Character Process
page 1
page 2
page 3
page 4
page 5
page 6

I am taking Modeling for Game Design this quarter and our final project is to model a character. I modeled a horse for the vehicle project, and the process page for it can be viewed here. I want to go even further in depth into my modeling process. I am doing this for two reasons. First and foremost is so that any potential employers looking through my work can get an idea behind my workflow. The second reason, is because so many people model WRONG! Not that my way is the only right way (but it is a right way), but that I found it to yield superior results to the process I have seen my classmates and online peers use. So to begin, I want to show how most people model.

Most people start with a box, and then start extruding like so...


The reasoning behind this seems sound at first-to keep everything in quads of course! That seems to be the mantra today quads quads QUADS! While quadrilaterals are indeed good, they are not the most important part of the model, because whether its an 80 poly game model, or a 80 million micropoly displaced Lord of the Rings cave troll, no one will see your wireframe. What they will see, is the topology of the shape. So people need to start emphasizing shape first, THEN get to making their faces all 4 sided. If you get too caught up in keeping quads you will lose the artistic flow of the model.

Now you ask, "How then shall I model?"

Good question. Here is A way that I found that works for me. It is not the end all, but it hasnt failed me yet. I don't beleive I am the first to use this, but I did not directly copy anyones process when I thought it up.

Start with a cylinder.

Thats right, I said a cylinder. Why a cylinder? Well because the human or horse or dragon or alien or insect or (insert anything organic here) form is CYLINDRICAL! Why in the world would you create extra work for yourself by having to tweak every face that you extrude from a model into a round shape, when you could have started with a round shape in the first place. There are two cylinders you can use, the 6 sided and the 8 sided. 8 works best, but if you are going for really low poly, 6 will work. The reason 8 works so well is that the human body is divided into regions, much like the earth. There is a left and right side, and a front and back. the 8 sided cylinder will give you a line dividing the character down the center (useful for mirroring over) and a line down either side. The remaining 4 lines give you the 3/4 lines that define the roundness of the shape.

Ok, now on to a real world example. In true cooking channel fashion, I have a model already completed to the exact point I want to illustrate. How convenient!

Place your cylinders in front of your reference images (always use reference images). Scale them to the basic length and size of what they are going to end up being. In this instance, the light blue is an arm, the green the torso, and the yellow is the leg. for now, dont mirror the torso yet.

In this step I grab each cross section (a cross section is the ring of vertices or edges traveling perpendicular to the tube, and fibers are the lines and vertices traveling parallel to the tube) and scale them to fit that part of the reference image. Do this in both the side and front orthographic viewports. the arm and torso are finished, but the leg is still only halfway done. Not only does it need more definition, but it also has a case of "tube syndrome". (tube syndrome is when the cross sections remain perpendicular and the fibers remain parellel to the original cylinder. If you notice in the arms and the torso, the cross secions have been rotated to better fit the item they are defining. In the shoulder they rotate to define the deltoid. In the torso they rotate to define the chest. Doing this can add loads of definition to your model, without having to add more poly's. On the leg, the cross sections have been rotated to define the buttox from the side, but from the front, they are still very much flat. they still need to be rotated to help define the leg.


THIS APPLIES TO HIGH POLY MODELING TOO! I cannot count the number of crappily made models I have seen here at school and on the internet, where the wireframe looks just like a checkerboard.. The lines do not follow the contour of the muscles and body at all. This results in poor deformations, as well as slower rendering times, because you have to add more and more cross sections to get the desired shape. This way will take longer, but it will yeild superior results to an object with "tube syndrome". You will have to massage and carress your model into having proper topology in it's surface, as well as its wireframe. But people who take this extra time, are the ones who have great models, instead of mediocre models.


Notice how much better the leg looks, now that the cross sections are rotated to help define the leg. It is subtle, but it makes a big difference. Once your cylinders are in their rough proportions, and you are happy with them, Split the torso down the middle and mirror it over, so that you only have to do everything once. I combined the three cylinders so that I could join them.

In this picture, I am showing how I define the stomachs on all my characters. The cross section orginally traveled where the red line lies. The yellow line shows where it has been pulled to. This allows you to define the lower edge of the rib cage, and the obliques (V shaped muscles traveling towards the groin)
Here I have further defined the abs and the obliques, as well as the lower edge of the ribcage.
Sometimes inm order to make a model flow correctly, you have to totally ignore the current flow of fibers and cross sections, and redirect them. I had to do this for the pectoral muscles. The yellow line shows where I made cuts, and then welded vertices, and drew new lines in order to accomadate this new definition.
Here I added even more lines, and really made the chest read as a chest instead of a rough shape extruded out, which is what many people try to do with their muscles. If you dont know what it looks like under the skin, if you dont know how it inserts, and where, then do not try to do a model that has defined muscles. Period. Do not attempt to make up muscles either, and claim that it is an Alien or a Monster, and it doesnt matter because the muscles are made up. You have to know correct anatomy before you can make up your own. This is where life drawing sessions are key.
In this step I started defining the buttox, and lower back muscles. Notice how I am disreguarding where the current lines lie, and will redraw, or remove existing lines in order to help make the correct flow.
In order to define a muscle, or any feature for that matter, you have have to set its outline, or valley. The yellow lines are the "valley". then you have to draw the actual definition lines, or Peaks. The red lines are the "peaks". The valley's make it so that when you bring out the peaks, and then scale them , the definition is limited to within the valley lines. Many times, people will draw peaks, but not valleys, then when they scale the peaks, there is nothign to contain their definition, so the muscles end up wonky. (wonky is an official term, coined by my great drawing 2 teacher)