A very general question

When modelling an object with lots of different parts.

Is it better to, have one mesh with lots of edge loops, and all the parts extruded out of it.


to have separate simple meshes, with less edge loops, that are not joined, but are grouped by say, joining the meshes or parenting or something.


you want to create a door. you want to keep the door simple, but if you want to add a handle and letter box, then its going to complicate the mesh by adding more loops, but the advantage is, its all one object.

your other option is, model a simple door, leave holes for letter box and handle, then create separate objects, which will have a higher poly count,then insert them in the holes. a rectangular hole for the letter box, in which slots the rectangular letter box.

Option 1, I imagine being better for simulations, like the game engine, I imagine if somethings not connected, then it would fall off?

Option 2, would be an cleaner, more suited to still images. I dont know if a group object would animate or be usable in physics sims.

so, is this something you are conscious of when modelling? Is it better overall, to subdivide everything from one cube, or to have seperate grouped/parented objects?

I prefer the second options definetely specialy for character, doors, windows.

My general preference is to model as realistically as possible … if objects are separate in reality, I like to keep them separate in ‘virtual reality’.

A doorknob, though ‘connected’ to the door, is a separate object. I find that it makes other possible steps a bit easier to manage - unwrapping & texturing, animating, etc. The difference also tends to show in the shading/render, depending on the levels of detail and resolutions.

I’m not quite sure how this affects GE stuff, but simulations in general can be done with both methods (not sure how they might affect performance, though).

In the end, I say “Do what works for you in any particular project.”

Having a lot of separate objects joined into a single mesh can be a problem as it sometimes makes it hard for Blender to understand which way to orient normals (i.e. know what is inside and outside).

Just my five pennies.

but in which way joined? which is the correct way to group objects together if you dont want to have them as one mesh, but want them to “stick together” in the physics engine.

You press the handle against the door and press ctrl- j to join them into 1 mesh
Then press Center New in object mode and then enter into edit mode and press ctrl- n to recalculate normals

Yeah, you can join two objects without actually connecting any of their vertices … so no need to add loops (unnecessary mesh detail) and avoids unrealistic shading where they meet.

I prefer the second option too, for all the reasons already said, but allow me to add one more. If you ever intend to animate the objects, having them separate and parented (if needed) is the only (or the best) way to do it.

If you’d join the Knob/ Handle mesh to the Door mesh you’d get only one mesh, and I guess the only chance you’d have there wouldn’t be a much practical one: to make an armature to move the Knob/ Handle, which, I guess, would deform your poligons when you’d rotate the knob or handle…

Thanks for the answers. Im still curious as to what method of joining one should use for the physics engine. I wouldnt want to model seperate parts then press P and they all fall apart.

Parent them.

Select the object you want to “attach” to another (i.e Door Knob) then Shift+select the Parent object (Door), press Space -> Object -> Parent -> Make Parent, and then when the door rotates or falls down or whatever, the knob goes along with it, without leaving it’s place on the door.

Remember, the last object selected is the one that will control (hold) all the others. The others will be attached to that one. So if you select the knob, then the mail box, then the peek hole, etc, and after all those are selected you select the door, all of them will be able to be moved freely on their own, but will move along with the door if it’s moved.
(I’m just not sure if you can parent multiple objects at once, but I think you can)

Although I’m not the best person to talk about that stuff right now I think that’s the (best?) way. I’m still catching up from a year of Blender abstinence and can’t remember many stuff.

oh okay, I always forget which way around to select them >.< pressing P whilst you have them selected also works. but thanks for the clarification.

opps I meant CNTRL-P

indeed. I didn’t remember that anymore. Thx :smiley: