Same thing here; tried a bunch of different drivers, another pc, another project, even tried windows 11 to see if it happened there too… just a whole bunch of different things; and there’s no replication for the bug, it just happens sort of happens to textures and/or their mipmaps after a while on projects running D3D12.
At first i thought VRAM intensive projects might have something to do with it, but it even happened in a project that is absolutely not VRAM intensive.
Saving/refreshing the texture gets rid of it, but it’ll just happen again to another texture, the only way to get rid of it for a good amount of time is an engine restart.
No idea what causes this. Would love to officially hear from Epic on this issue.
I have this issue, turning the texture MipMaps to “NoMipMapping” solves the issue and seems to be related to that. Repeat for any textures affected. I’ve found it present in just about all texture map types.
Not ideal if you rely on mipmapping for textures, but does fix the problem.
I don’t mean to hijack this thread - I apologize in advance - but this is not unique to UE5. A large number of World of Warcraft players are seeing similar corruption in DX12 mode; for example:
I’ve also had this issue appear in Atomic Heart once (which is a UE4 title), but I believe this was when a copy of WoW was running in the background.
Has anyone been able to reproduce this issue on a Radeon or Intel card?
Also, if you have WoW installed, an easy way to reproduce it is to dual-box 2 accounts. After about 20-30 minutes, it’s almost guaranteed. I’m on a 3070, so I have 8GB VRAM. With no games open, but some browser apps open (Chrome, CurseForge, Discord, etc) open, I tend to be around 20% VRAM usage. One copy of WoW alone will eventually experience the issue, but two copies reaches nearly-full VRAM very quickly. Meanwhile, the textures on-disk are good - a restart of WoW can show the same scenes fine.
As such, I suspect this is a memory pressure/texture loading issue - perhaps some VRAM is being deallocated but not written to before texture decompression? It might be less about total VRAM usage and more about allocation/deallocation or multiple requests at once?
taking away the atmosphere, direct light, clouds, etc.
Redo a texture.
Very randomly when closing and opening the game.
Note: When the corrupted error texture shows and I exit the world and rejoin the world the error still shows. When I view the texture within the mesh, the corrupted error shows on one of the textures. I would also get dark black lines when playing and walking, the lines follows my player. If anyone has the awnser to this problem, Please let me know.
I have another solution: I just change the Texture Group to something else and then back to the original. Seems to fix the issue as well. - still anoying because it happens randomly to textures which were not “fixed”
Just wanna leave a bandaid remedy here, you can fix the corrupted textures or other assets by going to the Asset Browser, right click on the asset and click Reload. Of course this does not prevent textures from getting corrupted again.