Home Technical Talk

Texel density

Lorenzo444
polycounter lvl 5
Offline / Send Message
Lorenzo444 polycounter lvl 5
Hi guys i am working on my showreel. In the first part i will show a desk with some objects on the top of it. I thinking on use two different texel density. One for the objects (close up) and the other for the desk, but i have a problem to understand if is better to have an high resolution texture like 4096 px and a texel density of 10.24 px/cm for example or an 2048 px texture with high texel density? and also what do you think about have two different texel density ?   I don't know if they are stupid questions but i stuck in this process because i want that everything is ok. Thank you

Replies

  • zachagreg
    Offline / Send Message
    zachagreg ngon master
    If it's just a desk and some props don't even worry about texel density yet. Texel density is to make things look homogeneous not stand out due to a concentration of resolution in one given area. Just make them look good and have the same texel density on all things. Also you're confusing what texel density is it seems. As a bench mark most 3rd person games will use a TD of 5.12 px/cm. If the thing is going to be right up in your face like FPS guns 10.24 px/cm is more sufficient. For movies and VFX this goes even higher.

    Check this guide about TD by Leonardo.

  • Lorenzo444
    Offline / Send Message
    Lorenzo444 polycounter lvl 5
    zachagreg said:
    If it's just a desk and some props don't even worry about texel density yet. Texel density is to make things look homogeneous not stand out due to a concentration of resolution in one given area. Just make them look good and have the same texel density on all things. Also you're confusing what texel density is it seems. As a bench mark most 3rd person games will use a TD of 5.12 px/cm. If the thing is going to be right up in your face like FPS guns 10.24 px/cm is more sufficient. For movies and VFX this goes even higher.

    Check this guide about TD by Leonardo.


    ok but the risk on doing the same texel density for every objects in my case is that i will have a perfect texture for the desk and fuzzy textures for others object because  if i take the texel density of the desk for example and i use it for the other props , the uv shell of these became very small and if i do the reverse, i take the TD of props and i use it for the desk , the uv shell of this one became very bigger.
  • Alex_J
    Offline / Send Message
    Alex_J grand marshal polycounter
    It's very easy to get stuck in a trap of overthinking things, which I think a lot of beginners do when it comes to this stuff. Because there is all this math.

    But really, it is simple stuff. The best way to learn, at least for me, is visually. Take all the numbers and math out of it, and just export your scene to game engine or however you are rendering it. Walk through the environment and see what jumps out at you. Some thing is too fuzzy, what to do? Well, it just needs a little more space for it's UV's right? You don't need to copy paste the precise texel density from it's sister props -- just scale it until it looks best. Maybe that means you got to shuffle your UV's around. Maybe, if you absolutely cannot keep that new texel density and fit all your things into the same UV sheet, then you know you need to split up your materials.

    In other words, you can start by just eyeballing it. The more you do it, the better you'll get at judging how much texel density to give to this or that, and how much materials you need.

    It is easier to start with everything separated and don't worry about materials or texel density at all. Then, once you've got the art figured out, start combining these textures as efficiently as you can. By iteration, you will find out the absolute lowest resolution you can get and still look good. And this way, you are able to productively move forward in your work, rather than sitting there tying brain in a knot for too long without getting art done.

  • zachagreg
    Offline / Send Message
    zachagreg ngon master
    Alex is talking about what I was saying. The technicalities of the art will bog you down and hamstring you like no tomorrow. If it looks right it is. A desk and a cup can have the same texel density. This doesn't mean they have the same map size. This also gets more involved based upon which workflow you're using. I'm assuming everything is going to be uniquely unwrapped, baked, and textured. So if the desk is a 2k texture and the lamp is 1/4 the general size of the desk then the lamp gets a 1k or 1024 texture.

    Again the technicality of this gets messy and is always dependent upon how the end product looks. If it looks right and runs right then it is for a portfolio piece. Because now you have the same TD but what if your camera is right next to the lamp clearly you want to up the TD of it then, but then now your TD goal is off again. A desk and an object on the desk are both playable objects they are not background objects.
  • Lorenzo444
    Offline / Send Message
    Lorenzo444 polycounter lvl 5
    zachagreg said:
    Alex is talking about what I was saying. The technicalities of the art will bog you down and hamstring you like no tomorrow. If it looks right it is. A desk and a cup can have the same texel density. This doesn't mean they have the same map size. This also gets more involved based upon which workflow you're using. I'm assuming everything is going to be uniquely unwrapped, baked, and textured. So if the desk is a 2k texture and the lamp is 1/4 the general size of the desk then the lamp gets a 1k or 1024 texture.

    Again the technicality of this gets messy and is always dependent upon how the end product looks. If it looks right and runs right then it is for a portfolio piece. Because now you have the same TD but what if your camera is right next to the lamp clearly you want to up the TD of it then, but then now your TD goal is off again. A desk and an object on the desk are both playable objects they are not background objects.

    It's very easy to get stuck in a trap of overthinking things, which I think a lot of beginners do when it comes to this stuff. Because there is all this math.

    But really, it is simple stuff. The best way to learn, at least for me, is visually. Take all the numbers and math out of it, and just export your scene to game engine or however you are rendering it. Walk through the environment and see what jumps out at you. Some thing is too fuzzy, what to do? Well, it just needs a little more space for it's UV's right? You don't need to copy paste the precise texel density from it's sister props -- just scale it until it looks best. Maybe that means you got to shuffle your UV's around. Maybe, if you absolutely cannot keep that new texel density and fit all your things into the same UV sheet, then you know you need to split up your materials.

    In other words, you can start by just eyeballing it. The more you do it, the better you'll get at judging how much texel density to give to this or that, and how much materials you need.

    It is easier to start with everything separated and don't worry about materials or texel density at all. Then, once you've got the art figured out, start combining these textures as efficiently as you can. By iteration, you will find out the absolute lowest resolution you can get and still look good. And this way, you are able to productively move forward in your work, rather than sitting there tying brain in a knot for too long without getting art done.


    Ty both of you for your help
  • Brandon.LaFrance
    Offline / Send Message
    Brandon.LaFrance polycount sponsor
    Not directly related to your question, but I think worth mentioning. Unless you're in a field that explicitly requires video to properly showcase your work (animation, rigging, VFX...), a showreel is probably doing you more harm than good. Stick to well-composed still images for your portfolio.
  • gnoop
    Offline / Send Message
    gnoop sublime tool
    There is absolutely zero point to waste  texture space  for a  texel density dogma .    Decades ago with so limited vRAM that time it wouldn't be even a question.  Today more vRAM allows to have this  extra  "advantage" of equal texel size.

    But is it   really an advantage ?    Human brain perceive details not equally. Your sight usually focus on eyes and lips  if you look at a face  and edges and silhouettes of outside world .     Contrast details in general.
        
    If you have a plain flat table having  zero contrast details  and some cup with tiny golden ring   it's pretty natural to give that ring  a bit more texels.    
      
    Just don't do it in excess   and where you really need it.      Texel density  variation could be quite a tool if  clever used.

Sign In or Register to comment.