someone has better idea why it doesnt work for some events and how to resolve it?

[misty glade]

Event.EventType isn't a string, it's an enum, although I'm not sure if that's your problem, but I imagine it might be worth verifying that Type.GetType works like you expect it to on an enum

although what you're doing seems ... weird. What are you trying to actually do..?

[fast falcon]

Calling create? Rn the project uses a render texture I made in the assets folder. Do I need to create one at runtime?

[bleak citrus]

ah, I don't think that's needed..

[fast falcon]

I just tried calling RenderTexture.Create() in the awake method, but that doesn't seem to fix the issue

It works if the camera renders directly to the render texture, but then it won't render to the screen.

[fast falcon]

Okay, I've done some more testing and copying a different render texture in the OnPostRender method works, and calling RenderTexture.active.IsCreated() returns true. But for some reason I can't copy that specific texture when the game is actually built.

[hazy bear]

eventInfo is actually my custom type so I know whats inside XD
Im actually trying to, lets say, expose UI events in inspector in form of UnityEvent

[misty glade]

gotcha

Thought you were maybe just hooking into IPointerDown etc events and maybe there was some enum to string implicitly happening that was messing up your reflection

[hazy bear]

dont get me wrong, I actually managed to do it and it works sweet, but only some specific events dont seem to work, but ill try make custom wrappers that register them in a way they supposed to be registered (I mean with no reflections) and then expose them or smth (idk yet)

even inspector itself looks clean for which im proud of

[dusty wigeon]

There is already a ton of solution on github. You could just snatch one or copy one.

https://github.com/Baste-RainGames/SerializableAction_Unity
https://github.com/Siccity/SerializableCallback
https://github.com/wlgys8/SerializableEvent

[hazy bear]

yeah that could be cool extension to what I already have, but I dont mind using native UnityEvent, and it isnt a big problem to replace it in the future if I need to, the whole problem revolves around registering callbacks to UI elements with reflections

[dusty wigeon]

I am pretty sure I sent you the solution #archived-code-advanced message. What was not working ?

[hazy bear]

oh this, sorry i forgot that I ever asked this question XD Thanks for the solution but later I understood that if I want to have "duplicate" funcionality for UnityEvent in inspector this would be much more work, but now Im thinking that this is a very good excuse to use one of these alternatives you have sent

[hazy bear]

copy created in such manner isnt serialized, and there is more data needed to be copied that isnt straight forward accessable, either more reflections or fooling around serializedObject

[hazy bear]

but one of these may be easier to copy, but this is definitely not a worry for now for me

[dusty wigeon]

If you want serialized you want to use https://docs.unity3d.com/ScriptReference/Events.UnityEventTools.AddObjectPersistentListener.html

https://stackoverflow.com/questions/40655089/how-to-add-persistent-listener-to-button-onclick-event-in-unity-editor-script

[hazy bear]

I love and hate how many such helpful methods are hidden somewhere in a little classes no one knows about...

[dusty wigeon]

Most of them are hidden behind a google search though. But yeah, they are really spread between a lot of class. Keep an eye on those "Utility"/"Tool" class, Unity likes to have a lot of them for the editor.

[tawdry plinth]

Hi, is there a way to get gameobject before their awake call for each new scene?

[sly grove]

"get" it for what purpose

theoretically script A can reference an object in its own Awake before some other script B runs, if that's what you mean

[tawdry plinth]

for injecting dependencies

[sly grove]

One way would be to use the Script Execution Order settings and make sure your dependency injector runs before everything else

Another way is to explicitly call MyDependencyInjector.InjectFields(this); in Awake for each script that needs things injected

[tawdry plinth]

it works for first scene , what about others which will be loaded later

[sly grove]

it will only work if the dependency injector is in each scene, not DDOL. If it IS a DDOL, you could have a stub object in each scene which just invokes the dependency injector in its awake using script execution order settings

e.g.

public class DIStub : MonoBehaviour {
  void Awake() {
    DependencyInjector.Instance.InjectFieldsForScene(this.gameObject.scene);
  }
}```

[tawdry plinth]

thanks

[dusty wigeon]

Callback invoked when the first scene's objects are loaded into memory but before Awake has been called.

https://docs.unity3d.com/ScriptReference/RuntimeInitializeLoadType.BeforeSceneLoad.html
https://docs.unity3d.com/ScriptReference/RuntimeInitializeOnLoadMethodAttribute.html

[sly grove]

That's only for the first scene though

Sadly

[dusty wigeon]

Yeah, didnt see that it needed to be done for each scene. I'm using it for Singleton Init, though it could have been the same use case here.

[shrewd hemlock]

https://hastebin.skyra.pw/jizutuqaqi.java Hi I wanted to ask again maybe here this time. I currently have this code from a tutorial, and I tried to apply it to a 3D model's torso and I was wondering how I would change this such that the torso properly follows the mouse?

[shrewd hemlock]

An image for my previous comment

[untold moth]

Does it not follow it now?

*rotate towards it.

[visual hollow]

Hello! I'm trying to add depth support to Rigidbody2D, so it can ignore collisions with colliders on different Z levels. What I have is a script that controls a kinematic Rigidbody2D by checking collisions with a contact filter and moving the body via MovePosition. In the screenshot is my test with two BoxCollider2D using my script. Following is my code.

using System.Collections.Generic;
using UnityEngine;

namespace Game
{
    [RequireComponent(typeof(Rigidbody2D))]
    public class CustomRigidbody2D : MonoBehaviour
    {
        private Rigidbody2D rigidbody { get; set; }

        private Collider2D[] overlapColliders = new Collider2D[8];

        private void Awake()
        {
            rigidbody = GetComponent<Rigidbody2D>();
            rigidbody.bodyType = RigidbodyType2D.Kinematic;
            rigidbody.gravityScale = 0;
            rigidbody.interpolation = RigidbodyInterpolation2D.Interpolate;
            rigidbody.collisionDetectionMode = CollisionDetectionMode2D.Continuous;
        }

        private void FixedUpdate()
        {
            ContactFilter2D contactFilter = new ContactFilter2D
            {
                useDepth = true,
                minDepth = transform.position.z,
                maxDepth = transform.position.z,
            };

            int overlapColliderCount;

            if ((overlapColliderCount = rigidbody.OverlapCollider(contactFilter, overlapColliders)) > 0)
            {
                for (int i = 0; i < overlapColliderCount; i++)
                {
                    Collider2D otherCollider = overlapColliders[i];
                    Rigidbody2D otherRigidbody = otherCollider.attachedRigidbody;
                    ColliderDistance2D colliderDistance = rigidbody.Distance(otherCollider);

                    if (colliderDistance.distance <= -0.01f)
                    {
                        Vector2 force = 0.25f * colliderDistance.distance * colliderDistance.normal;
                        otherRigidbody.position -= force;
                        rigidbody.position += force;
                    }
                }
            }
        }
    }
}

The question is: how can I improve this to make the collision mimic the behaviour of a regular Rigidbody2D (no overshooting, velocity transfer, drag, etc)?

