HISMC bug, always shows lowest LOD

Just noticed I forgot to comment here, thanks very much for fixing that so quickly Tom!
I saw your commit directly after it appeared in the 4.15 branch, was very happy to see that it was actually not intended that the screen size couldn’t be more than 1 :slight_smile:

The new screen size calculation for both static meshes and HISMCs is great by the way, I still remember how it had issues before with the screen size changing the further the mesh was displayed towards the edge of the screen. No problems like that any more in 4.15.

Hey , there was a fix added to dev-vr on 2017/03/17 by Ryan Vance that fixed a bug with VR that different LODs are sometimes visible per eye. Now, as always when something related to LODs is changed, HISMCs seem to be special, and the fix did not fix the same issue with HISMCs. So with HISMCs, it still happens that the left eye sees a different LOD than the right eye.
Could you take a look at that? :slight_smile:

Hi John, I entered a ticket UE-43763 for this.

Thanks !

Just to be sure, UE-43763 mentions “Instanced static mesh”, I never tested though if ISMCs do have the same issue. ISMCs also aren’t really relevant since those aren’t really meant to be used with LODs I think.

Its a huge issue with HISMCs, since all foliage are HISMCs and I often see that the left eye shows LOD2 thats’s a mesh, and LOD3 thats the billboard, and I hope you can imagine how it looks like when one eye sees a mesh and the other eye a flat billboard, it completely breaks the stereo effect of VR and quickly makes you sick.

And since performance is really important in VR, its obviously no option to use static meshes for all foliage :slight_smile:

Sure, I just modified the title to clarify. I saw some offline discussion which mentioned the eye consistency was only working correctly for regular staticmeshes and not for ISMC, HISMC or skeletal meshes.

Worked like a charm for me. I can’t believe I actually found an answer to this. Except I only needed to increase the scale to 5 not 10,000.

Hey , could you take a look at this new, quite horrible, regression with HISMCs in 4.19 and 4.20?