Wednesday, April 14, 2004

animator by name, animator by nature?
As regular readers will know, I am married to a chap I refer to as 'the animator'; however, since neither of us are in full-time employment right now you may be wondering just how much animating the animator actually does. This is where it gets complicated...
The animator wants a job working for a computer games company, specifically he is looking for work as a 3D modeller, concept artist and/or animator. That's a very broad set of skills and it is going to take him a long time to get proficient in all of them. To achieve this ambition he is constantly working on improving his showreel and adding new content to his website. It is quite natural to think that 3D animation, since it is done with computers, is easier and less laborious than traditional 2D animation. Ha bloody ha, let me tell you there is much labour that is 'hidden' in 3D animation*.
Step 1. Build your model. This can be done by using either 'nurbs' or 'polygons'. I like to think of nurbs a bit like modelling with virtual clay, while polygons are more like building with lego - the animator thinks that this is rubbish, but the analogy is a useful one for us outsiders. The animator works mainly in Maya (but, as I understand it, most software packages work in a similar way) and prefers to use polygons over nurbs. I'm not a 3D modeller and it is tricky to explain even with the software on the screen, but I'll try to convey the compexity of the task. To begin with, the animator designs a character on paper. Once the look is set he draws two scale-accurate versions on squared paper - one front on and one in profile. These then get scanned in to be used as guides. He imports these pictures into Maya and places them in a cross + thus. Now the fun starts! Using polygons, he painstakingly begins to build his model using the drawings as an outline: each polygon gets extruded from an edge until he gets a hollow head, body or whatever he's modelling. There can be many thousands of polygons in a model. It takes a long time to build even the most basic model, with anything remotely sophisticated taking much, much longer: Sil-9 (see Soluus), for example, took around three weeks to complete.
Step 2. Blend shapes. If you intend using your model for lip-synch you will need to set up a series of 'blend shapes'. Blend shapes are pre-set facial expressions. A very basic set of blend shapes will require at least 14 copies of the head (Gollum in LoTR may well have used over 100). With each head, the animator carefully deforms the geometry of the model to create an expression (angry, happy, scared etc) or mouth shape (oh, ee, woo etc). Each blend shape takes around 10-15 minutes to set up - a full set will take the best part of a day.
Step 3. Rig your model. Once you have built your model, to make to move you have to give it a skeleton; this process is called rigging. If you are going to animate your model successfully, all the joints need to be placed in exactly the right position - otherwise the model's movement will look unnatural, no matter how skilled the animation is. Also, the axes of each joint need to be correctly aligned - with the x-axis follwing the direction of the joint and the others placed acording to the three-finger rule (a bit like we were taught in science lessons: the thumb points up, the index finger out and the middle finger down). Models can need a lot of joints, Sil-9 has somewhere in the region of 80. In most games companies, models are rigged by a someone who specialises in rigging - and, by extension, knows a good deal about anatomy (although sometimes rigging can be counter-intuitive, with extra joints needed in the lower arm, for example). Once the skeleton is complete the skin of the model (its surface) is bound to the rig, either rigid (a given point is weighted to one joint only) or smooth (allows the weighting of a given point to be applied across multiple joints). The animator then pre-programs sets of movements - so that if the wrist moves the rest of the arm follows, for instance.
Step 4. Weighting. This is where things get really awkward. Each joint has a certain sphere of influence and - to make the model appear to move naturally - it is necessary to assign influence to each joint. Weight maps tell the computer how much influence each joint has. The animator selects a joint and 'paints' weight on to the surrounding 'flesh' this method is more of an art than a science and, with much of the body affected by multiple joints, can be an arduous process of trial and error. Hair and fabric move according to particle dynamics and, yes, these too need to be programmed in.
Step 5. Texture mapping. Right then, if all has gone according to plan, we now have a lovely well-modelled and rigged character, but it looks crap because it is made out of a matt grey substance. We need to create colour, texture and surface detail. To do this, essentially, the computer takes the 3D geometry and creates a 2D flat surface - this sounds so simple, but in practice it is anything but. There are examples of finished texture maps on Soluus, but it takes a long time to get to this stage. The texture map begins life as a series of complicated and difficult to identify fragments - imagine what a box looks like flat, now try to imagine that multiplied by 10,000 non-uniform planes, or take my word for it, it is complicated. The animator stitches these fragments together to create a recognisable patchwork skin. There are several ways to colour this skin in - the animator uses photoshop. There are other things to do with the way light reflects and is absorbed by surfaces that also take hours to set up - I can't even begin to explain how these work.
Step 6. Animate. At last! After weeks of work the animator can finally begin to actually animate. The animator uses the key frame animation method. There are 25 frames per second (European standard) and a basic 1.5-second walk cycle has approximately 18 key frames. Unlike hand-drawn animation, the animator works his way through the cycle moving, for instance, just the legs then goes back to the beginning to move the arms, then the breasts, then the head...etc. It is only at this point that 3D animation begins to have advantages over traditional hand-drawn animation - the animator can instantly check his progress by 'playblasting'; whereas, with trad animation, it is only once the drawings have been 'line-tested' that any problems can be seen.
Step 7. Lighting If the lights aren't on no one can see your animation! There are whole books devoted to the subject of lighting CG environments - suffice to say it takes a very long time to set up anything more than the most rudimentary lighting. The best lighting enhances the realistic feel of the environment and helps us to suspend disbelief.
Once you have watched someone go through this incredibly laborious process to create a second or two of animation, Finding Nemo or any LoTR scenes featuring Gollum are suddenly transformed into masterpieces beyond compare.
The animator is currently working on a new character, based on the basic body of Sil-9 with considerable refinements, for lip-synch, walk cycles and what he calls a 'cut scene' (those bits in computer games where you get played an animated bit of story before going up a level or on to a new senario). So when the new stuff is finally finished - probably in another week or so - I hope you'll join me in being mighty impressed by how much hard work is involved.

rant alert
*If the animator wanted to he could buy in ready-made models (although I think he'd still need to rig them) or use software like Poser to help him out - I don't really need to tell you that this is frowned upon do I? One of the most frustrating things for both of us is that having spent over £5,000 on putting him through an intensive six-month course at Central St Martins he seemingly still has a very long way to go before he is of a good enough standard to find a full-time job in the industry. He's not the only one either, despite the course literature quoting a 90 per cent employment rate on completion, the majority of his classmates are also struggling to find work. Also, St Martins only teaches animation - all the other stuff (the modelling, rigging etc) he's had to teach himself and, while it would be great to be able to concentrate purely on animation, a showreel featuring models made out of primitives isn't going to impress anyone, no matter how beautiful the animation.

No comments: