Upgrading to Unity 5.2 - UI Problem with Raycast Target

Hi,

Maybe I'm misunderstanding something simple here, but I just upgraded to Unity 5.2 this morning and I'm now having several issues with many of my UI buttons.

After troubleshooting for about 3 hours I discovered that some of my buttons fail because the touch seems to interact with objects beneath the layer my button is on. This was previously never a problem. In the screenshot below, there are areas of the Yes and No buttons that don't respond because it seems the click / touch interacts with objects beneath the popup menu.

In this example, I have a Confirm screen with a semi transparent disabler background that previously prevented interaction with the objects below. Now with 5.2 it seems that in some odd cases a touch can interact with objects below the disabler.. weird.



I'm going to continue experimenting to provide more clarity here, but if anyone knows something simple that changed in 5.2 that would cause this issue please let me know.

 

Some additional information about this situation:

In the screen shot above, I'm using two canvases:

Canvas Foreground:
Contains the blue popup menu and semi-transparent black disabler screen

Canvas Background:
Contains the shop options that you see in the background.

It seems that the "Yes" / "No" button problem can be fixed by unchecking the "Raycast Target" on the Image component in the gameobjects in the background. That of course is not a viable solution since those objects need to remain interactable.
I must be missing something simple that has changed in 5.2. Will keep digging.

 

I'm using canvas render mode "Screen Space - Camera." It looks like the issue goes away if I change to render mode "Screen Space - Overlay"

Does this make sense to anyone? It looks like I would need to make substantial changes to support Overlay.

 

Hi Titan having the same issue. Finally got other scripts upgraded to 5.2 and now all my button clicks don't work. I don't really have the option to use Overlay as I have particles that are part of my UI. Any more information on this at all? I will keep digging around to see what I will find but need this sorted. Seems like a huge oversight if this is a bug.

 

Ok good to hear it's not just me. Here's another example where buttons don't work:


Clearly the tabs at the top of the interface are in front. In this example, the black gradient (setup as an Image component) on the rear layer had the "Raycast Target" checkbox checked by default. If checked, none of the buttons in the foreground work.

 

Maybe not related, but after upgrading my project to 5.2 mouse stopped interacting with buttons. Fixed this by removing checkbox "Force Module Active" in EventManager's Touch Event Module (script) gameobject.

 

Yea I looked into that, but mine is already unchecked. Couldn't find great documentation on that either btw.

 

My setup is having multiple canvas'
one for common UI (including back, home and help buttons)
another for standard views
another for popups

It seems all touches are hitting the popup before hitting the common UI even though the sorting layer, and hierarchy order means the touches / mouse input should hit the common > Popup > standard.

Been bashing my head against this for a while and for now means I will roll back to 5.1 but there are some horrible bugs that 5.2 fixed.
Will have to look into this more tomorrow when I have brain power.
Will post on here if I find a fix or will create a test project and submit a bug report.

 

I have the same issues with buttons and also DropComponent stop working for me (it looks like another panel must block the one with DropComponent), I have to dig in later. For now I rolled back to 5.1.3.

 

Maybe it's this same issue? http://forum.unity3d.com/threads/solved-update-to-5-2-ops-problem-now.353548/

He solved by "adding a fresh EventSystem".

 

I don't think that is related, I did try recreating the EventSystem component just to see but no change... still having issues. If I can't figure it out in the next few hours I will be rolling back to 5.1.3

 

Still no luck here, I've reverted back to 5.1.3p2, will wait for patch or new details.

 

I'll get you a project shortly... need to reinstall 5.2 to create the issue =)

 

i already report it. i guess you guys miss include this one from previous version.
check eventsystem's raycastcomparer. when i check the bitbucket unlike web source, zip version has a bug which was existed before 5.1.

 

Ok here is a reproduction project example:

If you click the Foreground button in the area that overlaps the Background button, you'll notice that foreground button doesn't work. It seems related to how many panels nested the button is within the hierarchy. You'll notice this difference in my two projects.


In the broken example, I've nested the button multiple panels deep:

Download Broken Example:
https://drive.google.com/file/d/0BwL7QqsvL2_EcEhKMUVIZDZyVFU/view?usp=sharing



In the working example, I've nested the button only a single panel deep:


Download Working Example:
https://drive.google.com/file/d/0BwL7QqsvL2_EVlBHTG8tbkQxSFk/view?usp=sharing

 

Same problem here. That's frustrating.

 

It's Case 726055 currently, and it is happened before 5.1 at case 707026.

 

I managed to solve my problem by following @Shii advice (touch module). I had another eventsystem being created at runtime, which is weird (we use several scenes merged together, it may come from that).
Seems related to touch control in some ways.

 

Hi, same problem here with UNITY 5.2

Deleting and adding EventSystem object in hierarchy mysteriously fixed that problem.

 

I tried deleting and readding the EventSystem object with no success.

 

I submitted a bug report (and suggested fix) last week. I'll post a somewhat briefer version here.

All 2D-based raycasting done by Physics2DRaycaster and GraphicRaycaster is borked because they use Raycast/RaycastAll (designed for coplanar and "flattened" coplanar intersection -- like laying string on a piece of paper with a box drawn on it) instead of GetRayIntersection and GetRayIntersectionAll (designed for non-coplanar intersection, like throwing a dart at the paper). This causes a number of issues related to mouse pointer collision when used with in-world objects or world-space canvases, namely:

1) Collision depth is screwy (the issue the OP seems to be experiencing). Since raycast blocking by 2D objects is done by the faulty raycasting, you get situations where rays may either be arbitrarily blocked or permitted to pierce.