And is there another way to make a Rigidbody2D care about depth?

My end goal would be to have a single tilemap with multiple levels (z position) and enemies across all levels that can only collide with enemies in the same level.

[untold moth]

Why not just use 3d physics if you need depth?🤔

[visual hollow]

Because tilemaps don't support 3D colliders, and 3D physics can't interact with 2D.

[untold moth]

Then just use a different collider for the tilemaps. And if you use 3d physics, then you wouldn't need it to interact with 2d physics.

[visual hollow]

My game is procedurally generated, so generating collider meshes on the fly is kinda expensive.

[untold moth]

Otherwise you're just reimplementing 3d physics in an awkward way.

[untold moth]

While it's not cheap, it's totally possible.
Minecraft managed that somehow. And it was written in java.

You can use jobs or compute shaders for the heavy processing if needed.

[visual hollow]

I was going that way, but it smelled like overengineering, so I'm checking if there's any easier solutions. I don't need a full custom rigidbody implementation. Non-overlapping collisions would be enough to me actually.

[untold moth]

Well, you don't have a way to do that without relying on layers(only 32) or butchering the collisions and stuff.

[visual hollow]

Perhaps I could do a simple custom CharacterController2D using raycasts and tile checks.

[untold moth]

If you want to handle collisions manually, you can just use triggers.
There are many ways to approach it, depending on how much of the physics functionality you want to be working.

[visual hollow]

This is a test with 3D tilemap colliders I made.

Will try to optimize it with Jobs, just need to understand it better firstly.

Thanks for your help!

[next marsh]

I'm getting an error with the Burst compiler:
G:\Unity\Space-Gaem-Aw-Yeah\Assets\Scripts\RTS\Physics\Manager.cs(110,13): Burst error BC0102: Unexpected internal compiler error while processing function `IL_0008: call System.Void System.ReadOnlySpan`1<UnityEngine.Vector2>::.ctor(!0[]) args(IL_0001(ldloca.s), IL_0003(ldsfld))`: System.NullReferenceException: Object reference not set to an instance of an object

Manager.cs (110, 13) being:

[BurstCompile]
private static void _step(float timeIncrement, int count)
{
    var velocitySpan = new ReadOnlySpan<Vector2>(Velocities);

Velocities being:

public static class Manager
{
  private const int BodyCap = 150;
  private static readonly Vector2[] Velocities = new Vector2[BodyCap];
...```

Here's the whole trace log:

Using a couple day old install of Unity 2023 (2023.1.14f1)

Seems to compile fine when I comment out the [BurstCompile] tag on the function call.

[sage radish]

Do you need to read the data from that array or are you just using it as a reusable temporary buffer for your method?

[next marsh]

Just reading from them

Oh you mean the array Velocities?

[sage radish]

Yes

[next marsh]

That's being read/written too extremely often

[sage radish]

But do you read the data that was previously in it, or are you overwriting it immediately?

[next marsh]

Oh damn wait I see what you mean

It didn't occur to me I was writing to the array I was spanning

Thank you-- I can't test now but I'll let you know

[sage radish]

Burst doesn't support managed arrays, unless they are static readonly, in which case it will interpret it as a constant array, which is evaluated at compile time.

[next marsh]

Ahhhh

Well damn

Long-lifespan native array it is?

Wait can I span a native array?

[sage radish]

Yeah. You could still wrap a span around that though.

[next marsh]

Oh wow

Thank you

[sage radish]

But I would still create the span outside the method and pass it in

[next marsh]

As far as I can tell Burst doesn't like Spans being passed in

Lemme retest

[sage radish]

Hmm, you might be right

[golden lodge]

Hello I'm working with asperite importer, how can I enable a layer midway through my animation,

[shell horizon]

does anyone know if there's a way to directly get the array of values for a given key in a native multi hash map?

without enumerating

[sage radish]

That's not possible because there isn't an array for each key. It's stored as a linked list, so it has to be enumerated.

[shell horizon]

Ahh, I see. Thanks

[tawdry plinth]

foreach (var (key ,service) in _services)
{
if (service != null)
continue;
_services.Remove(key);
Debug.Log($"Removing service of type {key}.");
}
context : key is of class Type and service is c# object
I'm removing null services but it's not working, i'm sure that the service of Type Key where is a monobehaviour is destoryed.
After this code when i access this service this gives me Missing Ref exeption any idea about this ?

[upbeat path]

what type is _services?

[tawdry plinth]

private readonly Dictionary<Type, object> _services = new Dictionary<Type, object>();

[upbeat path]

you cannot remove an item from the object of a foreach anyway

[tawdry plinth]

ok i will fix that but the issue is why it is not null when it's destroyed

[sly grove]

it's because you have the value type as object

it will NOT do unity's special check for destroyed objects

If they're all MonoBehaviours you can make it Dictionary<Type, MonoBehaviour>

worst case: Dictionary<Type, UnityEngine.Object>

[upbeat path]

or just UnityEngine.Object

foreach (Type key in _services.Keys.ToArray()) {
if (((UnityEngine.Object)_services[key]) == null) // do something

[tawdry plinth]

@upbeat path @sly grove thanks

[real grove]

basically

You have to

[upbeat path]

I have to what?

[coarse compass]

So I was working on a property drawer for an interface that gives me a dropdown of all implementing types. It's as simple as making a field like this:

public class MyMono : MonoBehaviour
{
    [SerializeReference]
    public IMyInterface Obj;
}```

The problem is that I would like to be able to reference components and scriptable objects as well, but unity doesn't like it when you use any UnityEngine.Object with SerializeReference. 

Is there any way around this? I would really like to get this to work with ANY implementation of the interface without restriction, and without making a wrapper type.

[austere jewel]

No. I have not seen, nor can I think of a workaround that doesn't involve a wrapper type

[coarse compass]

Unfortunate. Thanks!

[dull epoch]

Hey everyone.
I am trying to write a shader for the Terrain to make it appear Blocky like it is made out of Voxels. But i will face problems with the Terrain Collider if i do that.
Is there a clever way to overwrite the Terrain Collider ? Since i know the maths where which block should be, it should be easy to bring that into a terrain collider ... right ?

