Mirrored uvs but no seam....

Chase

I'm not complaining, but I'm wondering why this is. I test baked two meshes in Max and am viewing them in UDK. One mesh has mirrored uvs and a copy that has no mirrored uvs. They look the exact same. If this is the case I'd obviously want to keep mirroring when possible. I'm in the process of matching texel density of my model so before I go through editing uvs I want to know why I'm getting these results.

Here's the mirrored uv model. I put hard edges based on my uvs, making sure the mirrored parts shared the same smoothing group:


I applied a random tiling diffuse and am not getting a seam:


However, the seams are showing up where the hard edges are, which should be expected. I mention the Handplane thing cause then I don't have to worry as much about smoothing splits/uv splits. I know there's more to it then just this so maybe I should have just left that part out:


Am I not getting a seam where the mirroring is happening because I used an Average Projection? I have read up on Average Projections vs Explicit Projections in Earthquake's thread, but must not be understanding them fully. Average Projection means the cage is going to ignore the smoothing splits of your lp so the bake doesn't have these gaps. By gaps does that mean I won't get seams? Explicit Projection means the cage will transfer these smoothing splits to the normal map, and when you apply the normal map you'll see these gaps? As always thank you again guys!

passerby

hard edges will be hard unless you got a normal map to override shading.

What you got in pic 3 isn't a hard edge as far as normals go, it is that the texture is not matching up.

i also think you misunderstand how handPlane works, it wont fix the texture not matching up at the seam, all it does is take a worldspace normalmap, and convert it to a tangent space normalmap with the tangent bias for your engine of choice.

as far as normals go there are many ways around the seam problem, handplane is one of them, or just using a averaged cage projecting, and spilting normals at your UV seams, and than exporting the tangent bias with your fbx to udk.

timotronprime

Chase wrote:

By gaps does that mean I won't get seams?

By seams do you mean gaps?

This is a gap (along with aliasing):


Your tiling diffuse wouldn't affect if it has a gap or not. A mirrored UV doesn't automatically mean it will have a normal gap. You set up the projection correctly so there is no seam. Or it's a flat surface anyway and the vert normals have the same pole for both sides, thus not having a visible normal seam.

For your 3rd image, there is a hard edge there but it is avoidable and Handplane has nothing to do with fixing normal gaps, it just helps with converting your normal maps to work in sync with the target engine's normals algorithm.

An averaged projection mesh overrides your game mesh's vertex normals and uses those for the baking process.

Explicit projection does not mean it will put gaps in your normal maps; it means it will use your game mesh's normals for the projection instead. Consequently, there will be likely be normal gaps caused by the resulting map as the game mesh isn't necessarily built with averaged normals for projection purposes (hence, another mesh is created for this).

I'd suggest you go back and read the first few pages of that thread and actually experiment with a simple mesh to get a better understanding.

EDIT: seeing passerby's reply, I was assuming you have a normal map for those corners in the 3rd image. If there isn't, then that's a UV seam that has nothing to do with using a normal map to fix it.

Chase

Passerby - I've done plenty of researching and watching Alec's videos already. I was posting from my phone I just didn't explain what I know about Handplane fully. I know it converts an object space map to a tangent space thereby giving you a tangent space map as if your baker and engine were synced. You're also not forced into having as many uv islands as you normally would meaning you can stitch uvs you couldn't otherwise stitch if you weren't synced or using Handplane. You still want to have smoothing splits at the uv splits cause they're still free. About the seam not being a seam, then am I just not gonna have seams with an Average Projection? With a uv modifier applied you can see where the seams are (green lines) based on where the uv shells have been split. I've done plenty of googling about unwanted seams showing up and I always thought a seam occurred based on where you split the uvs. It might be easier if I upload the fbx.

Tim - By seams I mean gaps if that's the same thing. Otherwise I don't know what these "gaps" are. By saying "setting up the projection correctly" do you also mean because I used an Average Projection? So the cage is like it's just one whole smoothing group where Explicit Projection is based on where the sg's are on the in game mesh? I don't have a normal map for the Diffuse. Just a normal map for the bake of the hp for the in game model. If I am getting this right you guys are saying that as long as I at least use an Average Projection I won't need to worry about normal map seams. The only seams I have to worry about are the hard edges are. Or am I off base again...

timotronprime

A split normal is result from using a hard edge (Maya) or different smoothing groups (Max).

Gaps in your normal map are a projection artifact caused by using a split normal/hard edge; with a cage, it uses the cage's normals instead (which are smoothed), thus eliminating that artifact. If you use a cage, you shouldn't have to worry about gaps on hard edges in your normal maps.

Chase

Awreome. That makes sense. I knew about the cage being smoothed but didn't fully understand the rest til now. As long as you're using a baker that uses an Average Projection then you don't need to worry about having gaps in your normal map.