Subdivision displacement stops working when mesh is moved in sub-object mode.

[LudvikKoutny]

Hi,

I've discovered quite severe displacement bug which almost drove me crazy today. When using subdivision mode of displacement, as soon as you move mesh any significant distance in sub-object mode, displacement immediately stops working. It appears as if pre-sampling of quads to subdivide was performed with pivot point offset being completely ignored.

The problem here is that on the same mesh I am using displacement on, I am also using displace modifier with distance texture, and distance texture works in world space, not texture space, so if I move my mesh in object mode, mapping created with Distance map is on the wrong place, therefore moving it in sub-object mode is only option.

I can not use 2D displacement because it does not support non-trivial map networks, and I can not use 3D displacement because it instantly fills up my RAM and makes my OS crash, so subdivision is the only option.

Therefore, I really need to know how to get this working.

Here is an example scene with two planes. One plane has pivot at its center, while other, right plane, has been moved in sub-object mode, so its pivot is somewhere else:

I am attaching the scene as well
DisplaceBug.zip

[Miro]

I can reproduce it and will log it as a bug, but in the meantime, as a workaround, you can try to turn off View-dependent and reduce the edge length or just reduce the edge length.

[LudvikKoutny]

I can reproduce it and will log it as a bug, but in the meantime, as a workaround, you can try to turn off View-dependent and reduce the edge length or just reduce the edge length.

Thanks, but this was just a simple repro scene. In my production scene, I am displacing 1.6*1.6km terrain where camera is about 2 meters off the ground, I am not sure turning off view dependent would be a good idea in that case

[Miro]

That's a shame. It's about the only workaround I found. Have you tried turning off Static geometry when in 3d mapping mode, to keep the memory consumption lower?

[LudvikKoutny]

That's a shame. It's about the only workaround I found. Have you tried turning off Static geometry when in 3d mapping mode, to keep the memory consumption lower?

Given how many polygons would uniform tessellation produce on 1.6km^2 ground, I don't think that even V-Ray's out of core rendering could handle that, and if it did, my rendertimes would probably slow down to crawl. I can try it, but I doubt it will work.

[LudvikKoutny]

Yep, so I tried and no luck, light cache just hangs at the start and estimated rendertime starts to grow towards years.

[vlado]

Thanks for the scene, looking into it.

Best regards,
Vlado