Lightmap seam along inner edge

So, I have some environment props all prepped for texturing and I thought I’d lay out the lightmap UVs today - I read at least ten articles on correct lightmapping techniques yesterday just because of painful past experiences. I have a building part which has extruded wall details - iteration after iteration, I decided to cheat a little and separate those extruded bits off of the mesh and give them their own lightmap space. That seemed to fix most of the bleeding, but now I have this:

As these bits are pretty simple I decided to stick to the default 32 resolution, but cranking it up didn’t give me any improvements. The second UV channel definitely has enough padding, I think I left 5-6 pixels between all shells:


Normals are pointing the correct way, soft/hard edges seem to be correct as well (although I did try hardening the soft edges and vice versa, no change though). I snapped all edges to the grid divisions, and the grid’s set to 0.03125 (=1/32). I baked the lights at production quality, and I just can’t think of anything else that may be causing this :frowning: I even tried auto generating the UVs in UE, but we all know how well that works.

Any known solutions?

P.S. Some may say that edge is unnecessary - yes, once I extruded those bits off of the main mesh they became redundant, but the point remains, that this shouldn’t be happening as far as I know.

Based off the mesh screenshot it doesn’t look like the UV’s are correct. Also make sure your object is static and not movable.

I’m not sure what you mean by ‘not correct’? The 1st UV channel is fine (and as far as I know doesn’t influence light maps?), and I followed documentation/article guidelines for the 2nd channel UVs that I posted. The object is set to static as well.

Just tried the meshes out in 4.5 (I’m currently working with 4.3) and I even backtracked to UDK - both give the same result, so it’s definitely something with the mesh, but I cannot imagine what it is! I went back to the 2nd uv channel and gave it an extra pixel for the border padding (so there’s 2 pixels around the edge of the map, as I understand that would make it 4 in between maps when they are squashed into 1 during baking?), but I still get bleeding on almost every duplicate of the object, just in different places. I don’t understand what can be causing this cause I think I have a clean unwrap for both channels, all the verts are snapped to the grid and the grid’s set up properly.

I even went on to test a single cube with 222 divisions and I just went crazy with the 2nd UV channel, no snapping, different shell sizes, and guess what, that baked out perfectly! :rolleyes::confused:

When removing the detail parts of the mesh…could it be that the lower was somehow detached…meaning, are there double vertices between the upper part and the lower part, that are in fact not connected, effectively making it two different sub objects in 3ds max?

Hm I don’t know anything about sub objects, 'cause I’m using Maya for modeling. Although I did make sure to run a mesh cleanup and merge all the vertices just to see that the mesh is clean though.

Now I’ve tried not grid-snapping the edges that are supposed to be soft. Still the same old light bleed, i.e. me running out of ideas and google articles to read :frowning:

what about smoothing groups? xD

if that doesn’t help…can you send me the fbx for further analysis?

They are done and there. They are correctly exported in the fbx as well.

Just found out that I can visualise lightmap density in the viewport, the mesh on the left is the source - I transfered the UV attributes from it to the other ones in Maya; they are all essentially the same mesh though, so I’d assume the density should be the same, but it seems very distorted on some meshes, could this be what’s causing the bleed?