For some reason the lighting on one of my pieces is acting up. I've rebuilt a bunch of times, made sure I have no overlapping or inverted UVs (for the lightmap). The pieces I'm referring to are obvious. All I did was replace the normal map and color_ID in Substance, so every color is exactly the same.
Diffuse showing that the colors are exactly the same
The lighting issue, after a lighting bake(The section in the background is easier to see)
For testing purposes, I even replaced the top piece with the bottom one, and the lighting seems fine.
And if you look at the materials. They are exactly the same in the editor.
Bottom Piece
Top Piece
The bottom is what it's supposed to look like. I've even moved the top piece to where the bottom piece is to test, and the lighting still ended up the same. Not sure what to troubleshoot next.
Replies
Do you for any instance use mirroring in your UV's? If so, try moving one of the pieces a full UV space to the side.
You could also try checking again the normals in the mesh, one of the surfaces could have its normals skewed and is affecting the normal map.
In the case it's not the normals, try checking the lightmap UV's and upping a little the resolution. Also make sure there's proper spacing between the islands.
Keep up the good work. o/
Bottom Piece
Top Piece(one with issues)
I'm not mirroring any UVs. I also made sure in my lightmap UVs that there are no overlapping or inverted UVs. I bumped the lightmap res from the default 64 to 128 and it made no difference.
EDIT: Showing the flipped Y normals instead.
If it's not even that, then you got me. Perhaps one of the pros around here can give you a better answer.
edit: also try checking the mesh itself. In max just put an edit normals stack and see where they're pointing.
Doesn't seem like anything is out of the ordinary.
You're doing all the islands as a single smoothing group, this is usually bad for these kind of surfaces because it introduces too many gradients in the flat parts, which are prone to errors like bit depth compression banding and others.
Try separating all the angles > 90 and flat surfaces in their own smoothing groups.
On a personal note: You seem to have a lot of wasted UV space on these pieces. Any reason why both parts can't be on the same texture?
I didn't get that seam when everything was smoothed. Also, the bottom piece seems to be working out fine, and it is also one smoothing group which is weird. Or maybe the bottom piece is also incorrect, and can look better. I did split up the smoothing groups on the top piece though, but I did not rebake and apply to the texture. I will try that next if you guys think that also might be an issue.
It did get a bit brighter, but still not the same. Based on this lighting set up, they should be receiving about the same amount of light, correct?
Yeah, in the beginning I think I was just overthinking how it would be a hassle to combine the maps. But just thinking about it now, all I would have to do is model and unwrap both pieces separately like normal, attach them, layout the uvs as accordingly, detach/separate them, and bake them individually in xnormal. Then just combine the normals in photoshop. That the best way?
You're getting the seams in the normal maps because you set the smoothing groups / hard edges but don't have a UV split.
This can completely elimitate the gradients in your normalmap, which are causing the shading issues.
But yeah, you should read that sticky normalmap thread.
Also, combining in ps is not the best way.
I've skimmed through a lot of it. That whole 70 something page write up. And that's where I got confused. I DO have a UV split in that corner. I always have.
I'll try that method of getting rid of the gradients from that thread. Although, I would have never had such strong gradients if there wasn't a seam on the hard edge.
This is the same normal applied to this
UV split where the hard edge is, and a different smoothing for the front and the side faces.
Also, why does the bottom piece, which has a strong gradient as well, seemingly look fine in terms of shading from the top piece. And maybe a better way would be Substance Designers normal combine node, and just export. Thanks for all the help btw
This is with 16px edge padding. Bevel just keeps getting bigger.
16x Edge Padding
Something is not right lol
Yup, just triple checked. Exported my LP and HP form 3ds Max. Brought them into xnormal, used 16x edge padding, opened the same mesh in Marmoset. Applied the normal, same result.
I'd highly suggest though that add a small bevel in your 90 degree edges, to better match the shape of your high poly.
Bakes
Triangulate your LP before you make the cage, maybe when you export the auto triangulate is causing a difference in low poly and the cage, then you'll have an error in xNormal, the geo must be the same. It's kind of a pain to be honest haha.
Create a Cage in 3Ds max and try that one out
1. Make sure that the LP follows the silhouette of the HP as close as possible.
2. reset xforms
3. select the low poly and apply a push modifier
4. increase the push untill the entire HP is overtaken
5. Export the LPWith the push as a cage.
6. Import HP and LP
7. Right click on the LP and use external Cage
8. Try to bake!
[ame]https://www.youtube.com/watch?v=ciXTyOOnBZQ[/ame]
@repete I'm sorry haha. I have watched that video like 6 times already though, I was doing everything right!
I figured it out. I decided to record me doing it so you guys could see I wasn't doing anything wrong. As I was recording it, it magically worked. But I wasn't satisfied with that, so I looked at what I did different. Turns out, I was exporting my cage from 3ds Max without tangents and binormals checked. Relieved that problem is fixed yet pissed off that it was something so simple. Sorry for the trouble, thanks to everyone for helping me out!
Oh, I read through your thread a bit, it's very helpful, thank you! But I was more so looking for a reason as to why my normal map wasn't working as intended, as apposed to finding a way to get rid of my gradients. I know there are a multitude of ways to do everything, and the way I was doing it (hard edge, with a smoothing split and uv spit) was one of them, but it wasn't working when it should have been. Thats the main reason I didn't use that method. I'm sorry.