Home Adobe Substance

Substance Painter - Persistent Baking Issues

Ginsplosion
polycounter lvl 2
Offline / Send Message
Pinned
Ginsplosion polycounter lvl 2
Hi there,

Just to preface this I've done substantial searching for the following issue and tried a whole multitude of potential adjustments, tweaks and fixes to no avail. As I can't attach my models here is the link: 

https://drive.google.com/drive/folders/1YHnZfX_DDLqnMM77z0s9pXW_BZDEchLn?usp=sharing

I will say that my high and low are constantly changing whilst I experiment at my end but the two uploaded here are reflective of the results shown below.

So generally I learnt how to bake in Maya but spending 10-30 minutes per bake just to have to adjust things wasn't working and I've heard nothing but good things about Painters baking capabilities.

So here is my bake result in Painter (all default settings). This is using the attached low and high FBX (by all means give it a spin yourself so you can see the issue up close and personal). If you can't spot it in the first image I've put another in to show it's the corners before they turn in by the pillars.





So, I've read a lot of various baking technical based forums for similar issues and as far as I'm aware I've covered all my bases:

  • My low has smooth edges across all of it's areas that the high intends to bake its fine bevels from.
  • Maya doesn't have smoothing groups per-say it just uses high/low definition of edges (its Mayas equivalent).
  • If you check my seam placement on my low you'll notice the seam isn't on either of these problematic corners.
  • At first I had my seams on these corners as the area behind the cylinder I shrank significantly due to lack of details/hidden areas and thought it could be related. Now you'll notice I moved the seams in an edge to test this; same problem.
  • The edge is now sewn onto the larger sections and thus shares their resolution - same error.
  • I thought maybe the cylinder is causing some strange bleeding (cages etc) - took it off of both the high and low and had the same issue.
  • Both models are placed exactly atop one another.
  • I generally always soften the edges on my low as I've read that's half the cause but still get the above errors.

Unfortunately this is literally stopping me from continuing with this crate, done all my prep for this in an earlier Painter test for this asset but I'm not gonna go push the details before I can get this error fixed for obvious reasons. I must have sent my FBX for both the high and low back to Painter about 30 times now with adjustments in hope to identify the issue.

Another example of a similar issue across various other hard surface bevels, the majority tends to be a really good bake with some horrendous errors scattered around:



I have no idea what's going wrong as far as I can see everything is in order from the uvs, the general prep/export - no errors on said meshes etc. It's most likely something I'm just unaware of but it's kinda off putting at the moment.

I'm pretty adept in various 3D software and modelling but have never really had much focus on baking throughout my studies, hell even Substance is relatively new to me unfortunately but I'm in the middle of learning 30 things whilst teaching on the side and I'd really appreciate any help here. I appreciate there's probably something fundamentally wrong with what I'm doing I just cannot understand what it is.

Regards,

Wayne

Replies

  • Ginsplosion
    Options
    Offline / Send Message
    Ginsplosion polycounter lvl 2

    Additional info:



    Shell A/B and their respect UV shells,  now initially B was tiny because it was practically hidden and wasn't going to have much of a detail metal filler. Because of this I ended up at the same conclusion that the lack of resolution joining that edge would be crippling the quality on one side and thus causing the bleeding.

    However I sewed that edge (incase I'm sounding confusing that's the red line) and moved that seam a face or 2 in (behind the cylinder) and made sure the adjacent face was now the same res as face A.

    This gave me the same error leading me to believe I still had reasonable resolution on my UV layout to bake the bevelled edge even on the lower shell.

    ---

    I have had some progress by ensuring every uv shell border was a hard edge leaving everything that wasn't defined as soft:


    Now this trades off one worser evil for a lesser. This causes the hard dotted-esque line but does somewhat form the bevel.

    I guess I'm on the right path now but I feel like I'm wasting a ton of time using trial and error/best guess approaches with what I know as opposed to nailing it with the correct baking workflow due to lack of better knowledge.

  • poopipe
    Options
    Offline / Send Message
    poopipe grand marshal polycounter
    You'll need to make sure uv splits have hard edges and that texel density is consistent. 
    Also arrange it so that UVs are straight (as much as possible)  and that you have a decent amount of padding (8px @2048 as a minimum) 
    If you're being really thorough you should pixel snap your UVs as well. 

      In a perfect world you would  just average normals and the fully synced baking pipeline would take care of it  but the sad fact is it just doesn't work like that in real life... 
    That said, even if it did  you'd be screwed when it came to making LODS. 
Sign In or Register to comment.