Substance Painter, 3DS Max - UV seams with smart materials
polycounter lvl 6
Hello there, so as my title says, I'm am having some issues with the UV seams being highly visible within Substance Painter. I am only a beginner using the Substance programs and also with the texturing process and have a very small understanding of the baking process.
What I have found is that there is a visible seam line when I place a smart material onto a mesh that has been unwrapped within Substance.
Pictured above is how that looks with the uv layout as well.
Now to my understanding, a high poly model does not need to be unwrapped for the smoothing to work and it should possibly have a singular smoothing group? However I am not entirely sure on that last part. For the low poly, Am I correct in saying that the smoothing groups can be separate? or does the mesh have to have one smoothing group?
Now if I place a smart material onto the mesh it breaks up the sections by the UV areas. it also seems to ignore the corner smoothing from the normal map as pictured below
This is fine however I noticed that by changing the projection method from UV layout to Tri-Planar Projection, the smart material seems to do some areas correctly but there is still seams, as pictured below:
So my question is, how do I go about fixing this, or is there any specific things I can look for to aid in my fixing of this.
Thanks.
What I have found is that there is a visible seam line when I place a smart material onto a mesh that has been unwrapped within Substance.
Pictured above is how that looks with the uv layout as well.
Now to my understanding, a high poly model does not need to be unwrapped for the smoothing to work and it should possibly have a singular smoothing group? However I am not entirely sure on that last part. For the low poly, Am I correct in saying that the smoothing groups can be separate? or does the mesh have to have one smoothing group?
Now if I place a smart material onto the mesh it breaks up the sections by the UV areas. it also seems to ignore the corner smoothing from the normal map as pictured below
This is fine however I noticed that by changing the projection method from UV layout to Tri-Planar Projection, the smart material seems to do some areas correctly but there is still seams, as pictured below:
So my question is, how do I go about fixing this, or is there any specific things I can look for to aid in my fixing of this.
Thanks.
Replies
There is also the wiki
To answer specifics.
1) No, a high poly doesn't need to be unwrapped. Side note - It can be however, such as if you wanted to transfer a texture by baking.
2) You can smooth the high poly however you want. Generally most subdivided models are 1 smoothing group (averaged normals).
3) Low poly smoothing groups/normals are important to take advantage of. The rule here is that when you split smoothing groups, you must have a UV seam where the normals are split. The reason why becomes clear after understanding the process. More on this below.
4) You can theoretically use 1 smoothing group for the low, but it has drawbacks in practice.
5) Tri-Planar Projection is having some problems due to your normal map, but also note that the way it works means the camo pattern won't be seamless the way you may think. It will be blended projections of the same texture from the cardinal xyz directions. You may already realize this but it's worth touching on.
Now on low poly shading with normal maps.
If need be, while reading this take a gander at information on how baking works by casting rays, and how normals (for models AND maps) work. Trying to condense this info is hard lol. I've written some on this stuff before so hopefully this here covers the bases and less documented information:
Essentially what it all comes down to is that with tangent space normal maps (the kind we all normally use) the smoothing (normals) of the low poly mesh dictates the values (color) in the normal map. A TS normal map is actually computing the per-pixel difference between the high poly and low poly's mesh normals. The closer the low poly's normals are to the high's, the less difference you'll see in the final normal map. Keep this in mind!
The goal is to control the low poly's smoothing so that you don't make the normal map do too much work and have large, smooth gradients in it. These gradients often don't play well with texture compression, mipmapping and LODs, can be considered harder to work with, and can become inaccurate without 16-bit color. The tangent basis must also be synced or else strong normals are wildly innacurate.
What you do to manage is split the low poly's normals at big angle changes. Some people do this simply by some threshold like 45 or 60 or even 90 degrees. At an advanced level you can set smoothing groups manually. The catch is that if you don't also split the UVs at the same locations you'll have normal values bleeding over into the wrong areas, creating seams. Thus, the more smoothing splits you have the more complex your UVs can become.
So theoretically you could use 1 smoothing group on any low poly model, but it's problematic in practice/production when you push the normal map too far. Obviously if you have something like a simple spherical shape where there are no drastic angle changes 1 smoothing group would work totally fine. Like I said it's all about making the high and low mesh normals similar from the start!
Finally:
One question to know to help fix your bakes is whether or not you are using a cage, but go ahead and try to follow this post up with more tests!