Official "Meet the Devs" -- Post Your Questions

Can one of the devs please speak to the problems with retrieving the current touch position on mobile? There appear to be multiple open questions about it and I have been attempting to resolve the issue as well. See this post.

 

Many people want to create local multiplayer we now the way using the Input Manager and Player Input but many like myself want to create it using the C# Generated class how can we achieve this?

 

Will you be incorporating touch gesture support any time soon? Particularly highly customizable touch gestures.

Examples: Swipe, drag, turn, flick

 

How do you go about making a rebind controls menu? Including saving and loading the rebinds between play sessions.

 

Will XR and New Input System ever reconcile their differences?

 

Hi @Rene-Damm I know you have been working on pressure and tilt support for pen tablet, like a Wacom.
I'm mainly interested in the use for editor tools, not so much for runtime.

Questions:

• Is Wacom pressure and tilt now working properly in the editor?
• Do you have any new examples planned to show hot to use the Wacom pressure in the editor tools.

https://docs.unity3d.com/Packages/com.unity.inputsystem@0.9/manual/Pen.html

Last time I checked it dit not work for me in the editor tools.
https://forum.unity.com/threads/input-system-for-pen-stylus-use-with-editor-tools.679960/

 

Sorry if this is too angry, but why does it take 5 years to make a new input system? Entire game engines can be built in that time frame. Why is does it take Unity so long to make packages that are functionally incomplete compared to asset store alternatives that have significantly fewer resources?

 

My questions :

- is it possible to assign multiple local players to a single device, ex : two players sharing a keyboard or a gamepad (yes, I want to do that)

- still on local multiplayer, how to properly handle UI interactions with one of the players having the priority over the others ? ex : player 1 can navigate in all menus and options, while others are restricted to certains menus parts.

 

why version 1.0 is called a preview ?

 

Will be glad to hear about TOUCH events on the Android platform.
Especially ONE tap filter - to skip first tap, if we are trying to get DOUBLE tap. I cannot understand, how to filter it with the new InputSystem.Also cannot catch Long Tap (Press) on the screen.
Thank you!

 

Im trying to customise the system a bit and add new interactions. Im very lost about what a context, action, phase, `performedAndStayX` are. Some more documentation on these concepts would be very much appreciated.

 

Upgrading an 4.7 project onward I faced, in a short duration, many generations of breaking changes and change in programming patterns.

Is anything done to slow down this trend? If so, what?

 

Another one :

- a simple demo on how to create a Custom Device in order to use it as a Bot would be nice

 

This ^^^ guy just gave me a genius idea:

A group of Unity devs make a game in a few days while non stop on Twitch.

Game idea is locked but each comment is answered.

One account, Unity devs in the background, and all questions fielded by one experienced person who is low key, I'm thinking someone in the "quality of life" team, preferably who has been with the company long enough that he doesn't fear repercussions

benefits:
the people can pick your brains
you can see pain points
It's fun

 

Will you create 3d and 2d physics that will run on gpu?

 

Please talk about complete touch controls and possibilities with touch inputs. And talk about optimising 3dgames for mobile platforms without having to sacrifice the quality of the game. The utmost important thing is please tell us how to use the profiler for optimising mobile games

 

Hi,

We are experiencing a strange issue with the new input system on macOS. Whenever we take a screenshot using cmd + shift + 4 when switching action maps, the new input system seems to soft crash (no errors in log or indication that anything is broken) until we switch the current action map again. Stangely, left shift will still work.

To give some further details, things will work normally again once the game screen is unfocused in editor (e.g. when one clicks the console). We've experimented with filtering what inputs are accepted in a single action map and having seprate action maps for UI and Gameplay. The only thing that seems to help is switching to a dummy action map and before switching back to "fix" the soft lock.

Any ideas into how to prevent this?

 

What's the best way to reduce draw calls?

 

What's the state of compatibility with DOTS?

 

- When can we expect it to be a Verified package in a LTS version?

- How can we detect in code that the new InputSystem API is present, so we could ship cross-compatible components and Asset Store products without the end user facing build errors?

 

Will we see it in the YouTube channel? I was not be able to watch it on Zoom.

 

Same here ! I have missed the schedule

 

Same here. I registered but could not attend. Any chance it was recorded?