[crisp parcel]

Heya! Question on compute shaders
To bugfix my code, I ported the shader code over to cs, by merely replacing the math calls (add .math), and keeping the rest (practically) intact

Any idea why the cs output and compute shaders output may differ, by LARGE amounts? I've triple-checked my code, diff-checked it twice, but can't find any misinputs or changes in code between both

(Note: I can't show the code, but can list which functions I am using: anything quaternion related (using hlsl quaternion code / manually ported version to cs), normalize, clamp, dot, length, abs, a bunch of acos/sin/... , lerp, /

[untold moth]

At this point you better of implement your own terrain system that generates the blocky mesh.

[dull epoch]

yeah i was afraid of that answer :/

[untold moth]

Maybe the issue is with how you get the results or interpret them?

It's really hard to help with that without seeing the code.

[crisp parcel]

True that.. I'll try a bit harder tomorrow, and if stuff doesn't work I'll create a simplified script to show-off the problem

[hoary merlin]

Hey there, working on a mod for a game (Yes the game supports mods and Harmony)
I'm looking to add a file browser to my mod allowing the user to import files.

I've tried using OpenFileDialog and SaveFileDialog, whilst these work, they only seem to work when the game is running in windowed, and not when the game is running fullscreen (Otherwise these open behind the game window) so I need to have something that will run directly in the game.
I found this Runtime File Browser which works absolutely brilliantly when using the editor, although I'm unsure / unable to port this to a mod as scripts can't be added to AssetBundles.
When seperating the scripts and AssetBundles I get the following error The referenced script on this Behaviour (Game Object 'FileOperationConfirmationPanel') is missing! (This is one of many)
Obviously I could manually hardcode it to fix all these script errors, but there are hundreds of them.

I'm looking to see if anyone here has a solution to this problem, or if I'm going to need to build a file explorer from scratch. Thanks!

[junior stone]

Hi! Is there a good way to keep track of all the Renderers in a scene, even though gameplay code I dont control might add or remove them randomly?

At the very least is there like an event I can listen to, to flag something in the scene has changed and I should do another search for all renderers?

[wraith storm]

Hi, I've been using this asset for, hmm, maybe 4 years now :
https://assetstore.unity.com/packages/tools/utilities/file-browser-pro-98713

I can't really give you an opinion on how it compares to the competition, I've not really thought much about it since I started using it. It just works I guess 🙂

[wraith storm]

You can find all components using :
https://docs.unity3d.com/ScriptReference/Resources.FindObjectsOfTypeAll.html

But I think this will also return prefabs in your assets folder, or something like this, you might need to filter the results somehow to make sure you are only looking at Renderers in your scene maybe.

[junior stone]

Thanks, thats what I do, but it has some pretty crazy overhead

[wraith storm]

Would Tagging be a solution in your project?

If you made a tag and then made sure all your Renderer gameobjects had that Tag, then you can find them super fast using GameObject.FindGameObjectsWithTag()

[junior stone]

so I was wondering if there was another strategy. Normally I'd be like 'Hey I can just add/remove them from a list as they get created', but I dont see a way to do that

I guess the real problem is that Im writing a SRP thats meant to plug into other games, so Im limited on my ability to demand they do anything

[wraith storm]

I'm not sure of any other method, other than have a root object in your hierarchy, and then transform.GetComponentsInChildren<Renderer>()

Maybe you could maintain a list of all root gameObjects, and then use the GetComponentsInChildren on each of those?

[junior stone]

yeah thanks, I thought it might be the case.

[wraith storm]

Hmm, another option might be to make a sub class of MonoBehaviour, called something like MonoBehaviourRendererTracked

And then in the Awake() method, you just check to see if there is a Renderer and if so add it to a static List.

But that would only work if the Renderer is there already when Awake() is called

[wraith storm]

What's the use case for you needing a faster way to find the Renderers? Maybe there is another solution that would work for your project.

[junior stone]

I have a SRP, and it needs to feed data to materials for a forward renderers materials like the number of lights

so essentially I have to maintain those relationships every frame

[regal olive]

for people that host packages on private github repositories (say something like com.mycompany.core), what solution did you go with to be able to have unity successfully find the repository? Since I'm going to be working with multiple people off and on I'm not crazy about either managing SSH keys or including some credentials file, but I'm wondering if there a better solution to get around this.

[dusty wigeon]

Usually, Unity does not needs to "find the repository". It does not even knows that it has one. You just send the link of your repository and make sure that other members have the appropriate permission.

[regal olive]

when I try this I keep getting the error that the git repo can't be found, even though I've double checked the path and everything. Is there something I'm missing in the setup here?

[dusty wigeon]

That standard Git issue I would assume. Simply Google it and you gonna find a gazillion possible issue. (Remove Unity from your research)

[regal olive]

ok yeah looks like that was it, now that the repository is being found, any idea how to resolve this error? An error occurred while resolving packages:
Project has invalid dependencies:
com.systemsheavy.core: Version 'https://github.com/systemsheavy/sh-unitypackages?path=/Assets/com.systemsheavy.core' is invalid. Expected one of: a 'SemVer' compatible value; a value starting with 'file:'; a Git URL starting with 'git:' or 'git+', or ending with '.git'.

is the path messing things up here?

[stuck plinth]

yeah i don't think you can reference a subpath like that in the manifest? fwiw i've always found it easier to add dependencies as git submodules, you can add them straight into the Packages folder - if the manifest is at the root it just finds them automatically, otherwise you can add them to the manifest as "embedded" and set the path that way

[solar condor]

Heya - techartist here:
What is a good use case for Texture2DArrays instead of multiple Texture2D's? I kind of see how TexArrays are powerful, but I'm having a hard time thinking about how to harness the flexibility of the index-based sampling on top of uv's in a way which isn't just just "read from multiple textures that are actually in the same file now!"

Or is that particular feature that much of a performance gain?

[sage radish]

If the texture you need to sample is dynamic, that is you have to calculate the index in shader, then Texture2DArray is very convenient. The alternative is... what? A big switch statement?

[solar condor]

What do you mean with dynamic? Like, something regenerated at runtime?

[sly grove]

In a shader no less 😰

[sly grove]

More like if your character has several states and you need to know which state to sample based on what's going on currently

[sage radish]

Maybe you want the position of the pixel to determine which texture to use, or the vertex color, or something else that you can't know ahead of time.

[solar condor]

hmmmm, I see

[sage radish]

If the number of textures aren't too crazy, then you can get away with just sampling all the textures and lerping between them. That's usually how triplanar mapping works. But that doesn't scale very well.

