The Making of a Game character
Written by: Anna Larsson
My character was made for a capture the flag demo, with a free camera and only nonpayer characters. We wanted to make an innovative demo so I didn’t want to make a cliché female fighter, at the same time I wanted her to be interesting and match the surrounding elements. This document will describe roughly how I did and how it went.
I chose to make the concept an extremely small part of my project, it is a very important stage and normally I wouldn’t rush it but we were under time pressure and I knew that I had to
choose what I wanted to focus on. Ultimately I thought it would be better to centre my attention and effort on the modeling and texturing part, I’m really happy with my decision.
Still, in a perfect world I would’ve liked to spend more time coming up with a more refined
concept. There’s not much more to say about this stage of the character making, I just drew fast and it’s not very thorough.
A central part of modelling is to get the topology right, I heard it’s an agonizing experience
trying to animate a model that doesn’t have a proper edge flow, so I took some time before I
even opened Maya to study different characters and what kind of topology that works well.
I think it’s also important to bear in mind
before you start, what parts of your model should have more detail then others. Is the face going to be animated? Will the character ever be viewed up close? My character has no facial animation and although she can be viewed up close I have a very harsh polygon restriction and thus she will look angular if you look too close.
I used cylinders as a starting point for the torso, legs and arms, then I used a front and a side view of a woman to get the shape and
proportions right, which is important. I wouldn’t want to give Magnus, the animator of our project, a character that i.e. had a kneecap that was placed too high or low on the leg.
Because of my polygon limit, the detail on my model had to suffer a little. But I was trying to make up for it by sculpting a very high poly model and then apply it as a normal map, sadly you can’t do anything about the contour, the body will still look angular if you look close.
The texturing was the hardest part for me, I didn’t want to use photos however drawing your
own textures and making them look believable takes a lot of skill. I’m pretty happy with the
result but I think it could’ve been better, I guess practise makes perfect. To make it easier for
myself I baked an ambient occlusion
map from my high resolution mesh, I
can really recommend that. It makes it
easy to see where all the details are
suppose to be and also you can apply it
as a layer over the texture when you’re
done, which made my textures look
better. I used two 2048 pixels texture
map, I though more would be a waste
since you mostly saw the characters
from a distance and I really felt like
more detailed was uncalled for. I
should add that I also made a specular
map to make the metal look shiny and
the cloths less so.
What went right:
I delivered what was expected, and I did it on time. I learned more about the do's and dont's of making a character and experienced what it's like to work in a team, we've worked in groups before but it's a bit of a new experience every time. I’m happy with all the important decisions
that I made in the project.
What could've been better:
I think my textures could have been much better, it’s not that I didn’t do my best, it’s just that
it takes skill and experience to create stunning textures and I need to keep practising on that.
I could’ve had a better dialogue with Magnus, the animator. Due to some unforeseen trouble with the Engine documentation and his animations we didn’t have time to perfect certain
I could probably have made the model more high quality, we thought at first Nebula engine wouldn't be able to handle it so I made it as low-res as possible but later on we discovered we probably could've used more polygons.
The concept could’ve been better.
If I would do it again:
I'd make a model with a higher resolution, I think 8000-9000 triangles would be appropriate. If I could do it all over again I’d also like to try an other engine, just to see the difference. If I
could do it all over again, I think I would’ve made some changes in the face topology.