 

I think its good news, Is there any place on the forum I should look out for when this is implemented MGUI and UI Elements? I know that on Mac OS the pressure already works for some time

https://docs.unity3d.com/ScriptReference/Event-pressure.html

Good to hear is coming to windows too, so more tools will be able to make use of it by default

 

Thanks for all these answers

 

I was wondering if you had any examples or tutorials of modifying the Player Input Manager? I'm try to better understand how that works as I'm building a split screen shooter and need more control over what happens when the player input manager spawns a player when a button is pressed to join.

 

A way to bind dpad and wasd separated? It seems that the Input System associate "One Device --> One Player" handling, but i don't think should be like this. You can go in example with one controller 2 player and so on... or one big custom input device that handling two player separeted, but it's a single device.

 

@Rene-Damm Thank you so much for your reply! Indeed, it was clarifying at several levels. I'll begin adopting the the new Input System straight away.

 

I checked out several of the Unite Now offerings this week, and it seems like y'all are experimenting with various formats, which is cool.

For me, the format of this session was FAR AND AWAY the BEST of the lot. Please do more like this in the future!

• Live demo
• Multiple Live speakers
• Moderated live Q&A
• Recorded to YouTube for those who missed it or want to re-watch it
• A forum for follow up

Wonderful!

 

For mobile games, how should we think about using InputManager vs. using IDragHandler and related input interfaces in scripts?

Will/Does InputManager replace all those interfaces? Do we use some combination?

Would be nice to see an example of how to use InputManager with a basic mobile game that uses all the common mobile touch inputs.

 

hey guys. I'm currently working with the actionasset approach but I couldn't find a way to disable an input scheme (mouse/keyboard) or change a default scheme and prevent switching, esp for testing this is pretty much annoying for me...

AND:it would be nice if we could change some input settings without having to stop the player, change some settings, wait for recompile and then restart the player. that's damn ridiculous.

So I'd say instead of BAKING data into the Input class, how about you put it into a json file, where it belongs? Obviously if you have to generate some code based on that data, do so, but much of that baked data is just a method call, like normalizing a vector or adding a deadzone. or many other actions I haven't found yet. you don't need to regenerate a class for that, and slowdown the whole workflow. (I'm not sure but you can probably externalize this whole process into another module and just add it to the whole (Player, maybe Editor) assembly once it is finished regenerating, this way you don't have to recompile THE WHOLE GAME, so In short I think the Input Actions Window shouldn't be much more than I json editor, that calls an external regenerator, whose generated Module gets added to the executing assembly)

If you need some help, I'm happy to ;-)

 

Thank you, but still can't catch Press/Long tap of Android touch screen.
This code is not working for me.
P.S. How can I see Debug output in Editor, while I run the application on the device? Is the only way to see log from Android - using Android Studio?
I don't have any logs this way.

 

Action: "ActionLongTap"
Properties:
Action -> Action type: Pass Through
Action -> Control Type: Touch

Bindings:
Primary Touch [Touchscreen]
Binding: Path = <Touchscreen>/primaryTouch
Interactions:
Hold/ Slow Tap/Press - nothing catches. This way works only Tap/Double Tap.

Code (CSharp):

1.   public void LongTapLogEvent(InputAction.CallbackContext context)
2.   {
3.     Output("...LongTapLogEvent\n"); // This is not working!!!
4.     if (context.performed)
5.     {
6.       var control = context.control;
7.       Debug.Log("Slow Tapped ****");
8.  
9.       if (control.parent is TouchControl touch)
10.       {
11.         Debug.Log($"Slow-tapped at {touch.position.ReadValue()}");
12.       }
13.     }
14.     else
15.     {
16.       Debug.Log("Long tap::Somewhere here");
17.     }
18.   }
19.  

 

Thank you! Will wait for the announce.
I hope it will be soon.

 

@Rene-Damm I was looking at the samples that can be downloaded from the package manager and the Editor window sample seems to be missing. It's there in the video at 11:26, but It's not there in the package manager now. Do you have a direct link to it?

And please, please, tell me this isn't true. I have been waiting forever to be able to read input from a gamepad in editor mode for years, and now that we finally have it you guys are thinking about cutting it?

 

Finally a demo where it is actually discussed that the default "SendMessage" method is abysmal at best!

 

The problem with that is that the input part of the Event system that we usually have to rely on for reading input in the editor is an outdated, ill-conceived piece of trash. I mean, the whole Event thing is bad, but especially reading input. We can't actually do useful things there.

So no matter how poorly the new input system integrates into the editor, it would be better than what we have.

