Visual Analysis of N64-era Zelda Games #1 (Models)

This post is to explore what the models and assets of the N64 era games look like and to discuss how to achieve this.

Straightaway the most obvious comparison to do is Link the protagonist.

On the left is Link from the Nintendo 64, he is made up of 696 polygons. The right is Link from the GameCube, this one is made up of nearly 3000 polygons, a 4 times increase in geometry.

The same can be seen for most 3D assets between the games. The most obvious being the main characters as they have more polygons reserved.
Other objects such as foliage and mundane items all have as few polygons as possible. This is because at the time hardware restrictions were a lot more important, the N64 just was not powerful enough to have soo many detailed models on screen.

To achieve the same quality will require a mix of two methods, one is a quicker automated way but will more than likely produce some unfavourable results. The second would be to manually edit original Windwaker models to reduce their poly counts to appropriate counts.

Method 1:

The first method is a simple modification that can be applied to objects in MAX to optimise them, it reduces the amount of faces in an attempt to optimise them. But does have unfavourable results, especially for detailed models.

As you can see the polygon count is severely reduced. If you look closely at the lines around his torso, they're not even, other polygons too are now weirdly shaped. The end result in game would have undesirable shading, considering this model in its current state has less polygons that the N64 model.
Another drawback to this level of optimisation is the size of the polygons. Characters need to have enough polygons or small enough polygons to make animations look more realistic. Because of the new reduction in polygons in the above model, there is now a lot of space around Link's waist, which means if any animation shows him bending then these faces will not bend because they're too long. Instead they will bend at the end points, making him look very rigid.

Method 2:

The other way of reducing polygon counts is to do it manually. It's much more tedious but by doing so allows me to recreate and move lines in a more natural way.

If you compare this render to the previous one the difference in edges is quite clear, it looks clean and is symmetrical in most places. It even has enough polygons around his waist (as compared to method 1) to allow him to have more fluid animations.
This one was edited by hand by myself and took a few hours, simply because this is one of the most important models so extra care was taken. The overall polygon count is far less than its original 2874 on the GameCube and does now more closely follow the graphical style of the Nintendo 64. the model could be reduced further, which I may do, but for now he is much better.

I think using both methods for converting game assets will be beneficial. The optimise especially for testing, anything else such as important models I can do by hand.