Problems with baking high to low poly between marmoset 4 and substance painter
Hi, I’ve made an optimized model from a high one using the least ammount
of triangles possible and using UDIMS. Before exporting it I’ve
triangulated it and exported as a fbx
The baking were made under Marmoset Toolbag 4, flipping the normals in the Y axis.
By the way, the mesh and UV mapping it’s the following, the retopology it’s handmade:
I'm using blender and exporting as FBX
The baking were made under Marmoset Toolbag 4, flipping the normals in the Y axis.
Some fixes were needed to be made, after fixing It looks with not artifacts at all
After that I’ve exported the texture set with correspondent names, when
I’m checking if everything is fine, I’ve stumbled with artifacts that
are not seen in marmoset.
By the way, the mesh and UV mapping it’s the following, the retopology it’s handmade:
I'm using blender and exporting as FBX
Replies
http://wiki.polycount.com/wiki/Normal_Map_Modeling
You could bake in Painter see if it comes out any different, if so compare maps.
Is the lowpoly shaded all smooth? If so, you could harden edges at steep angles that have UV splits anyways, to reduce the lowpolys shading gradients (as gnoop wrote).
It looks like you have not split smoothing groups where you have UV splits on hard edges like trigger area causing shading artifacts.
Oh my God I've should learned that before starting
I've flipped the Y Channel for DirectX.
With "harden edges" are you talking about Mark Sharp?
So, I should mark seam along hard edges, like here:
http://wiki.polycount.com/wiki/Normal_Map_Modeling#Smoothing_Groups_.26_Hard_Edges
You could also use Face Weighted Normals, where the model has bevels.
http://wiki.polycount.com/wiki/Face_weighted_normals
Generally you want to avoid shading errors in the lowpoly model, because these often cause extreme gradients in a baked normal map, which are bad, see: http://wiki.polycount.com/wiki/Normal_Map_Modeling#Smoothing_Groups_.26_Hard_Edges:~:text=When a mesh,to counteract it.
By the way, I'm using one material per UDIM tile because Marmoset can't accept them
- Knowing how to model-for-UVs, how to model-for-baking, and how to UV-for-baking are skills not aquired overnight. IMHO you're biting off more than you can chew with this asset in its current state. Perhaps just focus on one part until you get it just right.
- Sharing your files is necessary for people to help you out. But you'll also need to share your high, otherwise no test bakes can be done.
- You have a lot of work still ahead of you modeling-wise as the asset is just not looking clean at the moment with poor shading, uneven polygon distribution and improper UVs. Try to find an equivalent asset from a game with similar specs for reference - Fortnite weapons are probably a good fit.
Whereas on yours the shading issues are immediately apparent :
Of course knowing how to do things cleanly may seem subjective when you're not used to it yet. So to get there you'll just have to share clean screeshots of your progress for poeple to point things out, little by litte. Baking comes later.
Here's the high model btw
Here is what you have at the moment (high, low wireframe, low shaded) :
This valve for instance just look way too tiny and easy to break. Same for the connection of the blade to what looks like ... a pressurized gas tank ? Of course these are comments on the design itself rather than the technicalities of baking but things like that pile up. Had this been based on a real item these inconsistencies wouldn't have been as much of an issue. In a way the visual design itself has consequences all the way down to the baking process.
In production a highpoly like this would receive an extensive paintover pass indicating where to thicken things and what to adjust.
On the low you are shooting yourself in the foot with all these interpenetrating pieces serving no purpose. This will lead not only to an unnecessary increased workload but more importantly to discontinuities in the textures, causing issues for baking but also for texturing/weathering. And overall there are many inconsistencies with how various edges are treated.
Personally I think focusing on just 3 parts and getting them right could be a good point of entry : building 3 clean and continuous lows, 3 well-crafted corresponding highs, all of that gathered on a solid UVs sheet treated as if it was a real asset.
Sure enough this dark area on the dagger shouldn't be there :