 

Hello,

I'm making a top down shooter style game. Using the new input system is great and make it easy to support both keybord and gamepad. Great job !

My question is how would you go about making a directional input relative to player ?

See, to aim I use the mouse to get a world position by casting aray from the camera to the world so I get the mouse position just fine for the keyboard setup.
But for the game pad, I use the rigth stick which just give a -1->1 vector2. For now I made acustom processor that project that input to hte border of the screen using Screen.width and heigth. That works fine in single player scenario but for local multiplayer , the screen is not the player's camera view so the input get a bit distorded.

Is there a way to access the Camera define in the player's input component from a custom processor, so that I can base the input projection to the player's actual "view" screen ?

Custom processor code :

Code (CSharp):

1. #if UNITY_EDITOR
2. [InitializeOnLoad]
3. #endif
4. public class ToPointerPosition : InputProcessor<Vector2>
5. {
6.  
7. #if UNITY_EDITOR
8.     static ToPointerPosition()
9.     {
10.         Initialize();
11.     }
12. #endif
13.  
14.     [RuntimeInitializeOnLoadMethod]
15.     static void Initialize()
16.     {
17.         InputSystem.RegisterProcessor<ToPointerPosition>();
18.     }
19.     public override Vector2 Process(Vector2 value, InputControl control)
20.     {
21.         value.x = (value.x + 1) * (Screen.width / 2);
22.         value.y = (value.y + 1) * (Screen.height / 2);      
23.         return value;
24.     }
25. }

I read some stuff about virtual mouse, can that help in this context ?

 

This is the mouse and gamepad stick input distortion. A mouse is a way faster and sensible of a game stick. So you need to normalize the mouse input when it's magnitude is > 1. (Because a vector2 of a controller is capped here, but a mouse delta doesn't)

Code (CSharp):

1. if(rotation.magnitude > 1){
2.       rotation.Normalize();
3. }

In this way you pick each decimal value between 0 and 1, and also set a maximum magnitude of the input vector2. (In case of a mouse it will capped max at 1)

If u want to set a sensibility global for each mouse and gamepad, you can multiply this vector after the check with a variable.

Also for your problem you really need to rotate the player for facing the forward vector direction.
You should not see as "My pointer is here, so look here" but "ok, i move the mouse in this direction, so look in this direction".

The input Vector2 say (in x and y), which direction you should put in the x and z of a direction vector. This vector could be retrieved by the Mouse delta and/or stick of a gamepad.

If u need to keep the camera around the player, simply normalize this looking direction, and multiply the normalization with a "Range variable" for looking around.

 

Maybe I was not precise enougth in my description. I'm using a features that if I'm not mistaken was not presented in the talk, the split screen with hte keep aspect ratio. So when I have 2 player in a local multiplayer game, the each have 1/4 of the full screen.

Each player control a tank so they can move in a directon (no issue there has all is in normalized vector direction ZQSD (yes I'm french...) and left stick) and the tank can aim in another direction.
For the mouse user it's much more intuitive to have the tank aim where the mouse actually is on the screen, so I take the screen coordinate and cast against he world from the player's camera. In that case, there is no issue because teh player can put the mouse anywhere he want on the screen including the 1/4 of the screen that's his.

On the ohter hand for a gamepad user, it's much more intuitive to have the tank aim in the same direction as the player old the right stick. The issue is that the rigth stick only gives a -1 to 1 vector 2 range. I could make it realtive to the player easily but it would mean have a different logic for the mouse user and the gmaepad user. So I need to convert the gamepad input into a pointer position on the screen to be able to apply the same raycast logic.
The issue become that I don't have access to the player's view port so the input gets mapped out of the 1/4 of the player's screen.

I did not try to make the logic the other way around (convert mouse position to -1 1 vector) but I suspect I'll end up with the same issue and the tank won't aim at the mouse position relative to player but relative to center of the screeen.

I see no way around having a reference to the camera in hte processors to be able to make inputs relative to a player's view port.

If you want to try it out, you can look at my full project on github :
https://github.com/WAYNGROUP/MGM
If you press play you can join by pressing any button on the keybord or gamepad. you'll see the issue with the gamepad used, if he joins first, when there is two player, the aiming is much more sensitive on the north west corner, and if joins second ,it more sensitive on hte north est corner.

A word of caution though, it uses the DOTS stack and my own packages (all open source and free) so lots of experimental work .