[plucky trellis]

They're useful for when you have an unknown number of textures, which can change at runtime, like shadowmaps for example. But the limitations of all textures in the array having to be the same size leads a lot of people to just make a texture atlas (which has it's own problems, especially with filtering, but whatever)

[solar condor]

Wait, isnt triplanar mostly for sampling the same texture, but across different uv inputs?

[solar condor]

An unknown number of textures which can change? Do you have a game example where this is done? Or is this in a similar line to what Praetor explained?

[sage radish]

You can sample the same texture three times, but you can also sample two or three different textures, which is not uncommon, such as for drawing a grass texture on top and a cliff texture on the sides of a terrain.

[solar condor]

Ohhh, I see

[sage radish]

It should become obvious to you if you would benefit from using Texture2DArray, because the alternatives will be very inconvenient.

[languid isle]

HI everyone!

Are objects in an array "lazy loaded"?

I need help because I need to make certain I'm not creating a resource train wreck in the future

[sage radish]

If you're talking about serialized arrays, then no. There's no lazy loading in Unity's serialization.

[languid isle]

Ok then I have a problem.

I'm making a board game where people can change the cells and pieces of the board in mid game. I was told that it might suck up resources if I put all of the graphic sets on a piece. I then said : "Ok, can I put the sprite renderer with different graphics on each object, and then fill the array with objects." They said they were unsure if all the graphics would be loaded or lazy loaded. I came here because they had to leave (it's like 3am their time), and now you answered the question.

So it sounds like they are all fully loaded and not lazy loaded.

So what would be the best way to have those graphic packs available for a player to swap in mid play without loading up 300 graphic packs and models into memory?

I was also warned off using a resources folder.

[sage radish]

Resources folder is also not lazily loaded. It's not recommended because everything in it will always be loaded at startup and remain loaded throughout the game.

[languid isle]

ok

[sage radish]

Unity has the Addressables package which can solve this problem.

[languid isle]

ok I know nothing of this, can you pls explain?

[sage radish]

The idea is that you build your assets into separate bundles, which you can then load lazily at runtime. Addressables makes it a bit simpler to load these bundles and find the asset you want from them.

I don't have much experience with it, so I'm not the best person to ask.

[languid isle]

oooo

I will send this info to my tutor

[tender helm]

Does anybody know what changes are in Burst 1.8.9? The changelog is blank; https://docs.unity3d.com/Packages/com.unity.burst@1.8/changelog/CHANGELOG.html

[finite pond]

Is there a way to access a SpriteRenderer's size inside Shader code?

I feel like I remember using it sometime in the past, but I can't find any info about it nor do I remember how I did it

[hoary merlin]

Hey there, currently working on an IMGUI menu, and I'm having some issues with usingGUILayout.BeginArea inside GUILayout.BeginScrollView

contentScrollPosition = GUILayout.BeginScrollView(contentScrollPosition, false, false, GUIStyle.none, GUI.skin.verticalScrollbar);
int count = 0;
foreach (var directory in currentDirectoryInfo.GetDirectories())
{
    GUILayout.BeginArea(new Rect(0, count * (height / 15), width, height / 15), Mod.Bundle.LoadAsset<Texture2D>("folder-icon"));
    if (GUILayout.Button(directory.Name, _contentButtonStyle, GUILayout.Width(width), GUILayout.Height(height / 15)))
    {
        ChangeDirectory(directory.FullName);
    }
    GUILayout.EndArea();
    count++;
}
GUILayout.EndScrollView();

At the moment this is the setup I've got - my goal is to have an icon alongside the button (but also allow the icon to be clickable) similar to the attached screenshot. Although the issue is, when using BeginArea inside BeginScrollView, the scroll view doesn't get populated, therefore there's nothing to scroll.

Is there a way to keep the visual of the icon and the text, whilst keeping them inside BeginScrollView?

[lament ruin]

Hey guys - I am currently unable to build & run any version of my game on the phone, as it always crashes due to gradle trying to run facebook java code from the FB SDK which I removed a few weeks ago. I always get this error:

Fatal Exception: java.lang.RuntimeException
Unable to get provider com.facebook.FacebookContentProvider: java.lang.ClassNotFoundException: Didn't find class "com.facebook.FacebookContentProvider" on path

Anyone encountered this before? I have removed the SDK from the project but it's still trying to run it in some way. So clearly I didn't catch all of it.

[sly grove]

you could always pass it in as a shader parameter from C#

[sly grove]

is there any more to this stack trace?

[lament ruin]

yeah - this is the full one:

Unable to get provider com.facebook.FacebookContentProvider: java.lang.ClassNotFoundException: Didn't find class "com.facebook.FacebookContentProvider" on path: DexPathList[[zip file "/data/app/~~uN2HgfnUxpXc6E7PfrBDLg==/com.ggs.dungeoncreeprs-yLGKbyfWrpzwEq7FRlvNIA==/base.apk", zip file "/data/app/~~uN2HgfnUxpXc6E7PfrBDLg==/com.ggs.dungeoncreeprs-yLGKbyfWrpzwEq7FRlvNIA==/split_config.arm64_v8a.apk"],nativeLibraryDirectories=[/data/app/~~uN2HgfnUxpXc6E7PfrBDLg==/com.ggs.dungeoncreeprs-yLGKbyfWrpzwEq7FRlvNIA==/lib/arm64, /data/app/~~uN2HgfnUxpXc6E7PfrBDLg==/com.ggs.dungeoncreeprs-yLGKbyfWrpzwEq7FRlvNIA==/base.apk!/lib/arm64-v8a, /data/app/~~uN2HgfnUxpXc6E7PfrBDLg==/com.ggs.dungeoncreeprs-yLGKbyfWrpzwEq7FRlvNIA==/split_config.arm64_v8a.apk!/lib/arm64-v8a, /system/lib64, /system_ext/lib64]]```

[sly grove]

there's no stack trace there unfortunately

[lament ruin]

this is all i have from firebase crashlytics

[sly grove]

Does you project have a custom android manifest file?

