Thank you for choosing our Travel Guide! Now you own the perfect collection of stunning information about Hyrule
– the land where Zelda: The Wind Waker
is located in. Before you go out exploring, please read these safety-Nintendodructions:
I. The first edition of this guide was crafted by Warby. I just extended it a bit and we’re both proud to present you this 2nd edition!
II. If you like what you see and want to support the experience, feel free to check out the new Support Simon-page.
III. It’s dangerous to go alone! Take this:
Now then, we wish you a great time reading through this guide and discovering some of the greatest secrets crafted by Nintendo – the ancient creators of Hyrule.
The shape of our
world has been seen as a disc and (since some years) is nowadays known as a sphere. The shape of the world where Hyrule is settled, has for form of a cube. The GameCube
The GameCube console was released 13 years ago (in 2001) which makes it even more respectful what Nintendo squeezed out of its hardware. We’re talking about cell-shading, cloth simulation, depth of field effects and more. The following pages will cover some of the greatest tricks and we hope you’ll be as impressed as we are.
Normally, 3D engines sort all objects and throw away what you can’t see. It might be surprising to you, but in our real world this isn’t the case. Stuff behind a wall stays there, even if you can’t see it. And so in the Zelda game! In this world, it seems that nothing gets culled. In any standard BSP scene
nothing would be rendered behind this wall:
But as you see here, nothing gets culled. Note: Based on the changing draw call count while rotating the camera, there is view frustum culling
It seems that the GameCube was faster in just rendering everything instead of sorting what doesn’t need to be shown.
Hyrule is full of refreshing air and rich green grass bushes. You can cut down every single one of them and you might suspect that there surely must be some batching/instancing technique behind that amount of bushes. But if you look at the scene drawcall per drawcall, you’ll notice, that every bush is separately rendered.
Warbys observation brought up the question “What exactly is a draw call?” in Simons head and he then spent two month writing about rendering
Below you can find an animation where drawcall after drawcall is shown – you can see that every plant is rendered separately.
The GameCube seems to crunch through those drawcalls very well and one reason for this is, that consoles have less clutter (driver, APIs, …) between the game and the hardware and the whole rendering process benefits from the less communication between software and hardware (read more about this in the mentioned article
One remarkable characteristic of the citizens is, that their eyes and eyebrows are always rendered on top of their hair (the mouth isnt). For some people it’s hard to get used to this “Manga Style” but you better do because the Hyrulians don’t like to get stared at.
The people in Hyrule love sport. Below you see one doing his morning exercises and he fits his feet in no-time to the stairs. Isn’t this impressive? The game is more than 12 years old and even today not every game has such a IK
-System implemented (but it’s worth noticing that Zelda was not
the first game with such a technology).
Additionally, the feet don’t only get fit to the height but also to the angle of the surface! A lovely detail in my eyes.
The residents in Hyrule love fashion! Even the monsters decorate their spears with beautiful moving streak-cloth-thingies. But simulating cloth is expensive and some skeptical tourists could think it is pre-calculated.
Today we know: It’s not! If it would be pre-calculated, it would not be able to adjust itself to various ground heights. Interestingly the cloth always assumes, that the ground it hits is flat (must be some kind of approximation / optimization. You can see this in the next picture, where the white line marks the actual orientation of the floor):
Another optimization seems to be, that characters and objects are ignored by the cloth. If you have eyes like an eagle you will spot that, in this example, the cloth doesn’t collide with the wall (the video-framerate was reduced to make it more visible):
Also non-static objects like this barrel (some Inhabitants use it as a hideout and sneak around) are excluded of the cloth simulation.
Last but not least: the fashion-piggy itself seems to be made of air from the perspective of the cloth system.
But all this is really nitpicking. It looks great and while wandering around in this beautiful world, nobody rally cares about those tiny details and we rate this sense of detail with 5 Warby & Simon stars.
Science is another big topic in Hyrule. Often executed is observing the universe by looking at the stars. If those would be based on textures, you would get pretty blurred results by looking at them with a telescope. But look at this:
The stars are made out of two triangles which makes sure that the edges always stay sharp! Another interesting detail can be observed by looking at the sky – but at daylight. By looking closely into the sun you get a nice lensflare and an even nicer HDR effect.
If you’re a professional archer, you can observe the HDR effect (and a red tint) by shooting a fire-arrow and enjoying the following explosion:
A small hint from our side: Look out for a clear sky when you want to observe the sun because clouds make the effect disappear.
