[UDK] Inconsistent specular from mesh rotation

Gyrfal

So I've been working on an environment that has modular tiles with a highly specular material on them. I'm using a few different textures to make a pattern on the floor with the tiles, and noticed that when I rotate them and bake maps, the specular seems to come from different directions.

I did some research, and found out that the specular is baked to channel 0 in UDK. Tweaked with swapping the texture/light map channels like some places have suggested, to no effect. After paring down as many variables as possible, this is what I've arrived at.




They are plain cubes (scaled up) from Maya. They have identical UVs in channel 0 and 1, that haven't been altered from the default UVs that the cubes come with. The light map is set to channel 0 and the texture map is set to channel 1 (not that there is a difference between the two). Switching them doesn't solve this problem. From what I can tell this is a result purely from rotating the cubes to different orientations.

So is there a fix for this? I've been at this for what seems like hours, and can't accept that rotating a mesh in UDK is not possible without making the specular look bad.

Help would be much appreciated. Oh, also I'm on the most recent (August) version of UDK.

Gyrfal

Forgive the bump to the first page again, but I really feel like it's just a matter of some option that I'm overlooking.

Eric Chadwick

Guessing here, but the UVs might need to be non-overlapping in the lightmap's UV channel for UDK to work properly. Have you tried this?

Gyrfal

Yeah, in the example it's the default UV layout when you make a cube, so four squares vertically and the last two faces sewn to the sides. Nothing is overlapping at all, thanks for the suggestion though.

sprunghunt

Gyrfal wrote: »

I did some research, and found out that the specular is baked to channel 0 in UDK. Tweaked with swapping the texture/light map channels like some places have suggested, to no effect. After paring down as many variables as possible, this is what I've arrived at.

You shouldn't be seeing this behaviour. I doubt it has anything to do with the unwrap as you can usually do more complicated stuff than this without breaking it. And broken lightmap UVs won't necessarily break the specular. It's probably something to do with the normals. I'd try making another cube.

what happens when you convert these to an interp actor (mover) or kactor? these use a different lighting system as they're dynamic. So specular will be calculated differently.

Gyrfal

Changing them to movers/dynamic fixes it...in a sense.



Also changing the light to a dominant point light does fix it.



So does that mean it's just because I was using a non-dominant point light? Should I not bother with non-dominant lights anymore?

sprunghunt

is the second picture static meshes with a dominant light? or are they kactors/movers?

I think you are supposed to have one dominant light per scene for proper lighting.

Gyrfal

The second picture is static meshes with a dominant, the top picture is with movers.

I might post pictures later, but in the scene I'm working on using a dominant point light makes a whole bunch of weird issues. The primary specular is fixed, but specular from bounce lighting off the environment is still messed up, probably for the same reason as the initial issue.