0
$\begingroup$

I've seen these models on sketchfab that I find good enough for a mobile game, but if you look at them in wireframe mode it looks like they are made of a succession of cubes (or very close shapes).

I'm far from an expert at modeling, but isn't it a very bad practice?

In blender, what's the difference between creating multiple objects and a single object composed of multiple meshes?

How will it impact rigging / animation?

If I would import them into a game engine (unity for example), can it make a difference (in performance/usability/shading) ?

In short, are these models bad?

Thanks for your help.

$\endgroup$
  • 1
    $\begingroup$ in the linked model, I'm not sure what you are seeing. Is you question just "what's the difference between creating multiple objects and a single object composed of multiple meshes?" $\endgroup$ – David Sep 4 '17 at 12:22
  • $\begingroup$ Yes, and how it will impact performance (rigging/animation has already been answered). Will each mesh be rendered separately (I'm particularly interested in how unity will handle it) ? If so it really looks like a bad practice and slow. Again, I know this single character will not impact much, but imagine this kind of practice for dozens of assets of varying complexity, it may begin to be a consideration. $\endgroup$ – Stnaire Sep 4 '17 at 12:40
  • $\begingroup$ How did you view them in wireframe mode? Is there a .blend somewhere, or is there a feature of Sketchfab I'm unaware of? Anyway, without seeing the wireframe all I can say is it looks like a rigid and low-poly art style. Nothing wrong with it from what I can see. $\endgroup$ – Mentalist Sep 4 '17 at 13:23
  • $\begingroup$ You can see the wireframe by clicking on the gear icon, then rendering, wireframe. $\endgroup$ – Stnaire Sep 4 '17 at 15:38
2
$\begingroup$

The BGE needs more time to process multiple objects. Therefore it is advised to join objects that form a larger part of the landscape (e.g. a forest).

On a character the requirements of the character handling will dominate.

From artistic perspective an object that deforms (skin) is better to handle as a single object and animated via armature.

If you have objects with rigid parts like a robot or a car where nothing deforms you might use multiple objects. Nevertheless the armature method is still applicable.

You might even need a mix of both. E.g. the deformable mesh for presentation and the rigid objects for collision detection (check if the hand touched anything).

This strongly depends on your needs and your focus.

Other requirements might be:

  • breakable (can be done with both, but different objects are easier to handle).
  • lots of instances (here you need to look at the performance).

Edit:

When you want to change items and clothes in-game you better have them separated. If that is not the requirement they can be part of the skin mesh.

$\endgroup$
  • $\begingroup$ An object in blender can contain multiple meshes, so if this character is only 1 object containing multiple separated meshes, will it make 1 draw call for each mesh or will the model be rendered in 1 call like with a single mesh? You say in you response that it will take more time, so I guess there will be multiple draws. I know for a model like this it will not have much impact, but isn't it a bad practice? I don't need to break it or create multiple instances. I'm learning and I try to avoid bad practices as much as I can. Assembling dozen of meshes to make a character feels wrong. $\endgroup$ – Stnaire Sep 4 '17 at 12:16
  • $\begingroup$ As far as I know within the BGE you have one display mesh (which gets rendered) and one physics mesh (which get considered by the physics engine). So you do not need about multiple meshes in a single game object. $\endgroup$ – Monster Sep 5 '17 at 11:03
1
$\begingroup$

These models seem be very well suited for a mobile game, they are quite detailed but relatively low poly also. For a mobile game especially these are very well suited for the environment. If this were to be used for PC game these would be much more inappropriate but due to the major restrictions this form of modelling can be quite beneficial performance wise (the texturing comes out quite well so the look of the game does not suffer either). For gaming specifically a low-poly model with better texturing is better (as it is done in said model), using a higher poly model will significantly lower the performance of the game (especially on mobile systems). This type of modelling with beveled cubes significantly reduces the mesh detail. For the usability in unity the sword and shield are quite simple, this is good as something like an inventory system would benefit from the simplicity of the models as they would not be resource intensive to spawn in (this refers to the type of modelling in general rather than the specific items in the said model). Overall, for a MOBILE game these are great models and would work well in the engine and provide an excellent balance between aesthetics and performance. For rigging it is just simpler as one cube = one bone but I will leave this open for someone else to chime in about the rigging aspect. There is my two cents worth, Happy Blending! :)

$\endgroup$
  • $\begingroup$ It is the opposite. The mesh detail have more impact on performance than texture detail. Large textures eat more memory (which squares up). Performance wise not all pixels needs to be considered at all time (in difference to vertices/faces). Nevertheless the models are low poly and low texture detail and should be very fine on low-end devices. $\endgroup$ – Monster Sep 4 '17 at 8:44
  • $\begingroup$ Are you referring to in blender or in unity or both? Despite the mesh detail/texture detail performance mixup was the answer on the right track? $\endgroup$ – NBoss Sep 4 '17 at 9:20
  • $\begingroup$ The model is very simple so I guess it will render lightning fast no matter what but I just want to know if creating an object from multiple meshes greatly differs from creating it with a single mesh. For example, to make a table, is easier to create a rectangle for the top, and 4 other rectangles for each leg instead of doing multiple edge loops to extrude the legs from the top. I also requires less geometry. But it feels very wrong to do it that way. I have the same feeling with the character. I feels wrong that each part of is kind of a cube of its own, separated from the rest. $\endgroup$ – Stnaire Sep 4 '17 at 12:20
  • $\begingroup$ For a static object like a table using the method of five different shapes would not be desirable you are correct, with something like the sword and shield they should be one mesh and only modeled from edge loops and such as you have said. The other thing to consider is what you want out of these models, a sword may be many objects if it needs to break or have a cinematic effect. Although a table would have more geometry being separate, other models may not, sometimes separating sections of the model can balance it i.e. a very low poly sword or arm is separated from a much heavier mesh. $\endgroup$ – NBoss Sep 4 '17 at 21:25

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.