Baking Problem in xNormal

anthonybp

Hi everyone,

I'm at the baking process with one of my asset for a project, but I can't get the baking to work with xNormal. I usually do not have a problem with this software.

Basically, xNormal shows me this error : 


I have a high poly and a low poly mesh using a cage. The cage have EXACTLY the same number of vertices, quads, tris and edges as the low poly model, but I still get that error.
     

I would like to know what could create this problem.

Thank you

anthonybp

I solved the problem. I just remade the cage

FourtyNights

xNormal is very picky about the vertex order between a low poly and a projection mesh (cage), since every 3D object has numbered indices for each vertex. So, it's a good idea to try to keep the vertex order the same when exporting.

frmdbl

Just forgo Xnormal, it's outdated. I used it a lot back in the day, but right with much better bakers available it's not something I'd recommend.

anthonybp

frmdbl said:

Just forgo Xnormal, it's outdated. I used it a lot back in the day, but right with much better bakers available it's not something I'd recommend.

What are you using ? Because I find that the normal map in Substance Painter aren't as good as in xNormal.

Obscura

From what viewpoint they aren't as good?

anthonybp

Obscura said:

From what viewpoint they aren't as good?

I'm just saying that based on the experience I had with it, it's not a fact. I'm still new to the baking process. Maybe I just did something wrong.

Shrike

anthonybp said:

frmdbl said:

Just forgo Xnormal, it's outdated. I used it a lot back in the day, but right with much better bakers available it's not something I'd recommend.

What are you using ? Because I find that the normal map in Substance Painter aren't as good as in xNormal.

 Substance is definitely a noticeable downgrade in quality from Xnormal, same as Marmoset or Knald is a noticeable upgrade

Obscura

Could you guys possibly post a comparison images, or something like that? Baking a normal map from a highpoly mesh with tangent space of the low poly provided is pretty much a standard approach nowadays, luckily. There is definitely a downside of using xnormal though. I believe it uses the standard algorhithm on the cpu, that is now possibly in real time on the gpu. So xnormal is significantly slower than any newer approaches.

EarthQuake

Xnormal is slow (CPU based) and generally a pain to use. You can still get good quality bakes out of it, but you'll spend more time setting them up and waiting for the bakes to render. XN is missing a lot of features and map types that you would expect from a baker these days as well. I used XN for many years, and really appreciate Santigo's efforts in making and supporting it, but it's hard to recommend it with so many better options.

One of the main benefits to Toolbag (and Knald) is that you actually see what you're doing, realtime updates, see the change as soon as you adjust something, etc. Plus Toolbag has bake groups/name matching, skew painting, and all sorts of other neat features.

Substance should be able to make good quality bakes too, but you're flying blind with it like xNormal.

sacboi

xNormal runs fine on my primary system, an entry level gaming rig but still manages to output content relevant to an occasional job or personal stuff I'm into. It's been my goto since 2009 which I still suggest to those just starting out with limited processing options, at least giving it an honest appraisal prior too upgrading towards nexgen solutions. Anyways for what it's worth the app has been a rock solid reliable tool.

frmdbl

What sort of a difference in Xnormal's favor have you guys seen? 

Edit: Here's a quick test I did, the 1st one is Xnormal,then SP and Knald, the textures are 4k and the highest subsampling available. 
It's the SP viewport, even harder to spot any difference in UE4.

Actually if you look closely the only differences are in SP's favor.

Shrike

Many other people on the substance reddit also have noticed that it does inferior bakes

I tried baking some curvature and ambient occlusion, and these came out terrible in comparison with marmoset or xNormal, the difference in detail and quality was enormous. Using Substance own curvature was so weak, it was almost not picked up by Substances Own curvature based shaders.

frmdbl

I don't know how every type of a map stacks up comparing every program, but if we're talking anecdotes, the SP baking has been working absolutely fine for me, and so do the curvature based (and otherwise) smart materials.

Visually checking out Xnormal's and SP's curvature bakes, the Xnormal one is stronger and has a sort of a gradient which I guess is cool for stylized assets, the SP's one seems more faithful to the original mesh, perhaps for a reason too.

anthonybp

frmdbl said:

What sort of a difference in Xnormal's favor have you guys seen? 

Edit: Here's a quick test I did, the 1st one is Xnormal,then SP and Knald, the textures are 4k and the highest subsampling available. 
It's the SP viewport, even harder to spot any difference in UE4.

Actually if you look closely the only differences are in SP's favor.

Then I'm doing something wrong. I will try to experiment a little more. It would speed up my workflow If I manage to get this quality