Weard vertex shading after bake in substance?

Zaikoni

So my workflow is that i create lowpoly first then i create highpoly and then i start unwrapping lowpoly and after i have unwrapped it i load smoothing grps from uvw shells. then i do explode bake based on naming in substance.

Substance after i baked normals : 
Material Mode in substance:


Heres is my lowpoly :

Highpoly:

Here is wireframe mode with hard edges showing in 3dsmax:
As you see the shading seems to appear in blade only.

musashidan

The shading artifacts are very apparent in your low-poly so that will remain after your bake. This can be less noticeable or avoided completely if using a synced(non-averaged) workflow but, from your info I'm presuming a UV/SG split(averaged) workflow, so fixing the shading on your LP is first and foremost. Although from your LP screenshot it looks like you don't have smoothing groups applied?

Zaikoni

musashidan said:

The shading artifacts are very apparent in your low-poly so that will remain after your bake. This can be less noticeable or avoided completely if using a synced(non-averaged) workflow but, from your info I'm presuming a UV/SG split(averaged) workflow, so fixing the shading on your LP is first and foremost. Although from your LP screenshot it looks like you don't have smoothing groups applied?

Here is my lowpoly with unwrap: https://gyazo.com/f71ac45e34f35eaf797acd30501f0a01 the reason it looks like it doesnt have smoothing grps because i load smoothing grps based on uvw shells hmm i wonder if i'am missing something here. So with this workflow UVW/SG (averaged) i need to have LP'S shading corrected? like then i would need to split uvws more in order to get that shading away. Beause i did few uvw splits and loaded smoothing grps from uvw shells again and the shading was better after that.

Zaikoni

musashidan said:

The shading artifacts are very apparent in your low-poly so that will remain after your bake. This can be less noticeable or avoided completely if using a synced(non-averaged) workflow but, from your info I'm presuming a UV/SG split(averaged) workflow, so fixing the shading on your LP is first and foremost. Although from your LP screenshot it looks like you don't have smoothing groups applied?

I gave this part it's own smoothing grp :

heres non wireframe :


as you can see the shading is much better still get really small ones but not that noticable like these when going on certain angle

Eric Chadwick

If you embed your images, it makes things a lot easier for people to solve your problem.

Just right-click an image and copy the full url, including the extension (.png or .jpg) then use the Attach Image button.

Zaikoni

Eric Chadwick said:

If you embed your images, it makes things a lot easier for people to solve your problem.

Just right-click an image and copy the full url, including the extension (.png or .jpg) then use the Attach Image button.

Thx for the advice i fixed that now.

Zaikoni

Update picture after i did that small smoothing grp change :

Zaikoni

Anyone know how to improve it more? There is still small shading errors in the blade.

Obscura

Control loops or smoothing splits would help. Though, I do it the other way around. I set up the smoothing first based on how hard the angle changes are, how does it shade, and put uv splits here and there based on the smoothing.

Have you seen this thread?
http://polycount.com/discussion/107196/youre-making-me-hard-making-sense-of-hard-edges-uvs-normal-maps-and-vertex-counts/p1

Zaikoni

Obscura said:

Control loops or smoothing splits would help. Though, I do it the other way around. I set up the smoothing first based on how hard the angle changes are, how does it shade, and put uv splits here and there based on the smoothing.

Have you seen this thread?
http://polycount.com/discussion/107196/youre-making-me-hard-making-sense-of-hard-edges-uvs-normal-maps-and-vertex-counts/p1

Hey Obscura! What do you mean by control loops? and do you mean with smoothing splits that i split my uvws ?
Do you mean control loops like this? because when i try adding loop with slice tool on lowpoly the shading fades away and looks good: https://i.gyazo.com/c11cff30eb4bb2ddb8762b829e3a7cb0.mp4

Obscura

Yes, thats a control loop / support loop. 
By smoothing split, I mean you put an another smoothing group/hard edge where there is a strong angle change. Like there are those rectangles on the handle, and I see its smoothed together with the rest of the handle. Then on the normal map you can see that area has some strong gradients. If you would split the smoothing on those pieces at the corners, the strong gradients would stop happening. I know, you would end up with a lot of uv islands then (since you would need to split the uvs too then). Please reah through that thread, it has a lot of good info about normal map baking with/without smoothing groups and all that stuff.

Zaikoni

Obscura said:

Yes, thats a control loop / support loop. 
By smoothing split, I mean you put an another smoothing group/hard edge where there is a strong angle change. Like there are those rectangles on the handle, and I see its smoothed together with the rest of the handle. Then on the normal map you can see that area has some strong gradients. If you would split the smoothing on those pieces at the corners, the strong gradients would stop happening. I know, you would end up with a lot of uv islands then (since you would need to split the uvs too then). Please reah through that thread, it has a lot of good info about normal map baking with/without smoothing groups and all that stuff.

