Home Technical Talk

Architectural Modeling and Bevel

garrettia
polycounter lvl 2
Offline / Send Message
garrettia polycounter lvl 2

hello to all of you. Some of the architectural models that I came across on Turbosquid did not use bevels, I was wondering while I was researching, I came across this article on turbosquid Holding/Support Edges


Holding/support edges present to retain shape after subdivision. This also applies to 3D text.


NOTE: For Architecture No holding edges, chamfers, micro-bevels, etc. required.


I was wondering if the criteria are different in architectural models, and that's why they didn't use bevel, I would appreciate it if you could indicate the reason.

And in what situations should we use bevel and in what situations should we not?




Replies

  • gnoop
    Offline / Send Message
    gnoop sublime tool

    My guess is that subdiv modelling is just a niche . For example I never had to rely on subdiv modelling through my whole experience in gamedev . Because extra shading issues it often creates and redundant geometry to workaround them. For Architect modelling it would rather be corners decals than subdivs imo.

  • kanga
    Offline / Send Message
    kanga quad damage

    I would be interested on an answer to this as well. When creating structures for unreal I bevel every edge that needs it because I am allergic to the paper cut no bevel modelling you typically saw with sketchup for years. No subdivision on these pieces. I use Blender and use extra edges to get the smoothing group 3d max result. I am only doing these models for myself at the moment so would be interested to hear what the approach is in commercial practice.

  • gnoop
    Offline / Send Message
    gnoop sublime tool

    Tiny bevels when they gets occupying less than 1 screen pixel causes some kind of troubles for rasterization part of a renderer . I am not sure how this important nowadays although.

    As I see people use either macro unique normal map with rounding ( troubles with texel size consistency and macro normal map blending accuracy), special decals on corners , some geometry placed to hide paper clip corners and combination of all of this.

    Sometimes you can just put a kind of tiling wooden plank from a trim sheet over building corners , So basically a bevel , just not that tiny . But it would have to stay through most of lods too.

    I usually do rounding in normal maps over hard edges where unique unwrap or trim sheets allows this or just let it be as "paper clip" because too much troubles to workaround this

  • garrettia
    Offline / Send Message
    garrettia polycounter lvl 2

    This subject is very confusing to me. The reason for this is why turbosquid says so. This is a modeling method, a modeling technique, why do they use it? Do they know anything? If they know anything, what is it?

  • Eric Chadwick

    Totally depends on intended usage.

    If for a game, then all kinds of optimizations are often needed.

    For the kind of architectural visualization rendering-only work we do, everything is beveled. 3ds Max has some great tools for automating this lately.

    Turbosquid architectural models are likely conversions of architectural plans, and as such they're not optimized for either workflow.

  • poopipe
    Offline / Send Message
    poopipe grand marshal polycounter

    To address gnoop's point about thin triangles.

    These are still bad, avoid them where possible.

    But

    Bevel a thing if it needs it. You can tell if a thing needs a bevel by looking at it from the viewpoint of the camera/player/whatever and if your brain says 'that thing needs a bevel' then you need a bevel.



    It's good to consider these things but it's not worth compromising the appearance of an asset (in the context of view distances etc) before you've been told it's a problem and honestly, when the tech art and render teams start worrying about a couple of thousand extra tris on screen you can be pretty confident that 90% of the project's performance problems have been solved

Sign In or Register to comment.