Baking in Substance: Bizarre Smoothing Problems

PattyWhacker

I'm reworking a project since I wasn't satisfied with some of the parts of my Scorpio model (please shoot me) and I'm finally at a point where I'm baking within Substance Painter. Today I decided to work on my model's arm. For some reason when I bake within Substance Painter, I get a result which looks like a smoothed version of the high-resolution model in Maya. The forearm in particular strikes me odd since when I baked in the past, I did not have this projection issue. I'm currently using version 2.4 of Substance Painter, and I have no idea if that's the reason behind this issue or not.

Bakes within Substance:





Non-smoothed model in Maya:

Smoothed model within Maya:

Topology of high-res model:


Here are the parameters when I'm baking in Substance in case this helps:



The example in Maya makes me believe that Substance Painter uploads a smoothed version of the high resolution model. I'm wondering if there's any way around this problem. I'm not sure if I have to re-decimated these pieces in Zbrush and make sure the model triangulates well, or if there's a way to set up parameters within Substance Painter to prevent these issues. Any feedback is greatly appreciated!

Froyok

What does the wireframe of the low-poly looks like in Maya ?

PattyWhacker

Here is the wire frame of the low poly model.

I've redone the high poly model, baked the model again, and the bake turned out fine. It had everything to do with the wire frame of the high-poly model. It's just strange that this is occurring now, because when I baked the armor pieces 8 months ago, my normal maps looked a lot better and I had very little issue.

Froyok

If you increase the polycount of the high-poly mesh, does the bake perform better ?

PattyWhacker

I fixed the problem. I re-decimated the high-poly model. The problem the first time around was that I needed extra triangles to hold the high poly model shape. It took a few hours, but I fixed the issue. I think I could try adding more topology on the high-poly mesh, that sounds like a better solution.