Any tips on efficient weapon modeling with no ngons?

Just looking to speed up my work flow using addons or just techniques I’ve never really used. Trying to learn to work more non-destructively.

I’m not experienced in modeling weapons particularly but any otherwise valid modeling technique should apply ?

1 Like

Yeah certainly. I just don’t like to say “hard surface” because I think people tend to assume I mean robots or futuristic looking default cubes. I have hardops and boxcutter and they seem to be powerful tools but I haven’t quite gotten the hang of them. I’m looking not just to make something that looks good but also to keep everything really clean and professional as I’m trying to put together an impressive modeling portfolio that shows that I can offer clean topology and easy to work with, organized, blend files.

1 Like

I see. I usually discourage people from using these addons for several reasons and I’ll say the same to you : they rely heavily on booleans and hiding away the topology from the user. Booleans as you may know generate ngons and that’ll only work on perfectly flat surfaces, so for most actual objects you might want to model they won’t help you much. If you’re going for the game industry you’ll also want to generate low-res versions of your guns and for that you simply need the topology knowledge. The way to go is identify tension lines, define them using as little polygons as possible, connect these as cleanly as possible, then only focus on giving the whole more definition, define curvature, pinching, and so on…

1 Like

Cool, thanks for the info!

That really is the problem, these are general statements on what is needed, but the trick is, how exactly you do this with a hard surface model using “pure quads” without ending up with a total mess of polygons to cater for the various details.
Personally speaking, I don’t think the Hops/BC “hides” the geo from me at all, far from it. I’m not sure why not using addons gives you any advantage, bar slowing your workflow and productivity down. If I’m making a character, then I’d want quads all the way, if I am producing something hard surface, which, by it’s nature, tends to a have a lot of flat surface areas, I’d want the tool to get the job done most efficiently.

1 Like

That’s what I mean, as soon as you need a surface that’s not flat, this entire workflow breaks down. Take a look at the models produced with hardops, they’re all the same sci-fi box with rounded edges and a few insets/greeble here and there. They’re not actually models made from reference on which control over edge flow is essential, they’re just hmmm… I don’t know, I guess freeform modeling ? without any constraints it’s easy to make anything look good, but here as I understand it @ImmortalZombie wants to model actual firearms from reference ?

Indeed that is the question. Any time that’s possible, use different objects. For instance a handle grip can be made as a separate object, and as such will not need to be topologically connected to other pieces, which would require way more definition that the part in itself needs to look good/correct. One foundational idea in poly modeling is loops. Keep a count of your loops, divert them to connect to other parts so they’re “reused” and don’t end up adding unnecessary definition to parts of your model that are already dense enough.

1 Like

Whilst I agree 95% of the examplers shown in the Hops videos are “free-form”, I personally do a lot of hard surface work, like things for Arch-Viz, household objects, and Hops/BC are an absolutely essential part of my workflow for that. Before using them, doing the object was, at times, frustrating, in the number of keypresses needed to get a job done.
They speed that sort of work up tremenderously. Again, yes, organic work, you need quads, otherwise chaos ensues, but as with many things, I think it’s a mistake to write off something because the examples shown are not really “real-world” items. When I look around me at the items in my room, I can see in my minds eye how I would approach them using the addons and how i can, at times, literally, cut corners, to get the best result is the most efficient way. Again, yes, all this can be done in vanilla Blender and even all with quads, but then you can be clogging up resources by having everything quads when it would be much simpler to have some carefully crafted NGons in there. (Again, organic modelling excluded).
What I would say however is, people need to understand NGons as well, it’s not enough for it to just be an “NGon”, you need to still think about how it terminates and connects to surrounding geometry. Especially when considering bevels and how they like to “resolve”, not all NGons are created equal. :wink:
(Actually, I touch on this in my video on the subdivide tool at the end, how a “bad” ngon can give undesired bevel issues where a “good” one resolves nicely).


Can’t figure out how to quote multiple people on mobile…

Anyway, I’ve got nothing against ngons but because I intend on possibly selling these models I can’t use ngons, as I want them to be able to export to other formats cleanly and my understanding is that some (or most?) programs don’t support ngons. I personally see hardops or box cutter or whatever as a useful tool for cutting in details such as screw holes but I haven’t gotten the hang of using those addons or booleans so I probably shouldn’t talk too much about them as what I’m saying may be nonsense lol

Now, right now I’m modeling an 1860 Henry rifle which is a pretty simple model to make but I’m also modeling the “guts” of the gun and focusing on clean beveled edges, etc. What I struggle with is for example, let’s say I block in the receiver of the gun, then I cut a hole into the curved bottom of it for the lever and trigger, I get that hole nice and pretty looking using edge loops and bevels, then later on I have to adjust the size of the hole while maintaining the smooth curved surface, but I can’t without like an hour of fiddling with the edges and sliding vertices. While I know how to do this and even maybe do it well, it’s just so inefficient. This I believe is where boxcutter/booleans shine but if I were to go that route, I’m stuck with shitty topology and possibly just as much work to correct that topology when it’s all said and done?