Perhaps it's being mentioned there

It would be called AndroidManifest.xml

[lament ruin]

Fatal Exception: java.lang.RuntimeException: Unable to get provider com.facebook.FacebookContentProvider: java.lang.ClassNotFoundException: Didn't find class "com.facebook.FacebookContentProvider" on path: DexPathList[[zip file "/data/app/~~uN2HgfnUxpXc6E7PfrBDLg==/com.ggs.dungeoncreeprs-yLGKbyfWrpzwEq7FRlvNIA==/base.apk", zip file "/data/app/~~uN2HgfnUxpXc6E7PfrBDLg==/com.ggs.dungeoncreeprs-yLGKbyfWrpzwEq7FRlvNIA==/split_config.arm64_v8a.apk"],nativeLibraryDirectories=[/data/app/~~uN2HgfnUxpXc6E7PfrBDLg==/com.ggs.dungeoncreeprs-yLGKbyfWrpzwEq7FRlvNIA==/lib/arm64, /data/app/~~uN2HgfnUxpXc6E7PfrBDLg==/com.ggs.dungeoncreeprs-yLGKbyfWrpzwEq7FRlvNIA==/base.apk!/lib/arm64-v8a, /data/app/~~uN2HgfnUxpXc6E7PfrBDLg==/com.ggs.dungeoncreeprs-yLGKbyfWrpzwEq7FRlvNIA==/split_config.arm64_v8a.apk!/lib/arm64-v8a, /system/lib64, /system_ext/lib64]]
       at android.app.ActivityThread.installProvider(ActivityThread.java:7513)
       at android.app.ActivityThread.installContentProviders(ActivityThread.java:7019)
       at android.app.ActivityThread.handleBindApplication(ActivityThread.java:6790)
       at android.app.ActivityThread.-$$Nest$mhandleBindApplication()
       at android.app.ActivityThread$H.handleMessage(ActivityThread.java:2132)
       at android.os.Handler.dispatchMessage(Handler.java:106)
       at android.os.Looper.loopOnce(Looper.java:201)
       at android.os.Looper.loop(Looper.java:288)
       at android.app.ActivityThread.main(ActivityThread.java:7918)
       at java.lang.reflect.Method.invoke(Method.java)
       at com.android.internal.os.RuntimeInit$MethodAndArgsCaller.run(RuntimeInit.java:548)
       at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:936)

[finite pond]

Yep, it seems like I was mistaken. MaterialPropertyBlock it is

[lament ruin]

i'll take a look in the files.

[lament ruin]

I do have a bunch of them and checked the packages, but all I can find are firebase related packages / unity appsflyer (i only installed the base SDK so far and didn't do anything with it yet) / and some android sdk packages. Thanks for the hint tho - i'm going to scour further for possible solution 😢

[long ivy]

if you can't find it in any manifest, look for any scripts you might have missed that might be inserting it during a build processing step (ex: https://docs.unity3d.com/ScriptReference/Android.IPostGenerateGradleAndroidProject.html)

[lament ruin]

I'm fairly certain that is the issue - since I can't find anything in my project. I can't find anything obvious while searching my code base but will try to look more.

[sly grove]

Dumb question but have you tried searching for "facebook" in your entire project?

[lament ruin]

Yes I did lol - that was the first thing I tried.

[sly grove]

you searched where/how though? In your IDE? In Unity?

Or with some other tool

If possible I'd use something like this:
https://stackoverflow.com/questions/16956810/how-to-find-all-files-containing-specific-text-string-on-linux

should work in e.g. git bash or mingw32 if on windows

[lament ruin]

i've searched with my IDE & in my Unity project. The only thing that comes up is a link from Appsflyer for deeplinking but that is not currently used for anything.

[lament ruin]

interesting - i'll take a look.

[loud wadi]

Hello, does methods in math run faster even without BurstCompiler?

[crystal panther]

anyone know how to access an hdr color's intensity value accurately in-code? I'm writing an editor piece to help with doing one thing over and over, and one of those things is setting a list of colors on an added script. The colors are static.

[untold moth]

I don't think so. Even if they do, the difference is probably negligible.

[loud wadi]

But math has SIMD optimization. Will that affects computation outside BurstCompiler?

[untold moth]

No. Single instruction multiple data is one of the optimizations that burst enables. Without it, it's not in effect.

[loud wadi]

What if I'm doing matrix multiplication?

[untold moth]

If you need to assign them via the inspector, use https://docs.unity3d.com/ScriptReference/ColorUsageAttribute.html

[untold moth]

I'm not sure. Test it. I bet that there isn't much difference.

[loud wadi]

Ok, thanks

[crystal panther]

i want to do it via code

everything is in a file. its being set through an editor window button.

[untold moth]

From what I understand, an HDR color is just a normal Color in memory. The inspector just makes the calculation for you.

It takes the base color and increases its brightness along all the 3 components.

For the exact formula you'll need to google

[crystal panther]

I've tried multiplying the color amounts by the intensity amount, and it made the color intensity nearly max.

It went from 1.4 to 8.7 by multiplying the color values

[untold moth]

Yeah, I'm not sure that's as simple as just multiplying the values by intensity.

[crystal panther]

I could use min and max exposure, although deprecated I think that affects it

I'll have to test

[loud wadi]

It's even slower hahahaha🤣

[untold moth]

Probably some overhead necessary for burst.

[knotty obsidian]

Henlo, Cloud code C# modules, cloud save data. I want to access 2 players' data via my game server, how would I do this? This is my current code

     var apiResponse = await gameApiClient.CloudSaveData.GetItemsAsync(
            context,
            context.AccessToken,
            context.ProjectId,
            playerId,
            new List<string> { PlayerInventory.Keys.MatchDeck }
        );```
 on my C# module. `gameApiClient` is `IGameApiClient` sourced by `GameApiClient.Create()`. `playerId` is a string, the other player's Id 
As it stands, the requesting user (currently anon login) does not have perms to check the other player's (also anon) data. I don't know if it is possible to make the request as an admin of some sort.
The C# module is called via Unity's `CloudCodeService.Instance`

[subtle nexus]

Odd question, does anyone know if there is a good way to allow for Bluetooth file sharing from and android build to windows and windows to android? I'm looking to send fairly small txt or csv files.

[untold moth]

You'd probably need to look up a plugin that provides that functionality or write your own(it would probably involve writing a native plugin and C# bindings)

[gusty tusk]

I enabled stack trace and notice I am getting leaks at a simple NativeArray<float3> allocation on TempJob.

