MikktSpace workflow

[klo-works]

hi guys,

since both engine unreal and unity sync with mikkt.

https://www.youtube.com/watch?v=gR3r7Xmhmlk&list=PLB0wXHrWAmCx994Cb7iRFSmupYHFw5DTx&index=21

I want to know if the work flow is right I watch the videos and there some part I need to clear it.

first the work flow (correct me if I am mistaken) :

- no need to focus on 90 degree you can attach uv map with 90 degree.
- make 1 smoothing group.
- export it as fbx with tri and with unchecked (tangets and bionormal)
- Baking must support Mikkt like (xnormal or substance painter or any software that support mikkt).
 - Never Ever test your normal map in max or maya test it in unreal or marmoset.

move to my question:

-is it A MUST to make one smoothing group ? if let say there is 2 different uv island and I want to give it different smoothing group is it ok ?

- is there a rule to make multiple smoothing group with the mikkt ? in which cause I make different smoothing group ?

-is there different when baking with substance mikkt and xnormal mikkt ? is it gonna be  the same result when putting it in unreal ?


thanks 

[Froyok]

-is it A MUST to make one smoothing group ? if let say there is 2 different uv island and I want to give it different smoothing group is it ok ?

You can have multiple smoothing group, but that will produce seems between them if you don't bake with a cage.

-is there different when baking with substance mikkt and xnormal mikkt ? is it gonna be  the same result when putting it in unreal ?

If you use the correct settings, you get pretty much identical results. I don't guarantee it will be 100% identical because I know we don't generate padding the same way or because we also perform some different calculations in certain cases. The main advantage here is that with our bakers you get the matching by name feature. Also our baker are a bit faster than xNormal usually.

Still, what you see in Painter should be what you get in Unreal/Unity.

[klo-works]

You can have multiple smoothing group, but that will produce seems between them if you don't bake with a cage.

cage I need to use it even with one smoothing group right ?



doing different smoothing group as in the image above will be ok and not have any problems right ?

If you use the correct settings, you get pretty much identical results. I don't guarantee it will be 100% identical because I know we don't generate padding the same way or because we also perform some different calculations in certain cases. The main advantage here is that with our bakers you get the matching by name feature. Also our baker are a bit faster than xNormal usually.

Still, what you see in Painter should be what you get in Unreal/Unity.

about padding it use gradient padding but how I decide edge padding is it call Dilation width ?

if I add the dilation width 16 it mean 16 pixel ?

 is this method  will be ok if generate MIP map since it use gradient padding ? basically is (Gradient padding better than the old way ?



  I ask a lot because I am planing to use substance painter as workflow for my fyp,so I want to understand it.


 

[Froyok]

The dilation parameter is indeed the padding and the slider indicate how many pixels. Note however that in addition to the padding we also perform a diffusion (sort of blurred background from the UV borders). So setting the dilation to 1 pixels is enough.

Anyway, the padding was just an example, no need to worry about it. What matters is if the mesh looks right in your target engine. The default settings of our bakers should work fine most of the time.

[klo-works]

The dilation parameter is indeed the padding and the slider indicate how many pixels. Note however that in addition to the padding we also perform a diffusion (sort of blurred background from the UV borders). So setting the dilation to 1 pixels is enough.

Anyway, the padding was just an example, no need to worry about it. What matters is if the mesh looks right in your target engine. The default settings of our bakers should work fine most of the time.

Thanks a lot I really impressive specially the baking it way fast and not that only,the result was really Nice compare to xnormal when I try in Unreal  .

enough asking I will start my fyp with substance workflow and if I find problem I will ask. Iam glad I buy it. 

[klo-works]

The dilation parameter is indeed the padding and the slider indicate how many pixels. Note however that in addition to the padding we also perform a diffusion (sort of blurred background from the UV borders). So setting the dilation to 1 pixels is enough.

Anyway, the padding was just an example, no need to worry about it. What matters is if the mesh looks right in your target engine. The default settings of our bakers should work fine most of the time.

one last question about padding.
you said setting dilation to 1 pixel is enough
how far I should make space between uv Island  ?


specially if Iam using LOD or MIP
dokeep use this rule:

128 = 2px
256 = 2px
512 = 4px
1024 = 8px
2048 = 16px

..

[Froyok]

The dilation is 1 pixel yes, but the diffusion will happen behind, so keep the space between your UV islands like you used to do.

[elowan]

even, when it´s like 10px @2k res?

[Froyok]

even, when it´s like 10px @2k res?

I'm not sure to understand what you mean.

[elowan]

Klo-works asled, how auch distance one should have between uv islands, because you wrote about 1px dilation was enough.

So I wonder, how much space should be left in between the islands. Should I stick with common rule (like klo-works listed) or can it be less pixels ?

Like 2048 texture size would normally be 16px space (rule of thumb) - would you suggest to get below 16px or not?

Kind regards

[klo-works]

^^^
yeah I want to know the answer too 


+ I test today with this object


xnormal and S painter


in xnormal it show little shading effect in 2048 and if using like mip 256 it noticeable





but in S painter the reult is clean even with 256 still clean and that impressive 



but I don't know why in xnormal show some shading is suppose to be fine since it use MikktSpace .

[Froyok]

Klo-works asled, how auch distance one should have between uv islands, because you wrote about 1px dilation was enough.

So I wonder, how much space should be left in between the islands. Should I stick with common rule (like klo-works listed) or can it be less pixels ?

Like 2048 texture size would normally be 16px space (rule of thumb) - would you suggest to get below 16px or not?

Kind regards

As I said, the dilation of 1 pixel is enough because there is the diffusion process that is applied behind. There two post-process with the bakers that happen to fill the empty areas between the UVs.

So no : you need more than one pixel of space between your UV islands. That's what I meant when saying "keep the space between your UV islands like you used to do". Keep 16 pixels on 2048 texture. Otherwise you will have troubles with the mip-mapping.

[rosenand]

You can have multiple smoothing group, but that will produce seems between them if you don't bake with a cage.

I thought that's what the Average Normals option in the Baking window / Common parameters is for - so that you can have multiple smoothing groups, but the projection rays still get averaged as if you had one smoothing group, so you don't get gaps and whatnot.