3
$\begingroup$

Recently, I am trying to create 3D models for my own game project. However, when I do some researches on game models, I come up with the term - game-ready models and I could not find any clear definition online. Could anyone tell me about what exactly is game-ready models and how to know if my 3D models (created by Blender mainly) is suitable for games.

From what I know by far, quad topology is preferred in Blender. However, for most of the game engine, triangle topology is preferred (Please correct me if I am wrong). Does it mean I have to manually change the models in Blender from quad topology to triangle topology before importing to the game engine?

Thanks for any helps.

$\endgroup$
  • 1
    $\begingroup$ Usually "game-ready" will refer to the texturing and the tri count of the asset. Having a heavily subdivided mesh with no topology work, with very high res materials and possibly no animation preparation would not be considered "game-ready". A lower poly, triangulated model with optimised materials would be a game ready asset as it has optimum performance in a game engine. In game asset pipeline with programs like substance painter, triangulation is required for texturing, use the triangulate modifier, it will quickly and easily convert to tris. $\endgroup$ – NBoss Feb 14 '18 at 10:04
  • $\begingroup$ @NickBosse Thank you very much for the answer. There is one question I would like to ask from you answer. You said heavily subdivided mesh would not be a game-ready models, does it mean I should avoid using subdivide modifier? $\endgroup$ – Jack Ng Feb 14 '18 at 10:16
  • 1
    $\begingroup$ @NickBosse never ever should you triangulate a 3D model in Blender for a game. The engines will do it automatically at appropriate phase when it's needed. But n-gons you should avoid although mostly they are exported auto-triangulated anyway. Substance painter deals with quads as well. $\endgroup$ – kheetor Feb 14 '18 at 11:40
  • 1
    $\begingroup$ There isn't a definition and the term is becoming ever broader as less experienced people start using it. "Game-read" models sold in places like Unity asset store/TurboSquid are often horrible when it comes to polycount and UVs, all that matters is a nice preview picture that gets the asset sold. But alas, most buyers are also not 3D-modelers (otherwise they'd be making their own models and not buying, right?) and they can't tell the good from the bad and just take the hit in performance. $\endgroup$ – kheetor Feb 14 '18 at 11:55
  • 2
    $\begingroup$ Game-ready = you can use it in game as is (without further manipulation). This strongly depends on the game (and game engine). In Blender it typically means, there are no modifiers and the materials are 100% compatible. (You also need to check armatures and animations - which often belongs to the model.) $\endgroup$ – Monster Feb 14 '18 at 13:16
3
$\begingroup$

Game-ready is a very vague definition. But across the industry it usually means being very aware of the performance impact of polygons and textures and use them as efficiently as possible:

Geometry:

  • Mesh is all triangles or quads(which can be easily converted to triangles), depending on engine. No ngons for sure. Avoid high valence vertex.
  • Smart use of triangle budget to ensure optimal performance.
  • A lot can be done with shaders and textures these days, so the geometry is likely going to be quite simple.

Textures:

  • Maximize use of UV space, wasted texture space is wasted memory.
  • Textures in power of 2 (1024x1024, 4096x4096)
  • Use texture size that's appropriate for the size of the object on the screen.
  • Usually a single object has a single material and single textureset(1 albedo, 1 normal, 1 roughtness map for a typical PBR workflow). This ensures that the object can be drawn with one drawcall for optimal performance.

Workflow:

  • Note most of these requirements only apply to the end result, as you are creating your models, you can use multiple textures, multiple modifiers, even ngons temporarily, until you are ready to export the final model.

  • Most game models benefits from having a detailed normal map, which uses a low-poly mesh combined with a normal map to represent additional surface bumps. So often you'd have to model the geometry twice, once as a high poly model, and once more as a low poly model, in order to create a baked normal map.

$\endgroup$
3
$\begingroup$

Game-ready = you can use it in game as is (without further manipulation).

This strongly depends on the game (and game engine) as well as the purpose of the model.

In Blender it typically means, there are no modifiers and the materials are 100% compatible.

On characters you often want a skeleton (armature) and animations. For example you can directly use the character (with armature and skin mesh) within the BGE. It still has some restrictions. Not all options that work in the Blender Editor will work with the BGE.

Also important is the purpose within a game. Is it a character with closeups or an asset that usually will not appear larger then a few pixels? How many of it will there be? How well does it fit the scene?

Yes, performance aspects have to be considered. How long will it take to animate, simulate and render the model in game? Remember there is a small time frame to do everything (not just the one model).

Finally what is "game-ready" for one is not necessarily "game-ready" to another.

Most-likely you do not have to change much to adopt it to your game. There are still many models that needs to be changed that much, that you can create your own one.

$\endgroup$
  • $\begingroup$ Thanks for the answer. However, you mentioned that there should be no modifiers which make me a little bit of surprised. If I cannot even use subdivide modifier/ mirror modifier, the speed of building the model would be slower and less efficient. Is there any significant problem caused by using modifier on the models and export them to game engine? $\endgroup$ – Jack Ng Feb 14 '18 at 17:02
  • 4
    $\begingroup$ 'No modifiers' just means the modifiers needs to be applied before the final export process. Usually, it's okay to leave a mirror or subsurf modifier in place since the export script can deal with applying it. $\endgroup$ – Mike Pan Feb 14 '18 at 20:55
  • $\begingroup$ The problem with the modifiers is that they can cause trouble. For example the mirror modifier did not work when used on a linked group. In-game the model was just the one half. To make it "game-ready" the modifier needed to be applied before using it in that situation. As less additional work you have as more "game-ready" the asset will become. $\endgroup$ – Monster Feb 20 '18 at 9:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.