1
$\begingroup$

For a few months I'm researching and making little tests to better understing and optimize a LowRes Workflow, And this "error" continue to occurs on this particulary model that I'm working on, It was baked using tangent space, I tested in Substance Painter and Unity to see if It was a Blender Related Shader issue, and It was less predominant on the others aplication that a tested on.

But that was enough chit chat, here's a screen grab.

Principle Shader and Normal Map

And thank you for reading! Have a nice day.

$\endgroup$
3
  • $\begingroup$ In which application did you bake the normals? Can you show the bake settings? Can you show the material nodes of the screenshot? Can you show the normal map? $\endgroup$
    – Leander
    Commented Jul 30, 2019 at 6:55
  • $\begingroup$ Did you set the Color-Space on the normal map image node to Non-Color (or Raw)? $\endgroup$
    – Jackdaw
    Commented Jul 30, 2019 at 8:22
  • $\begingroup$ Here man, I will Post some screenshots, and I baked Again I Guess The bake was set to linear and now It's Non-Color, But the seam still appears, but not so glitch like yesterday. $\endgroup$ Commented Jul 30, 2019 at 15:44

2 Answers 2

1
$\begingroup$

A likely scenario would be if the normals were inverted. That happens for example if the geometry is inverted(scaled by negative values) at some point during modeling to mirror it. You should go to edit mode (tab), select everything (a) and recalculate normals before baking them. The hotkey is ctrl+n in 2.79 and shift+n in 2.80. You can access functions related to normals in the ctrl+f or Mesh > Faces and Mesh>Normals menus in 2.79, while in 2.80 there is a dedicated Normals menu alt+n.

However in this particual case the issue is the normal map made with overlaping UVs for symetrical geometry. If you have a look at a normal map of a symetrical object, the normals are not mirrored - they differ on each side so if you mirror them you get an incorrect result.

enter image description here

$\endgroup$
5
  • $\begingroup$ The Normal Are fine, I cheked but forget to tell in my First Comment $\endgroup$ Commented Jul 30, 2019 at 15:45
  • 1
    $\begingroup$ I saw the screenshot of the normal map in your answer. See my updated answer. Please share all the relevant info like that from the beginning with your questions. That would save time... $\endgroup$ Commented Jul 30, 2019 at 19:13
  • $\begingroup$ Yeah, i guess nothing that can be done with the apparent seams along the car, but In This case the roughness with textures will make less predominant, and Sorry for not mention in the beginning of the question $\endgroup$ Commented Jul 30, 2019 at 21:59
  • $\begingroup$ You can make a UV map without overlapping UVs, re-bake normals and it will work fine. $\endgroup$ Commented Jul 30, 2019 at 22:10
  • $\begingroup$ I made this way to better optimizing the UV layout, so I only baked half of the model. $\endgroup$ Commented Jul 31, 2019 at 5:47
0
$\begingroup$

I guess It was the color space, I originally baked in 2.79 but I guess It was set in linear, Now It's set to Non-Color, but The seams still appears, Not so glitch Like before, But still there.

enter image description here

enter image description here

$\endgroup$
2
  • $\begingroup$ That seam looks like internal faces along the centre line, like this. Is the merge option enabled in the mirror modifier? $\endgroup$
    – sambler
    Commented Aug 7, 2019 at 8:30
  • $\begingroup$ Yes It is, there are no duplicated faces of any kind, I guess It's because I Baked only half of the HighPoly Model, but I'm not so sure $\endgroup$ Commented Aug 8, 2019 at 17:19

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .