Home Technical Talk

Question regarding Texel Density.

polycounter lvl 11
Offline / Send Message
Linfo polycounter lvl 11
Hey there lovely people!
Here's my question:
Let's say that the project I'm working on asked me to maintain the Texel Density at 1024 per meter (so a 1x1 meter tile has a texture resolution of 1024).
So far, so good.

The problem I'm having is managing UV space when the object I'm working on is either too large or too small to fit properly within this scale.

In the case of small objects, you can of course use a smaller texture size. But sometimes that will not work, because of the way texture sizes scale.
And in the case of large objects that cannot use tileable materials, I get a similar problem.
I will end up with either too much free UV space or I won't be able to fit all of my UV islands in the grid.

Ideally I would want to use all of the available UV space in small objects, and I don't want to be forced to break up larger objects into multiple material IDs because it becomes unmanageable really quickly. 

So, what's the best way to deal with these cases?

Replies

  • throttlekitty
    Options
    Offline / Send Message
    For starters, if you can make things fit by scaling lesser-seen UVs down, then do so if the impact is small.

    For large objects, breaking them up by material first is probably your best option, like if you're able to think of portions of the large stuff as being a collection of small things. Reusing uv space via stacking for repeatable elements is an option, or even changing the model to make better use of UV space. Without knowing the specifics here I can't really answer that for you, these things are always situational.
  • Linfo
    Options
    Offline / Send Message
    Linfo polycounter lvl 11
    Thanks for the quick reply!

    Here's a special case I can mention:

    In a previous project I worked on I was tasked with making a set of very large pillars, which had a ton of specific details carved into them.
    This meant I couldn't use tileable materials. 

    The problem started when we applied the texel density to their UVs. Some pillars wouldn't fit, others would end up with a ton of empty UV space, it was a mess.

    My solution to this was to be honest with you, pretty terrible, hence my question here.
    What I did was to unwrap all of the pillars together, to ensure they would have the same density, and then I started to manually arrange the UV islands into groups and assign them to a material ID.

    In the end, I ended up with about 8 material IDs scattered around about 5 large scale objects. And the way the IDs was assigned to them was very random, which make painting these a pain.

    In retrospective, I'm thinking that the general idea of what I was doing wasn't so bad, but perhaps I could have spent more time arranging the UVs so that the mat ID distribution made more sense.

    I'm just wondering if there's a better way to go about it.  

  • throttlekitty
    Options
    Offline / Send Message
    Maybe I'm misunderstanding what you mean by Material ID here? To me, that's a baked map representing materials for quick masking during the texturing phase. It sounds basically like you went the UDIM route for these pillars, making everything fit the best you could while making it relatively painless to work on. Having just done a similar thing with Substance Painter but for offline renders, I didn't have to worry about VRAM overhead from having a gaggle of maps, so I understand the pain of needing to switch sets while working.

    If it's not gonna fit, it's not gonna fit. :D Other routes you could have gone is having a single "plain" pillar with the carvings as overlaid geometry like decals, not super fun to manage all the edge blending, but it could work. Or have your working uvs set up so you can work easily, then bake them down to tightly packed game-ready uvs where you have all the pillars on as few textures as possible.

    Or redesign them so you can make better use of texture space, like having cloned tops/bottoms/backsides, but unique centers.
  • Linfo
    Options
    Offline / Send Message
    Linfo polycounter lvl 11
    Yeah all good points.

    In this case all pillars had individual details, so cloning wasn't an option.

    On the other hand, the decal idea is pretty good. That could have worked for this.

    Ok! I think get it now, I appreciate your help with this! :)
Sign In or Register to comment.