A basic modelling question

I’m not quite sure how to put this, but let me try:

Imagine you model a hard surface object, say a spaceship/car/house/whatever. It consists of many very edgy parts, organically roundparts etc … just like we for instance know spaceships from any star wars movie and other places.

Let’s just imagine a sidepanel (simple plane) of which you have a slanted cylindrical structure coming out, on which you may have some weapons or antennae etc. The whole thing is immobile, just a simple fixed structure.

How do you model the “cylinder on plane”? Do you carefully create a single object/mesh, cylinder and plane connected properly etc. Or do people just make a plane, then a second object, the cylinder, and then just stuff the cylinder into the plane, totally not caring much about exact position etc.

I’m having trouble understanding how the pros model the ships for games etc. That is, if they just “sloppily” stick together various objects to get an external shape looking pleasing, or if they carefully model a single mesh. I also only talk about rendering … if the insides of the ship matter etc, you obviously can’t just stick together whatever, as the things prodtrude to the inside.

I know game designers etc have to think about rigging or the limitations of the game engines etc to break up their whole ship into parts, which is different from jsut modelling a static object to render.

standard method is just stuff the cylinder into the plane. If you have you make it one piece, you can use the retopo / surface snapping tools. You can also use booleans, but I don’t recommend it. The results from booleans usually have a lot of bad topology.


this is one of my older projects, but this might work for you? use the same metod.

i recomend using “I” on faces, and “E” click, + scale down. that is the metod i used to get the proper shape.

i added more verts than nessasary so i could use sub survace to get a smooth look. :stuck_out_tongue:

Don’t generalise what I’m about to say to every game asset or model, but as far as I know:

For gaming they usually stick it in as it reduces the poly count (cylinder faces+one plane face instead of cylinder faces+plane faces made to link cylinder). Second reason is most of the time details don’t matter as players don’t get close views of the cylinder/plane junction.
If the vehicle has an inside, then they clamp the cylinder to the plane to hide the protruding bit that would go inside the vehicle. Another solution is to make an inner surface to the vehicle and stick the cylinder in between the outer and inner surface of the vehicle.
This is mostly true for hard surfaces (vehicles, buildings…) because pieces move without deforming. When the joint area has to deform, you usually need to join both surfaces with more care.

For stills, high-quality renders or cg animations, it really depends. As long the difference between a clean or smudged joining isn’t noticeable (object is far, no close-ups…), artists don’t bother in making a clean assembly.
On animations, people won’t pay much attention to these things if they’re not the center of attention as the image, characters, props, move a lot, can be blurred, and our focus is guided to the area of action rather than details.

It’s all a question of wasting time on useless details or/and gaining performance really.
You’re not going to insanely detail a spaceship that’s hidden behind a bigger spaceship, moves so fast that it’s blurred and appears as a blotch on the screen, and is 10 miles away, taking 10x20 pixels on a 1080p image…

for a game engine, the mesh really needs to be manifold, meaning all merged into one continuous mesh. this is due to the way real time shaders work, manifold meshes process much quicker than intersected and non-manifold “holy” meshes. however this depends on just how much performance overhead is needed at a particular point in a game to determine how much effort to put into making stuff manifold as opposed to just stuffing it together. talk to your programmers to determine this.

for animation only, as long as its a hard surface that doesnt need to deform, and you dont need anything special at the joint like a clean chamfer, stuff it together! again, a manifold mesh will process slightly quicker at render time, it usually doesn’t amount to much more than a few milliseconds per frame. so not worth the extra time to model manifold.

if your 3d printing or CNC machining… everything has to be mainfold and perfect. no shortcuts! lol