Modeling From Images

I am new(ish) to Blender, and I’m not so good at modeling, especially from background images. I have laid out the basic shape for a robot that I want to create, and its relatively simple, or so it seemed… I am just wondering if there are any tips/ tricks that could help me out with things such as keeping my model all quads and just basic clean up? Also, is there a way to get better at modeling in general? I’ve heard the “practice practice practice” answer many times, and I’m sure it would work, if I could find something to practice on? I can’t really get any models that I want so I have no idea of what to practice on… Sorry for being so newbish and all and I am very grateful for all answers!

Best would be if you post your reference image … that way people could give you specific hints and show you patterns they see.

I am kinda new myself and i know from personal experience that robots are never simple^^

Regarding tris and quads and oh yeah ngons … well you can’t say just use quads … it depends on your project.
Models that have to look organic and/or are animated -> use quads and avoid tris and n-gons
Architecture and other static stuff -> use whatever gives you the best result … tri quad ngons everything is fair game here
(at least that’s my rule of thumbs)

For your modelling skills … well what people say is true … practice is the only way.
BUT i would strongly recomend using as many tutorials at first as possible.
Simply because in the modelling process so many shapes can be done the easy or the hard way (with the hard one in most cases ending in a horrible looking result^^)
So i think it’s better to learn as much as possible from people with years of experience.

A good example are the tutorials on cgcookie on making a barrel or a soccer ball.
Both seem quite simple at first but once you get started and don’t know the tricks you find yourself in quite a mess.

Thanks, I’ve been using some self-made concept images or sketches ( whatever the kids call them these days) and I’m a little self conscious… One technique that I found particularly helpful was, after many minutes of brainstorming (but probably is already in use), using photoshop/gimp or any program that you can draw in, and just trace the image with lines and boxes, so I could find where to extrude, scale, rotate, etc. and that worked really well (sorry for answering my own question, I just wanted to help any other newbies). Another thing that I wanted to mention, is that if you’re use the Blender 2.64 test build, maybe this is just on Mac, but there is an issue with multires… you have to subsurf first before multires or otherwise your project becomes a ball of vertices and faces, and that’s why I thought I was having issues with quads. Thanks again for the feedback, I think the tutorials will really help! Another question, I wanted a cylinder placed on my model, but I’m not sure how to connect it? Or is it ok just to poke it through the top and just pretend like its attached?

Modeling is problem solving. When you’re new to it, it’s also slow because most steps require searching an answer, making mistakes, and topology is just a strange word. It feels hard overall. With practice, you’ll solve problems that hold you back the most and the exercises become more complex; “When you’ve modeled a human ear, the eye is easy” -Williamson.

If you can get away with that, not a big deal. Objects that are far away in the scene or otherwise hard to see can be left that way. But for models in general, connected parts needs to be connected in the model. If it has clear separate pieces, those are often easier to model as separate and you don’t have to spend time making it look like separate while it’s not.

You didn’t provide a picture/.blend to see the topology so it’s impossible to suggest anything specific, but usually some preparation is needed before you can merge (alt+m) or bridge topology together, reducing or adding geometry to make them match.
In a case of cylinder, it’s common to arrange some (6/8) vertices in your mesh first so that you can make a hole, then round the vertices which you then extrude out. Subdivision surface modifier adds to that, rounding it and completing the cylinder shape.

Oh and please, use paragraphs. If answers are what you want, those help you get one step closer to them.

Ok, thanks for all the information, I still feel a little lost, but I hope that feeling goes away the more I work on my model. I’ll work on some pictures or .blend so that I can better represent my problems in the future.

As for my current issues, there seems to be no end… Here are the images of the wrist of my robot right now and what I want the final shape to be.

I was wondering, how do I prevent too many weird shapes from being created after using the subsurf modifier on my mesh? I had to start over because I had too many issues with the n-gons and subsurf, it turned out all gunky.

Avoid n-gons and triangles when using subsurf modifiers … just as a rule of thumbs (altough triangles are sometimes realy difficould to avoid)

I usually start modelling with the subsurface modifier disabled or at lvl1 and switch inbetween to see if everything goes as planed.
Altough this method only works if you A. model the mesh bigger than it has to be in the end (subsurface kinda shrinks it a bit)
or you want hard to semi hard edges … with which i mean you want to have a 2nd edge loop close to all outlining edges.
That way it keeps the “defining” edges of your mesh stay the same and the mesh only has a more round and smoother surface once you increase the sub modifier.

Do you want to make a blocky robot or should he have rounded edges and stuff like deformed shield plates and so on?
Because if you just want a blocky robot you don’t have to set the model to smooth and can avoid most of the stress you would have with a subdivision surface modifier.

Another thing … dpeending on the level of details you want it might be a good idea to do the legs and arms as seperate meshes and concentrate on the “smaller” pieces. Modelling all at once usually gets messy (at least when i do it^^)

And you will most likly need some more detailed references … they don’t have to be all handmade … you can for example look for parts that would look nice in your final robot and use those images to build the part. (see above … spliting the model into smaller pieces and therefore smaller problems)

Oh and for the sub modifier … in case you need a single edge to be sharp but cant or don’t wnat to add another edge loop … you can always select the vertices and increase their grease value (hit n and on the very top you find what i am talking about)

Another neat way to do that is to add an edge split modifier but deselect edge angle and select sharp edges … that way you can define which edges you want sharp and which not. (ctrl+e & mark sharp)

I hope that helps you a bit … and keep us posted on your progress :slight_smile:

Alright, thanks a bunch! I was wondering about how to do my arms earlier… so I began to do them separately and now I am confused on how to add them back on. Like is there a method to attaching or is it just a “by eye” deal where I just fill in any geometry? And sorry for the enormous amount of questions, but should fine details be made separately, extruded out, or just sculpted?

Also, I wanted to make my robot mode “smooth”/ human like. I think I’m going for a steampunk type robot, and I think the materials still need a little tweaking… Here’s a screenshot of my restart on the model since the last one was totally messed up. Tell me what you think. And thanks again for all the help I deeply appreciate it.


Depending on situation, you might leave it as separate piece and perhaps parent it to main object (object mode, ctrl+p). Or you can join it (object mode, select both, ctrl+j) and leave it as floating geometry, or you would connect the vertices and edges.
Here I have a default cube and I made a handle for it. First one is separate and the handle is parented to the cube. Second set is joined geometry but those are not connected, just left there as is.
[ATTACH=CONFIG]190845[/ATTACH]

Sometimes you have to connect pieces to get the desired look or make it behave like you want, no matter what. And you normally should aim for uniform and connected mesh. It takes some preparation work to make connecting pieces match. I made another example.
[ATTACH=CONFIG]190846[/ATTACH]

Thanks for the speedy and immensely helpful replies!

Depending on situation, you might leave it as separate piece and perhaps parent it to main object (object mode, ctrl+p). Or you can join it (object mode, select both, ctrl+j) and leave it as floating geometry, or you would connect the vertices and edges.
If I wanted to make my mesh/model usable for an animation e.g. rigging, would parenting objects be the easiest way? Or would I have to connect torso and arm directly? I feel a bit lazy for asking, but since I’m not entirely confident in my ability to prepare and connect the arm and torso, I’m just looking for another way…

Again, sorry for the constant onslaught of questions, but I was wondering, is there way to select which parts of my mesh are affected by a subsurf modifier, so that I can preserve finer details?

Thanks Again!

Go to CG Cookie and just start doing the tutorials man… they will get you on the right path!

There are a few other good ones out there, but they have everything set up in an extremely comprehensive manner.

Ok. thanks!