Speaking of clouds: some foreigners think, that there is exactly “1” perfect swirl/whirl/twirl-cloud texture. Warby worked on multiple games where people tried to imitate that perfect Nintendo cartoon smoke cloud sprite texture. But newest discoveries tell us, that there are actually a whole bunch of different ones for all sorts of occasions:
Since the world of the game consists almost only of isles, driving around with a boat is kind of a forced hobby. It’s great but also dangerous if high waves push you around. The water in the game is basically one huge non-tessellated plane. So how is it possible that there are harmful waves you might ask? Well…as soon as you leave the secure waters an additional plane fades in (only visible in wireframe mode):
You can’t see this plane in-game yet, but as soon as the world decides to turn the water on, the plane fades slowly in and presents itself with a nice water-texture.
Several points shall be noticed: The plane is tessellated and therefore “real” waves can be produced. The center of the plane is always below the boat and moves with it (if you look closely you can see where the plane ends in the distance). To give the illusion of movement, the wave-texture is scrolled over the plane depending on your direction and speed.
Without Internet, the people in Hyrule have a lot free time, which some fill with catching fireflies. While running behind a shiny firefly, you may notice, that they seem to have dynamic light sources which could be kind of a performance hit.
The light tech was one of the most discussed topics, here’s an update. Thanks very much to Alex, Amandine Coget, Guillaume and James O’Hare for bumping me and Crazy Butcher and alfalfasprossen for taking the time to investigate and think about how it was done. Here are the results:
Instead of using Deferred lighting
, they used a stencil buffer
– which is a full resolution texture, masking out (using bright and black pixels), where the light actually needs to be rendered. To create this mask, they render 3 geometry spheres per light – their presence/intersection with the world create the final mask. THe following is speculation – for the 100%-truth you have to ask Nintendo.
Only the back faces
are taken from the 1st sphere. And only its obstructed
geometry (behind the torch, wood, wall, ground) is rendered into the mask (which means that even light spheres behind walls are taken into account):
The 2nd sphere is rendered – this time the front faces are used when they are not obstructed (which means that this time hidden lights are not added to the mask). If you look separately at this draw call, it looks like this.
In the stencil buffer this mask is overlayed above the one which was rendered first. The result looks then like this. The areas where the first and this draw call overlap, get even brighter. This bright area is exactly what we want.
Last but not least, the 3rd sphere is added to the scene. And within its range, the darker values are clamped and only the brighter values stay in the buffer. This means, that hidden lights are clamped to black since they were not brightened up since they didn’t make it into the mask during the 2nd step.
After those 3 spheres, a fullscreen quad is rendered above the whole scene with the color of the light. Of course, you don’t want to have the whole scene lightened up! Therefore you render the light only where the just rendered stencil buffer mask has bright pixels.
Here’s an example of the whole process. Notice, that some lights “disappear” from the scene, since they are hidden by geometry. Like mentioned above
, everything is rendered in Zelda. There’s no distance culling.
Games and Culture have a high standing in Hyrule and one way to live this is via shadow play. But it’s not too easy! In Zelda, only one light source at the time can influence the shading and shadow of a character. What happens when lights overlap? Well, the closest light wins the battle:
A nice detail: If a change happens, the light source seems to “jump/move” to the (now closer) light position. You can see, that the transition between old and new shadow isn’t happening within one frame but looks kind of smooth:
The shadows itself are really interesting too! In our world, everything casts a shadow. In the Zelda game only characters do! Every character in a scene gets its shadow rendered into a 128×128 texture. Only big monsters fill this texture to a major degree.
We can’t prove it, but it looks like that these shadow textures are put on an extruded version of the collision geometry. The “prove” can be found by a quick look at the shadow itself:
The shadow “cuts” the stairs. Exactly like Links feet do (which are placed on the collision geometry)! This leads to our heavy assumption, that the collision geometry is used for the shadow placement. Another clue is, that if we deactivate the drawcall where the collision geometry shows up, also the shadow is gone:
Interestingly, the collision geometry for that specific draw call doesn’t cover the whole level. It seems that somehow only special polygons/tris (near Link) are used. Here you can see Link while he runs with activated collision geometry drawcalls:
The vertices of this geometry don’t seem to be “stitched” together because you can clearly see gabs between differently oriented faces. This also supports our assumption, that the polygons were extruded to float above the surface.
Not yet enough details? Here’s something no common tourist guide would know: Carpets are geometry which “float” a bit above the ground. To avoid that your feet stick in the carpet, it has its own collision geometry (in addition to the ground below) which leads to a double shadow:
When a character stands in water, it looks all totally like it should. The shadow is on the water surface and all is perfect and how we would expect it:
But look how this impression changes when the major part of Links body is below the water surface (which doesn’t influence the shadow map generation since it was done as first render step without caring about the environment). It looks like the water is kind of transparent. Two effects by not changing the technique. Impressive!
Thank you for reading through the whole guide! We hope, that we could serve some useful information and we would be happy if you got motivated to start a nice tour to and through Hyrule.
For any further questions feel free to contact us! Below you can find additional sources for more information about Zelda, Hyrule and technical details.