Normal Map baking issue (compared to Substance Painter)

greenAnka

Hello! I'm trying to achieve a perfect result in normal map baking, however, I noticed an issue while baking it in Marmoset 3.08
on the pictures, you may see the difference of baking in Marmoset (M) and a perfect result in Substance Painter (SP)

do you have any ideas of what I might be doing wrong?

cheers.

EarthQuake

Try changing the tangent space in Toolbag to Xnormal / Mikk, this should match with the tangent space used in Painter.

greenAnka

EarthQuake said:

Try changing the tangent space in Toolbag to Xnormal / Mikk, this should match with the tangent space used in Painter.

this could work unless in the new marmoset version is already set to mikk

greenAnka

EarthQuake said:

Try changing the tangent space in Toolbag to Xnormal / Mikk, this should match with the tangent space used in Painter.

the problem is not how baked NM looks like in Painter (and vice versa). the wobbly NM from marmoset is wobbly in Painter they simply bake in different way as I can see.

EarthQuake

greenAnka said:

EarthQuake said:

Try changing the tangent space in Toolbag to Xnormal / Mikk, this should match with the tangent space used in Painter.

the problem is not how baked NM looks like in Painter (and vice versa). the wobbly NM from marmoset is wobbly in Painter they simply bake in different way as I can see.

Yes, this is why it's important to set the Tangent Space to Mikk. Sorry if I wasn't completely clear, but you need to set the Tangent Space in the Baker Object and rebake.

Also, make sure your low poly mesh is triangulated before importing into both Toolbag and Painter. If not, the quads may be triangulated differently be each app.

Some more info on general baking techniques here: https://marmoset.co/posts/toolbag-baking-tutorial/

greenAnka

I double-checked the tangent space in Marmoset: it was set to mikk both in baker options and lowpoly object.

I also checked your recommendation as for triangulation of the lowpoy. So the following experiment gave me such results:
1) I created the sphere with uv such as half of the sphere without uv cuts and the second half - full with cuts (see the attachments)
2) I triangulated the lowpoly sphere
3) baked both in marmoset and painter (i import everything in painter to check out, that's why all the screenshots are from there)
looks like no matter what the seams are always visible if the object becomes slightly metallic)

greenAnka

greenAnka

I watched your video "controlling the shading behavior"
so on uv splits you will always have a minor (but still visible) uv seam, right?
in this case, how do people in production split 1 face into 10 uv tiles without getting the seams in the end?

EarthQuake

Yes, generally speaking that is correct, wherever you have a UV split you will have a visible seam if you look close enough to see it. Or if the texture resolution is low enough to make it obvious.

Generally speaking it is wise to minimize the amount of UV seams (though balancing amount of seams vs distortion is important), to place them along natural seam areas (where there is a shift from one surface type to another), and to hide them in less visible areas.

greenAnka

...

greenAnka

thank you a lot for the explanations! very useful!

I've got one more question: 
I work in maya. before exporting the mesh to marmo I averaged the vertices, triangulated. but baking the normal map gave me the following artifacts. 

when I separate the collar from the coat that part is baked ok

when I took a look at the mesh in maya, I also haven't found any shading problems

greenAnka

EarthQuake

It looks like there is extreme shading on this mesh, I would recommend adding some hard edges along your UV seams so the mesh normals aren't so extreme.