I’m trying to model an object right now but I can’t figure out how I can build what I have in mind. There are always some kind of artifacts and I just can’t get it right. I hope somebody can help me with this.
These pictures show more or less how I want the shape, so they’re kind of mood images:
It’s a subdivided circle, and I pulled the center vertex a bit to the upper left corner and a little bit along the y-axis.
alright alright. But I actually want this shape to be the surface of a box, and that’s where I’m failing. Like the loudspeaker in the very first image of this post, it’s supposed to be within a rectangle. But when I try to build a rectangle around the circle or start with a rectangle and build a circle in it, it looks like this:
this is not a suggestion to How to make the sink, but it shows a really solid method. use a low polycount, witch is easy to work with. subsurface modifier to make it smooth, and supporting edges to controll the sharpness of the edges you wish to be sharper.
by the way, dont limit yourself with the density of the mesh. when i say “very low polycount” i refer to a good rule of thumb witch sais;
“Use as few polygons as posible, but enough to get all the details you want”.
also, QUADS QUADS QUADS! in case you didnt know this, when ur working with subsurface, use quads only. if it is dead unavoidable, then try to hide the trist in a flat area.
also try to keep every quad as “square” as you can, IF i is posible. this will deform the mesh better.
This is the wrong mentality, you should model something so that it looks good. Triangles are useful and with hard surface modelling I’d actually encourage their use as it makes life much easier. The only exception is that they should ideally be placed on flat or recessed spaces.
Also keeping all the quads square just adds to the confusion and makes a mesh unnecessarily difficult to plan out. It may be easy on something simple as you’ve shown but that becomes incredibly arduous when you’re working with a complex model. Make the model so it again, looks good and concentrate on cleanup at the end.
disagree…
its not hard to keep it 100% quad, but you might have to rework areas a few times to get a proper topology. also when you only use quads, you will end up with mostly loops, witch is easy to work with.
the only senario where i would use trists is for low poly models where no modifiers are being used.
Sorry but that’s just naive thinking to believe that using quads is the only way forward for high poly modelling. If you go to Polycount.com and inspect some of the work by studio professionals you’ll see that most if not all hard surface models they make with the exception of automotive work consists of tri and ngon topology. Of-course quad topology is the ideal scenario and should be done as best as possible but the beauty of hard surface modelling is that well, you have all those flat surfaces to terminate topology on. In-fact in the instance of cutting details into poly-spheres you may even find you’ll get less pinching with a corner ngon than completely quad topology.
As I said, for still rendering and game assets you should concentrate on the aesthetics of the model so that it looks good… With caricatures and animation where joints need to deform it’s more important to have quads and clean topology.
Not to hijack the thread or anything, here’s an Elcan I made recently - do you want to see the wireframe shot? I think you’ll be surprised then…
You are correct in your approach Netroxen. However, I believe that sticking to quads is good for beginners because it helps them avoid bad habits. Only when they master quads they will understand when N gons and triangles should be used I believe.
It’s like in martial arts, when you are learning there is this kick, and that kick, and that other kick, and they all have this very rigid forms, which one should pratice. However, when a master fights, a kick is just a kick, and it will be very fluid. However, this was achieved only with very rigid form at the beginning. The same happens with painters, they first learn by imitating real subjects, very rigidly, and usually then move on to more stilized works, or even abstract, effectively "forgettin. If one begins in abstract, it will most likely be just a mess on the canvas.
Klar, I understand however I found, at-least for myself that I found it eventually quite limiting always hitting walls on how I’d keep all quads. That Elcan I posted above has so many triangles in it, however you don’t notice them because I go ape shit on the smallest flat area where I can terminate unnecessary edge-loops. I think the biggest issue I see new modelers making is that they don’t properly understand how to insert edge-loops correctly. You can mask many errors on a model if you correctly apply edge-loops in the right place and add extra geometry where necessary.
I always tell new people to visit this thread as a matter of priority when starting out. Go through it, read through it and take into account other people’s mistakes. http://www.polycount.com/forum/showthread.php?t=56014
It sure is quite limiting but for beginners it makes for nice exercises anyway. Of course it may make something that could take 10 minutes to make take one hour, but at least they’re learning and it works. But yeah, they should first just follow tutorials and read those kind of materials while following along.
Thank you both for your time to answer my questions
And I think this is a quite interesting discussion about the fundamentals of modeling.
Actually I’m reading and listening to a lot of stuff about topology lately and it all sounds fine and dandy but when I have to start something on my own or recognize a problem, I don’t really know what to do. I think it’s a lot harder than it appears.
So, what I learned so far in this thread is, that I made an huge mistake in the beginning by actually modeling to much. It works better to start low poly, add the subsurf modifier immediately and let him to the whole work … just pulling the one or the other edge and make an edge loop here and there.
Yeah, but sometimes I don’t know what’s the best way to start. Don’t know if I should box model or do it edge by edge.
It doesn’t look like much and it doesn’t look cool, so I feel a bit dumb. But i did it over and over again tried to figure out how I could achieve the coolest topology.
You should copy other people’s work before going on your own. That is, follow tutorials and/or try to replicate other people’s topologies. You’ll save tons of time, and headaches.
We could all tell you how to do it exactly, or show you a finished picture, but next week you’d have some similar problem again.
Topology can be a hard subject but as simplified (without scaring you too much), it’s about understanding what forms you have to model and what kind of model structure corresponds to that. The form has to be clear before you can make decisions about topology, but that can be harder as it might seem.
The form is important. Topology issues can be hard at times but modeling the wrong thing to begin with is worse. When you know what form you’re making and have enough experience to plan the structure that conforms to that, then you know what modeling style/workflow you’re going to use before you even start.
This for example. Hopefully you have more references because the image you posted could be deceiving
The left part, which I assume is what the speaker looks from the back, might not be what you’re currently modeling. It might be just the lighting that makes it look like one corner goes further in than the other. But if the key light comes from top left like the shadows suggest, the inset could be symmetrical. Perhaps double check unless you’re already sure about the form.
If it is symmetrical, the structure could be something to this direction
Something that you need to be aware of if your work is for general distribution is that there are some applications, such as Daz Studio and Poser that make a mess of importing a model that has n-gons and give rendering artifacts where there are long, thin tri’s
This is what it looks right now: