irrlicht+full screen mode--> corrupted textures when alt+tab

[agamemnus]

When I alt+tab out of my program and back in, all the windows are corrupted -- in full screen mode only. They are made "normal" again by just clicking on them or moving them. I am invalidating the root window with recursion on (when the program detects that I alt+tab back in) in order to fix this, but I am wondering whether anyone else has experienced this and/or if this is normal.

[CrazyEddie]

This is with OpenGL or D3D driver, or both?

It has to do with loss of context when moving from fullscreen mode. Do you notify about display size changes when tabbing away from fullscreen and/or back to fullscreen? If for some reason you don't / don't want to do that, then the fix / work around you have is basically correct, although it does not guarantee invalidation of windows not attached to the root.

CE.

[agamemnus]

I only tried it with DirectX. OpenGL mode is a huge mess on my graphics card...

>Do you notify about display size changes when tabbing away from fullscreen and/or back to fullscreen?

I don't. I'll see if that changes anything.

[agamemnus]

I notified CEGUI about the size change.

Neither method always works...

I have tried the following methods:

1) size change notify
2) setvisible off/on.
3) deactivate / activate.
4) invalidate.
5) update.
6) render.
7) setting the sheet to null and back again to the root.

I think it might be related to the CEGUI Irrlicht module.

Just try it ... in Irrlicht... set full screen to true, and alt+tab (or whatever switches between the program and the OS) around a few times... idk, but you might get corruption.

[CrazyEddie]

I'll try and have a look at it soon (either tomorrow AM or Monday/Tuesday).

What video hardware do you have? Some tests I did for Ogre related corruption did not show up here for me, but I'm going to re-run those on alternative HW when I get the chance. So knowing the HW in use may help to narrow this down or identify common ground.

CE.

[agamemnus]

ATIx600XT.. on Windows.

[Zero]

I've got the same Bug with the CeguiOgreRenderer, so I thing the problem is in CEGUI itself.
I tried it for OpenGL and DirectX 9.
The same bug is for me when I change from Fullscreen to WindowMode.
I notify Cegui that the RenderWindow was resized and some Windows get redraw but one Window disappear nearly 100%...
So the Window should look


But this is all what I see after Fullscreen to WindowMode change.


Regards
Zero

[CrazyEddie]

Ok, guys. Obviously there are a handful of you who are experiencing these issues with corruption of one type or another. The absolute major problem here is that I've tested it on multiple systems and so far I have been completely unable to reproduce the results. I will do more testing, but something is not adding up and it's getting a little frustrating.

@Zero: You need to post the log, because I need to know what version of the code you're using, and also to see which resize values are being passed.

CE.

[Zero]

09/08/2010 14:29:41 (Std) ---- Version 0.7.1 (Build: Jul 23 2010 Debug Microsoft Windows MSVC++ 10.0 32 bit) ----
09/08/2010 14:29:41 (Std) ---- Renderer module is: CEGUI::OgreRenderer - Official OGRE based 2nd generation renderer module. ----
09/08/2010 14:29:41 (Std) ---- XML Parser module is: CEGUI::ExpatParser - Official expat based parser module for CEGUI ----
09/08/2010 14:29:41 (Std) ---- Image Codec module is: OgreImageCodec - Integrated ImageCodec using the Ogre engine. ----
09/08/2010 14:29:41 (Std) ---- Scripting module is: None ----

09/08/2010 14:30:58 (Std) Display resize: w=1024 h=768

I hope that helps.

[agamemnus]

Ok, a wild shot in the dark..

When CEGUI does the redraw when the user flips back to full screen, does it turn off mipmaps for that too?

[CrazyEddie]

@Agamemnus: I think the the mipmap setting is a property used when creating the texture; so whenever a texture is created the option for mipmap creation is set to false.

@Zero: I would recommend updating from the released 0.7.1 code to either a snapshot release from the stable v0-7 branch, or to use the stable v0-7 branch. While I certainly can't say that the specific issue you're currently seeing is fixed, there are definitely some rendering issues fixed (so you may update and find nothing changes, but then again you could get lucky ) If you're reluctant for some reason to use one of those options, there will be an actual 0.7.2 release made by the end of the month

CE.

[Zero]

Thank you, the 0.7 Branch solved all my problems

[Zero]

Okay, after some testing with the team, we found another bug, where all get corrupted. This happen when we resize the Window at runtime.
But like the other bugs, i cant click and the windows to come back.

Zero

[CrazyEddie]

I'm sorry to hear that you continue to get issues related to resizing the host window / display size. Do you see this in only your own app(s), or also in the CEGUI samples (if you have those compiled). If it's just your own stuff, I wonder if it might be possible to create a test app that exhibits the behaviour so I can run it here to see if I can reproduce the issue that way (since everything else I've tried has given successful results for me).

Also which version of Windows are you guys using?

CE.

[Zero]

I use Windows 7.

Zero