TempJob leak at address 0000023A3DAAC0B0:
0x0000023c203cadeb (Mono JIT Code) Unity.Collections.NativeArray`1<Unity.Mathematics.float3>:Allocate (int,Unity.Collections.Allocator,Unity.Collections.NativeArray`1<Unity.Mathematics.float3>&)
0x0000023c203cacd3 (Mono JIT Code) Unity.Collections.NativeArray`1<Unity.Mathematics.float3>:.ctor (int,Unity.Collections.Allocator,Unity.Collections.NativeArrayOptions)
0x0000023c203c9abb (Mono JIT Code) ProjectBrave.Runtime.Combat.WalkMoveSystemManager:OnSystemUpdate (System.Collections.Generic.List`1<ProjectBrave.Runtime.Combat.WalkMove>,int) (at C:/Users/Frava Office/Documents/Unity Projects/Project Brave/Assets/_ProjectBrave/Scripts/Runtime/Combat/Core/CombatObject/AI/Movement/WalkMoveSystemManager.cs:38)

This is the simple code

    public class WalkMoveSystemManager : BaseSystemManager<WalkMove>
    {
        private NativeArray<WalkData> _moveData;
        private NativeArray<float3> _outputData;

        public const int BATCH_SIZE = 128;

        protected override void OnSystemUpdate(List<WalkMove> dataList, int dataCount)
        {
            _moveData = new NativeArray<WalkData>(dataCount, Allocator.TempJob);
            _outputData = new NativeArray<float3>(dataCount, Allocator.TempJob); // Leak here at line 38

            for (int i = 0; i < dataCount; i++)
            {
                _moveData[i] = dataList[i].GetMoveData();
            }

            ManageWalkMoveParallelJob moveJob = new()
            {
                DeltaTime = Time.deltaTime,
                MoveData = _moveData,
                OutputData = _outputData
            };

            moveJob.Schedule(dataCount, BATCH_SIZE).Complete();

            for (int i = 0; i < dataCount; i++)
            {
                dataList[i].MoveAgent(_outputData[i]);
            }
        }

        protected override void DisposeAllNativeArray()
        {
            if (_moveData.IsCreated) _moveData.Dispose();
        }
    }

Not sure why this leaks, its a very simple system

Oh wait I forgot to dispose of it

Sorry my bad

[tiny pewter]

once you allocate a native/unsafe something then immediately put its dispose() in OnDestroy or somewhere else

[steel snow]

So i have a weird behaviour i dont understand

Im combining two meshes like this:

CombineInstance[] combines = new CombineInstance[2]
{
    new CombineInstance(){mesh = m, subMeshIndex = 0, transform = Matrix4x4.identity},
    new CombineInstance(){mesh = s, subMeshIndex = 0, transform = Matrix4x4.Translate(Vector3.up) }
};

Mesh final = new();
final.CombineMeshes(combines, true, true, true);
_filter.mesh = final;

but i want each mesh to be a submesh, so i initially gave them different submesh indices and set combine submeshes flag to false, only the second combine instance didnt even end up in the final mesh.

The only way i can get both meshes to become one is same submesh index and setting combine submshes to true but then i dont have the submeshes anymore only 1 submesh ... any one know why that might be ?

[long ivy]

what do you mean "gave them different submesh indices"? Set second parameter of CombineMeshes to false, rest of code stays the same

[steel snow]

i had submeshIndex 0 and 1 respectively

and flag was false

but only the first combine instance in the mesh appeared

the only way both appear is as the code shows above

[long ivy]

in the input? If s has 1 submesh and you specified subMeshIndex 1, it's naturally not going to appear

[steel snow]

huh

am i not specifying that i want mesh s to be submesh [1] of final ?

i presume thats what it means

[long ivy]

no

[steel snow]

oh

so i need both = 0 and then flag to false?

[long ivy]

yes

[steel snow]

ok let me try it

oh

i thought i had tried this before but then i realised i hadnt added a second material

kinda wish it would give pink shader error color if missing a material rather than simply not display 😄

@long ivythanks for clearing up the confusion on submesh indices though!

all working nicely now!

[steel snow]

got a weird issue with mesh uvs.. i have a 2d polygon that i generated procedurally... im setting the uvs to the world space of the vertices on the x,z plane yet when i apply my texture its as if its scaled so much that it's just one colour

i dunno where to begin to debug this problem

this is how i do it:

_polygonMesh = triangulation.Triangulate2D(p);
var uvs = new Vector2[_polygonMesh.vertexCount];
for (int i = 0; i < uvs.Length; i++)
{
    var v = _polygonMesh.vertices[i];
    uvs[i] = new Vector2(v.x, v.z);
}
_polygonMesh.uv = uvs;

this is how it turned out

offsets seem to offset the texture but incorrectly and tiling doesnt even do anything - this is using the default urp lit material

[long ivy]

what are the texture settings on your texture? Maybe it's being clamped

[long ivy]

create a Plane GO, put it at origin, apply this material, and run your uv code excluding the triangulation

also you might want to copy the vertices array, your loop is extremely inefficient currently

[steel snow]

yeh its inefficient it was a quick job to figure out why the uvs wont work

let me try what you suggested

[steel snow]

worked for the default plane 😭

what does that mean though for my generated mesh

surely its not the vertex values

[long ivy]

well it definitely looks like something is wrong with your uvs, so now I'd use that plane as a reference for what the uvs should look like and generate your triangulated mesh in the same place. Does your mesh have any scaling?

[steel snow]

no scaling

but the uvs in my generated mesh is using the same code as that plane mesh

using x;z of the vertex values

what other uv setup is there to do

[long ivy]

debug the actual values. Make a simplified triangulated mesh and see what the uvs end up as

btw you are not using world space for the uvs coords, they're in object space which is why I asked about scale

[steel snow]

oh my bad as for scale the transform is 1,1,1 and the material tiling is 1,1

not sure what other scaling there is

ive not applied a multiplier to the uv values

wtf

it just suddenly decided to work

🤔

[mortal ocean]

Hi everyone,

Due to the structure of my game, I need to use a radius value smaller than 0.05f, which is the minimum agent radius value. But as you know, Unity currently limits this value to a minimum of 0.05f. Therefore, I cannot set a smaller value in any way.

Is it possible to remove this limitation? I don't want to increase the size scale of the entire game just because of this problem.

Thanks.

[steel snow]

the radius is set to 0.05 not 0.5 in that image

[mortal ocean]

Sorry, I wrote it wrong. Actually it will be 0.05f.

[steel snow]

Any smaller and the navmesh agent steering will bug out

[mortal ocean]

Are you sure about this? In previous "Unity AI" versions, Agent Radius could be reduced to 0.01f.

[steel snow]

I'm not aware of that

Maybe the inspector is stopping you but in code you can still use a smaller value

[mortal ocean]

I tried this, but even when I change the radius value through the code, the radius value updates itself to the minimum value of 0.05f while creating the NavMesh Surface.

The problem is probably not on the inspector.

[sterile meteor]

Do you know of any articles, blogs, courses on advanced code architecture for large Unity projects? I understand that every project is different and may require a different architecture, but there are general concepts, such as for eg. Clean Architecture.

[dusty wigeon]

You can read on architecture, it does not need to be related to Unity. Whenever you deal with advance architecture, it might help to have a basic understanding of the technology you work with, but the most important is to understand key concept that transcend the technology in use.