Definitely seems to be the case, when I try to import my old stuff from LW, Ngons and curves (splines in lw), get “lost”. As a result, even for hard surface, I tend to cut my ngns into quads and tris, but as I mentioned above, more with a view to logical cuts tham amything else. I hate how auto-tri of ngons just makes an utter mess most of the time, so try to control it manually.

Oh they do so much more than that, but I think that is where everyone starts to use them. I am also a big advocate of MeshMachine, the refuse/unfuse/unf*ck tools are a godsent when working with exiting bevelled geometry.

The addons don’t do anything you can’t do yourself. Basically for booleans they rely on your geometry to be already subdivided because once you cut a hole in it, there’s no subdividing it again (or you’ll see it break down before your eyes). So you can do the same yourself, model that receiver, subdivide it then add a substractive boolean on top. It’s all fine but you can’t work with low-res meshes, basically it works a bit as in zbrush when using liveboolean : it’s not a workflow that relies on subdivision.
On the other hand you can use any addon that you like for blocking out the shapes, then retopo the whole thing as a final step. I often work with several disconnected objects first, then join them up once I’m sure they’re in their final position.

Good to know! I found this interesting post about topology, seems to explain how to use ngons pretty well.

Hm, I’ll have to look into MeshMachine I know of it but don’t really know anything about it. Kind of an in-between of Hops and Looptools?

Good info. I’ve modeled the low-res mesh which sub divides quit well. Need to figure out how to bake it cleanly as I keep having issues with the normals, everything’s unwrapped cleanly but I think I need to use a cage mesh or something, not too sure, haven’t dug into it enough. It’s been awhile since I’ve baked anything. Basically I’m doing your “typical” blender > substance > blender workflow, low poly mesh, export fbx, high poly mesh, export fbx.

Not really, it’s a whole different toolset, I’d recommend jsut watching the videos on YouTube by Machin3 on it. I’m not sure of any way natively to do some of the things it does.

1 Like

Very cool! I’ll check it out.

I think it would be good information to remember a few basic modeling tips.

First, parts of the process are not the end result. The pocess includes things like Booleans and addons that use them etc. Anything that lets you develop the initial shape I consider part of the “process”.

In the first part of any of the modeling process you are not usually concerned with your polygon flow or keeping things clean. Unless you want to be tweaking forever just to bang out your shape.

As you get more repetition with your work you start to learn various tricks to plan ahead. For example modeling two shapes you plan to use as Booleans so that certain edges line up and will be smoothly merged as a result when you join or cut with Boolean.

All complex modeling starts out with a more or less “scupting” phase where you are more concerned about creating the contours and complex joined edges that have seemingly impossible curves that result by two dissimilar objects merged. And then also eventually having that edge beveled.

The final phase of the process would be retopo if you are infact sculpting with millions of polys or, if you have by brute force jointed parts together, you could commit (apply Booleans or addons) and then clean up the result.

In my opinion you can’t really do any good modeling of complex shapes without considering these two phases.

If you are using a process that allows for non-destructive workflow, keep it non destructive as long as possible before committing. Same would be true with sculpting.

That is the process regards to modeling.

The second main point is a decision as to what parts of the model will be baked details with maps and which parts will be geometry.

So regarding selling models or delivering them down the pipeline with a team, try not to allow the final result to determine what tools you use to get there. It would be almost like saying, “I don’t sculpt because no one can ever use millions of polygons…”

Another process worth considering is actually modeling with subdivision surfaces and quads. Get nice smooth joining shapes and contours (which you can also initially join with boolen and then clean up) and the addition of nice bevels.

Then back up or save that mesh and apply the modifier to the mesh, then go back and delete redundant edge loops to reduce polygon count. (don’t do this if you only plan to render in Blender of course but only for meshes you export). This can save a lot of tweaking in the long run and give real nice results.

You can also combine any of these techniques in any combination.

Hope that helps.


This is probably the most efficient games oriented non destructive weapon tutorial, yet authored over the past 6yrs, essentially realising the power of working with live boolean operands / operations.

3ds Max/Zbrush: Proboolean + Dynamesh hardsurface workflow tutorial