Thx for the help i'll give it a read very soon I will try ur suggested fixes and show results soon.

Zaikoni

Obscura said:

Yes, thats a control loop / support loop. 
By smoothing split, I mean you put an another smoothing group/hard edge where there is a strong angle change. Like there are those rectangles on the handle, and I see its smoothed together with the rest of the handle. Then on the normal map you can see that area has some strong gradients. If you would split the smoothing on those pieces at the corners, the strong gradients would stop happening. I know, you would end up with a lot of uv islands then (since you would need to split the uvs too then). Please reah through that thread, it has a lot of good info about normal map baking with/without smoothing groups and all that stuff.

Btw on the handle part u mean splitting it like this?

Obscura

There too, but not only there. anywhere, where you have close to 90 degrees angle changes. Yes, that would be a lot of uv islands. Depending on where you want to render it at the end, you should do some tests in your engine, and see how it acts with those gradients. If you render in a synced environment, you may not need to put all those splits, but after compression, the gradients can make things worse. The thread that I linked talks about all this stuff.

If you dont mind slightly higher polycount at the end, you can solve this by putting those supportloops here and there, and it will get rid of the gradients across the big flat surfaces, and then you don't really need to put that many splits.

Zaikoni

Obscura said:

There too, but not only there. anywhere, where you have close to 90 degrees angle changes. Yes, that would be a lot of uv islands. Depending on where you want to render it at the end, you should do some tests in your engine, and see how it acts with those gradients. If you render in a synced environment, you may not need to put all those splits, but after compression, the gradients can make things worse. The thread that I linked talks about all this stuff.

If you dont mind slightly higher polycount at the end, you can solve this by putting those supportloops here and there, and it will get rid of the gradients across the big flat surfaces, and then you don't really need to put that many splits.

Thx for the fast response! I imported the model in unreal engine and here is how it looks like :



In my opinion it doesn't seem to look that bad ? What is your opinion though. and taking consideration that texture would probably hide it little bit.

Obscura

Yeah, not bad. See the darker, "X" shaped shades between the big rectangles on the handle?  Thats happening because you baked without triangulating the mesh before baking, and now its triangulated differently , than how it was, during the baking. So triangulate before baking. Unreal Engine has support for Mikk tangent space now, so you are in a lucky case because this whole smoothing thing is a bit easier and you usually don't need to split everywhere at the hard angle changes. So then, I'd say youd last unwrap/smoothing with the rectangles split from the handle should be good enough.

Zaikoni

Obscura said:

Yeah, not bad. See the darker, "X" shaped shades between the big rectangles on the handle?  Thats happening because you baked without triangulating the mesh before baking, and now its triangulated differently , than how it was, during the baking. So triangulate before baking. Unreal Engine has support for Mikk tangent space now, so you are in a lucky case because this whole smoothing thing is a bit easier and you usually don't need to split everywhere at the hard angle changes. So then, I'd say youd last unwrap/smoothing with the rectangles split from the handle should be good enough.

you mean the latest unwrap picture i sent i should proceed doing that split the rectangles from the base handle? and by triangulating do you mean by when i export my lowpoly i check triangulate option on it when i'am about to bake them in substance?

Obscura

Yes.

Zaikoni

Obscura said:

Yes.

Allright got it! The thing is that i watched this https://www.youtube.com/watch?v=Z9wgKy-F1Rw tutorial from chamferzone and he didn't mention anything about checking triangulating option so i never did it. but if i do re-bake and export my model with triangulation option on does that mean it would probably fix the the ( X)  looking shading problems? or atleast make them look better so i would not have to split uvws? Sorry about if i ask too much in detail things i'am really new to hardsurface still learning.

Obscura

Well thats 2 different things. The x shaped shading errors appears because of the flipped triangulation that happened between softwares. And triangulating upon export should definitely get rid of those errors. And you split the uvs to not get those gradient lines along the topology lines on your normalmap. Keep that latest unwrap, its good enough, now you eliminated the most of the gradients on the flat surfaces. 

I see you still haven't read through that thread so I suggest it the last time. It will clear out all of these confusions. Good luck.

Zaikoni

Obscura said:

Well thats 2 different things. The x shaped shading errors appears because of the flipped triangulation that happened between softwares. And triangulating upon export should definitely get rid of those errors. And you split the uvs to not get those gradient lines along the topology lines on your normalmap. Keep that latest unwrap, its good enough, now you eliminated the most of the gradients on the flat surfaces. 

I see you still haven't read through that thread so I suggest it the last time. It will clear out all of these confusions. Good luck.

So i did some improvements do the model including uvw splits and these are the results :

I will optimize the uvw space later i just wanted to test out the bake. The uvw splits in some places in the knive fixed every shading problem i had.