By example,

• SOLID + T(Tell don't Ask) Principle
• Cohesion/Coupling
• Design Pattern (Statemachine, Pub/Sub, etc.)
• Inheritance vs Composition
• Screaming Architecture

[dusty wigeon]

Usually, you do not want to get that small. My advice would be to change everything so that your unit is not meter but centimeter. (Multiply the size of everything by a factor) In other words, your 0.05 would be 5. Remember, Unity Unit is whatever you want it to be.

[mortal ocean]

It seems I have no other choice. Thanks!

[knotty obsidian]

My question was not resolved yesterday night, so I'm asking for this again, to see if today morning people can answer 😦

[astral kernel]

i want an object to follow exact player movement with some offset but i have no idea how to do that, help pls

[wraith storm]

Could you make the object a child of the player object, then it will always maintain the same offset.

[astral kernel]

no, like multidimensional baby in The binding of isaac

Player moves and after for example 2 seconds the object follow players path

[wraith storm]

In FixedUpdate() you could store the Players position in a Queue()

So somethign like...

Queue<Vector3> historyOfPositions = new Queue<Vector3>();
int historySize = 60 * 10; // 10 seconds if FixedUpdate is 60fps

void FixedUpdate() {
historyOfPositions.Dequeue();
historyOfPositions.Enqueue( thePlayer.transform.position );
}

public Vector3 GetDelayedPlayerPosition() {
return historyOfPositions.Peek();
}

Oh, sorry in FixedUpdate() it should be...

if( historyOfPositions.Count > historySize ) {
historyOfPositions.Dequeue();
}

[livid goblet]

Hey I'm getting this really weird behaviour where removing this debug line from my code causes the code to run differently

bool DoorLiesWithinIntervalX(float start, float end, TestDoor door)
{
float doorPosition = door.transform.position.x;
float doorWidth = door.transform.localScale.x;
float doorStart = doorPosition - doorWidth / 2;
float doorEnd = doorPosition + doorWidth / 2;

    bool entirelyLeft = doorEnd <= start;
    bool entirelyRight = doorStart >= end;

    Debug.Log("door number " + doorNumber + " doorStart: " + doorStart + " doorEnd: " + doorEnd + " start: " + start + " end: " + end + " entirelyLeft: " + entirelyLeft + " entirelyRight: " + entirelyRight);


    return !(entirelyLeft || entirelyRight);
}

Here is the code

[sly grove]

!code

[thorn flint]

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.

[livid goblet]

bool DoorLiesWithinIntervalX(float start, float end, TestDoor door)
    {
        float doorPosition = door.transform.position.x;
        float doorWidth = door.transform.localScale.x;
        float doorStart = doorPosition - doorWidth / 2;
        float doorEnd = doorPosition + doorWidth / 2;

        bool entirelyLeft = doorEnd <= start;
        bool entirelyRight = doorStart >= end;

        Debug.Log("door number " + doorNumber + " doorStart: " + doorStart + " doorEnd: " + doorEnd + " start: " + start + " end: " + end + " entirelyLeft: " + entirelyLeft + " entirelyRight: " + entirelyRight);


        return !(entirelyLeft || entirelyRight);
    }

Hey I'm getting this really weird behaviour where removing this debug line from my code causes the code to run differently

[sly grove]

Differently how?

Nothing in that statement would change anything as written

[livid goblet]

I am aware, which is why it's so strange

[sly grove]

(would recommend using string interpolation to make that cleaner though)

[livid goblet]

it is part of a larger algo, the algo breaks and creates a different outcome when I remove the debug

[sly grove]

that's very vague

I assume you're misinterpreting the situation somehow

or perhaps you are doing something multithreaded and you have a race condition

[livid goblet]

how am I misinterpreting, when I remove the debug, the outcome is different

[sly grove]

and adding the log slows it down enough to change the behavior

[livid goblet]

nothing is multithreaded

[sly grove]

you'd have to give more details

you're giving very little to work with

the only thing that could conceivably change is how long this code takes to run

so maybe you have something very time or framerate sensitive going on

[livid goblet]

I thought the same as well, but I'm sure it's not the case

[sly grove]

another possibility is that doorNumber < some of these things are properties and the getter for the property has side effects

[livid goblet]

I removed door number and several other parts of the debug and it worked fine

[sly grove]

are any of those things properties?

[livid goblet]

only upon removing " doorEnd: " did it break

[sly grove]

how are you so sure?

and again, what is the nature of the breaking

[livid goblet]

it causes the function to return false in a case where it is true

nothing is moving, none of the properties are changing

[sly grove]

I think it's more likely you are misinterpreting something about this situation than it is that the log statement is actually changing anything.

Perhaps this function is running multiple times for example and you are mixing up multiple iterations somehow

or maybe something about how you're checking the result is off

you can write a simple test case here where you write another version of the function with and without the log statement

pass the same data to both

and see if they return the same thing

[livid goblet]

I'll give that a try

even running on the same data, they are returning different results

I'm just gonna restart my pc

there are demons in the machine

[sly grove]

show the two functions?

[sly grove]

are any of these properties?
currentPointX nextInterruptionX door?

[livid goblet]

no

[sly grove]

I am noticing doorEnd and start are... either the same or very close to one another?

so this could be some kind of floating point precision issue

[livid goblet]

that was my thought as well

but how does calling debug change it that?

or affect the results

[sly grove]

not sure...

[livid goblet]

spooky

[sly grove]

especially since the debug.log is after the bools are already computed

[sly grove]

What happens if you just call the same function twie

[livid goblet]

the one with the debug or without?

[sly grove]

Debug.Log("First");
Debug.Log(DoorLiesWithinIntervalX(...));
Debug.Log("Second");
Debug.Log(DoorLiesWithinIntervalX(...));```

I guess either one

but let's say without

[livid goblet]

both versions call the same with and without

meaning, it doesn't change anything

bool DoorLiesWithinIntervalX(float start, float end, TestDoor door)
    {
        float epsilon = 1e-6f;

        float _doorPosition = door.transform.position.x;
        float _doorWidth = door.transform.localScale.x;
        float _doorStart = _doorPosition - _doorWidth / 2;
        float _doorEnd = _doorPosition + _doorWidth / 2;

        bool entirelyLeft = _doorEnd <= start + epsilon;
        bool entirelyRight = _doorStart >= end - epsilon;

        return !(entirelyLeft || entirelyRight);
    }

this fixes it smh

[sly grove]

still weird though

i'm kinda stumped tbh

the only thing I can think of is that somehow the Transform position or scale getter has side effects?