2) Objects behind the camera can still trigger collisions. Again, seemingly because the raycast method being used is strictly 2D (and thus ignoring depth), whereas GetRayIntersection is correct for intersecting a ray and a 2D object in proper 3D space.

3) Perspective cameras are borked, typically in favor of overdecting collisions for objects directly in front of the camera. In a perspective camera, a ray should be cast from the focus of the camera toward the pointer's world-space position on the near clipping plane, which is exactly what GetRayIntersection would do and exactly what Raycast does NOT.

The fix is to change Physics2DRaycaster and GraphicRaycaster to use the appropriate GetRayIntersection* method (I believe the 2D raycaster returns an array of hits, and Graphic only returns one, but I'll leave that to the engineers). I've been able to work around it by creating subclasses with those very changes, but they should definitely be codified. Efforts to instead update my local copy of the Unity UI dlls (obtained and edited via the Unity UI Bitbucket repo) have been stymied by the fact that UnityEngine.dll's definition for UnityEngine.Input.getMouseScrollDelta seems to call an undefined getter method or possibly needs to be decorated as an InternalCall? A bit fuzzier on that.

EDIT: It's worth noting that I found/worked around these issues while on 5.1.1f, I think? If you're ONLY getting the error in 5.2 then that's certainly strange to me.

 

this is what happened to me and it works.

 

I had the same issue.but It's working now. I just had to add a fresh EventSystem.

 

A lot of issues seem to popup in 5.2 - I can't get a visual caret on any of my input fields, dropdown menus work fine the first click, but after that don't drop down again.

I've added fresh event systems to no avail.

Is rolling back to 5.1 the best solution?

 

I rolled back to 5.1.3 for now will have to wait for 5.2.1 to be released which will hopefully solve some of the isues.

 

OnClick event seems to get fire only when you release the mouse on the image and the PointerUp event only get fire when you hold down your mouse on the image and slide out of the image and release the mouse). Plus, calling the gameObject.setActive(false) is firing up PointerUp event even if the mouse isn't on it. Need to do more test to be sure.

 

I have this same problem.

Also when the project is running in the editor. I can't drag objects around in the hierarchy to rearrange them. I wonder if it's related to this somehow?

 

Has this been released yet?

 

just tried 5.2.0p1 this is still broken and there was no mention in release notes.
half of the gui is not working. clicks go thru buttons. completely hopeless. whole canvases unclickable randomly.
in editor and in builds. had to roll back to 5.1.3p3

 

Yes, confirmed with the test project, this issue has not been fixed.

 

Ya p1 did not fix my companies project either....

 

5.2.1 has been released today though.
http://unity3d.com/get-unity/download?ref=personal

 

Just installed 5.2.1f1, tried the test project again, and the issue remains.

 

Ya its still not fixed...

 

The same here, moreover the performance of UI drastically decreased. Can anyone paste a link to issue tracker to bump up the votes?

 

I have same problem, changing EventSystem did not help.

 

I couldn't find 726055, but here is a link to the other:
http://issuetracker.unity3d.com/iss...aycaster-regardless-of-sorting-layer-settings

 

I have the same problem with unity 5.2.1f1... changing the EventSystem did not help...

 

Omg I was about to throw my computer by the window.... I'm happy to see it's a bug haha. I lost a whole day trying to fix this, that's the frustrating part.

 

this is starting to be completely hopless developing with Unity. currently ALL official builds of Unity are broken.
4 5.1 and 5.2

In Unity 4 the binaries crash on iOS. ( proof: http://forum.unity3d.com/threads/crash-upgrading-to-4-6-8p2-on-application-loadlevel.354899/ )

In 5.1 sprite packer is broken. ( proof: http://forum.unity3d.com/threads/cannot-load-atlas-during-build-enter-game-missing-file.340440/ )

In 5.2 the GUI is completely broken and it leaks memory. Your game will eventually crash in just by showin a text.
( proof: this forum thread )

Absolutely frustrating in searching the working Unity version.
My projects all have this problems. Can't do anything.

 

We are using 5.1.0f3 as our stable version at the moment.

The Gui memory leak is fixed in 5.2.1p1 , but there are a million and one other issues with it

 

Ohhhhhh snap... it works on the test project! I'll switch our main project over and test this out shortly.

 

Well, I migrated our game project over to 5.2.1p1 and we're seeing a ton of fun new problems. No idea what is wrong... so while the button click issue as demonstrated in the test project seems to be working, we're now having loads of rendering problems on the screen that I can't explain. Here are a few screen shots of what I mean:

Random black boxes appear in our interfaces?


Odd Artifacts appear in most interfaces:


Some interfaces are 100% totally jacked up.


More black boxes?



I'll post details about our setup in the next post

 

Here are specific notes about our configuration that might be relevant to solve this issue:

- We're building a 2D game with an orthographic camera
- We use several canvases to control UI layering. Each canvas is given a sorting layer to control render order


- The images above are taken from the Editor while setup for an Android build
- We're using the following 3 compressions depending on the asset:




- We use the Sprite packer to optimize and pack like-compression assets together to minimize our sheets

If there is any other info that would be helpful to track down this issue let me know. I'll be reverting back to 5.1.3p3 for now... again.

 

We do have some locations where an Image component does not have a sprite set until runtime... but not very many. An example would be in the case of a custom button where the user has a choice to select 1 of 3 mutations. When this interface opens, the system sets the sprites to the correct graphic for the mutations being presented. As far as for masking purposes, no, we're not using any empty images to my knowledge.

I've reverted back to 5.1.3p3 for now, but could try again if you have specific suggestions or when a patch is available.