[livid goblet]

I'm not sure either

definitely weirdest bug I've ever had

thanks for helping out!

[astral kernel]

it doesn't work

but it doesn't give any errors

[ancient sequoia]

Alright, Im getting some weird behavior in Unity 2023.1. My coroutine is not waiting for my async to finish. The code roughly looks like this:

StartCoroutine(function)

public IEnumerator function {
yield return AsyncFunction();

*do stuff that requires AsyncFunction to finish first (I need value2)*
}

async Task AsyncFunction() {
value = await AnotherAsync();
value2 = await AnotherAsync2(value);
*store value2 to memory*
}```

[sly grove]

coroutines do not work interchangeably with async

unless they introduced that recently

you can't yield return AsyncFunction()

that will just wait one frame

like yield return null;

[ancient sequoia]

Oh, it will only wait one frame? I thought it waited like with
yield return new WaitForSeconds(2);

[sly grove]

you would need to do something like:

Task t = AsyncFunction();
yield return new WaitUntil(() => t.IsCompleted);```

[sly grove]

WaitForSeconds is special

it only knows how to deal with special things like WaitForSeconds

It has no idea what a Task is

[ancient sequoia]

Wooooow, thank you for your help Praetor! I will try out your code snippet! Sounds like I need to convert the whole function to async rather than snippets.

Yaaaay it works! You are amazing Praeor!

[fossil girder]

I need some help with rendering issue.
When I drag the game to enlarge or reduce the windows size. The graphics does not get resize properly.
This only happens with xiaoxiao921’s DearImGuiInjection.
I assume the issue is with DX11, but I am not sure how I can fix it.

[subtle nexus]

Ok, thanks.

[dusky kettle]

Hey guys, I'm trying to get the water in HDRP to simulate in game time instead of unscaled time, and I think I've found the part of the code that I would need to modify to do that, but the problem is that this class is an internal class. Since HDRP is a package I don't see a way to modify it without outright copying the entire thing and changing the names, does anybody know if there would be a more elegant way to override this update function without ruining all future updates and compatibility?

this update function is in the internal class WaterSimulationResources and I found it in at Library\PackageCache\com.unity.render-pipelines.high-definition@14.0.8\Runtime\Water\HDRenderPipeline.WaterSystem.Simulation.cs

im on 2022.3.10f1

im struggling to find where this update function is called at the moment, and WaterSimulationResources doesn't derive from monobehaviour or anything for that matter

a WaterSimulationResources object is created when the WaterSurface method CheckResources is called

dont see any direct calls to Update though and i didnt see any calls to Update within WaterSimulationResources either

man so close yet so far, i cant really see a way to get around this

itd be a lot easier if there was a solution that i could use to search through

OH

I FOUND WHAT CALLS IT

unfortunately this doesnt look like the kinda thing i should be overriding

its the UpdateWaterSurface function in Library\PackageCache\com.unity.render-pipelines.high-definition@14.0.8\Runtime\Water\HDRenderPipeline.WaterSystem.cs

[austere jewel]

Use an assembly reference so you can add your own class to the assembly and do whatever you want

[dusky kettle]

🤔 how I do that

do you have some resource online you might be able to point me in the direction of

[austere jewel]

Unity assembly reference

[dusky kettle]

lmfao aight

[austere jewel]

https://docs.unity3d.com/Manual/class-AssemblyDefinitionReferenceImporter.html

[dusky kettle]

oooh okay

right hmm

let me try that

im trying to figure out which assembly the water stuff is in

if it even is just one

cos theres both runtime and editor stuff here but it looked like the code was accounting for both editor and runtime mode

oh theres like 5 more editor scripts but im assuming and hoping that theyre not relevant

well that's one problem solved

but another created

[dusky kettle]

i suppose i can just try doing a janky override here

idk if itll work

[austere jewel]

why are you making it public? The class must be the same or less accessibility than what it inherits from

[dusky kettle]

good point

i guess my problem is more i dont know if this override is gonna be very useful cos the override class would need to then be used

maybe i can override the object creation part in WaterSurface and take advantage of polymorphism stuff

its accessibility does make that an attractive prospect

id need to copy all my parameters over but thats fine

thats also a problem

okay maybe this is just the wrong approach

i dont know any other approach i could take tho

oh the new keyword

oh okay that sounds like what I need

seems like i didnt manage to fix much

i think its just not using the new fuinctions i made

since its not an abstract class

when it gets upcasted it just uses the base function instead of the new ones in the children classes

alright well theres the new problem

i need to figure out how to make a non-abstract class abstract

kinda

oh i can just use the exact same name???

but then it loses access to everything the base one does for some reason

[dusky kettle]

im looking real hard for anything i can hijack and am coming up empty

i realised that simulation.simulationTime and simulation.deltaTime were fully exposed and could be written to from literally anything

so i thought it might be possible to just to calculate the values in my own scripts instead

but unfortunately the simulation Update function is called right before it renders the water surface every time so im not gonna be able to get inbetween that without some more drastic changes to the entire system

this would be such an easy change if I just had access to the system itself

im not in a rush here it might just be better to make a feature request and hope they can make that change

that being said i think the shortest amount of time I could need would be a week, or i could be drastically miscalculating and could need months

idk where to even make feature requests of tiny scale like this

[full zealot]

i need some help

my car is driving backwards and i want to use Math.Abs (Vector3.forward * speed * Time.deltaTime); But the Vector3 keeps throwing errors with doing that

[half swan]

you would use abs on an element. Not the vector itself. It works on ints/floats/doubles

Also, try transform.forward instead of Vector3.forward?

[full zealot]

thank you it worked

if (Input.GetKeyDown(KeyCode.C)) {
cam1.enabled = !cam1.enabled;
cam2.enabled = !cam2.enabled;

what does this code do

mainly the !cam1.enabled;

[half swan]

It toggles the enabled state.
.enabled is a bool. So if it returns true (it is enabled), it SETS it to the opposite (disabled)

so it is swapping cameras

[full zealot]

ohhh ok thank you so much

you helped me alot

[sly grove]

! Is just the negation operator for bools. Turns false true and vice versa

[vestal condor]

anyone know why the text is getting a bit warped? using textmeshpro

you can see how the T curves slightly upwards or the top of the H isnt exactly straight

[light falcon]

how to fix references to a class if the class and file have been renamed? I recall there being an attribute that could do this

[regal lava]

https://docs.unity3d.com/ScriptReference/Serialization.FormerlySerializedAsAttribute.html

That?

[light falcon]

yes! but it seems to only work on properties