Supper Mario Broth: Special Zone, Issue 25 (Patreon)

Welcome, everyone, to the 25th issue of Supper Mario Broth: Special Zone.

I apologize once again for the delay between updates. Despite setbacks, I will continuously attempt to get back to posting Patreon updates on the schedule outlined in the tier descriptions until I succeed. Please note that, as always, you are entitled to a full refund of all your contributions that will be issued immediately and without question upon request. 

Today, I will take a look at handcrafted dioramas of courses from Super Mario 64 found in a Japanese guide for the game, and analyze them for their accuracy to the game.

Super Diorama 64

Super Mario 64 Complete is a 1996 Japanese guide for Super Mario 64:

It was published under the ISBN  4889914110. Unfortunately, the guides seem to be very rare; the copies I was able to find listings for sell for hundreds of US dollars each. There are also no complete scans available on the Internet, as for the majority of Japanese guides. 

Nevertheless, the curator of N64th Street, a pictorial blog about Nintendo 64 games, was able to obtain scans of the pages of the guide that depicted maps of the game's courses. All guide scans in this article originate from n64thstreet.com; my deepest thanks to the author for publishing these images.

The maps are different from any other maps of levels seen in both other Super Mario 64 guides (of which I have previously covered two in this article) and Mario game guides in general, as they are not 3D renders or vector graphics. Instead, they are photographs of custom-made dioramas of the courses. Please take a look at this image of Peach's Castle:

In what was undoubtedly a very time-consuming endeavor, the designers recreated the terrain with a variety of materials and even added a few of the characters that appear in that location in the game. I will compare each of the dioramas to the areas of the game using screenshots taken from the most similar angles to show which parts were translated into crafted form with near-identical accuracy and which were simplified.

Here is a screenshot of the castle grounds as seen in-game. Unfortunately, due to the game's optimization resulting in the culling of all objects that are even slightly outside the camera's field of view, the trees on the right side of the image are not rendered; although their shadows still show their locations. In order to get all the trees to display, the in-game camera would have to zoom out further than normally possible. In the cases of the courses, this leads to issues that necessitated me using a level editor to view them; details below.

Although the relative positions of the trees and other features of the terrain is different in the diorama, one impressive detail is the correct number of trees. There are four trees to the left of the bridge and five to the right, which is consistent with the game. The amount of windows on the castle is also accurate. 

Strangely, the cannon structure on the right seems to be made of cobblestone rather than brick. At first I thought this could be due to reusing it between dioramas, but it appears to be unique to this one. This, and a few other details, suggest that the models could have been made with an earlier version of the game as a reference; although this is merely speculation on my part.

The scans include all the main courses except Jolly Roger Bay and Tick Tock Clock; I presume dioramas exist for those as well, but have not been scanned. If you are able to find scans of those pages, please contact me and I will amend this article with the images.

Here is Bob-omb Battlefield:

From the first glance, it is clear that concessions had to be made regarding the size of the characters compared to the environment, and the number of characters that could be put into the scene at once. Goombas are missing from the group near the right edge, as well as Bob-ombs from the canyon area, among others. Still, all the unique and important characters, like the Chain Chomp and Koopa the Quick, are represented.

In order to compare this to the game, I could not simply move the camera around in-game due to the issues mentioned above. For this reason, I utilized a level editor called Toad's Tool 64, which allowed me to take the screenshots without most of the objects not being rendered. Unfortunately, the tool has some of its own quirks so that the images still do not reflect the courses' appearance truthfully, though I believe they work better than the ones possible to take in-game for this purpose.

(The online resource noclip.website does have maps available that would work even better, but they are of the Super Mario 64 DS version, which is too different from the original to be used in this comparison.)

Here is Bob-omb Battlefield as viewed in the editor. The accuracy of the amount and placement of the trees continues, although due to their relative size compared to the enviroment, they appear much more prominent in the diorama. If you look at the area with the wooden pegs in the scan, the trees and the Goombas are so much bigger than the pegs that they are barely visible. 

For some reason, the big black rocks (presumably granite) in the ground are replaced by light grey rocks, with some of them being missing entirely. Compare the prominence of the black hill on the left side in the screenshot with the light grey, smaller one in the scan.

Next, Whomp's Fortress:

The lower part here is rather accurate, but it becomes less so further up the course. Near the top, the Bullet Bill Blaster, arrow-shaped floating platform, and the platforms emerging from the tower are all missing, although the last ones could be due to the Whomp King being present and the scene being too cluttered otherwise. Notably, every single enemy that would be visible from this angle in-game is also present in the diorama, plus some extra characters like Hoot the Owl.

Interestingly, the level editor does not show the water in the pool or the platforms on the tower either, which almost makes me suspect that the designers received a level editor screenshot themselves to work from, although the abundance of details in the dioramas that would clearly be missing from any editor image also suggests that this is unlikely.

Due to Jolly Roger Bay being missing, next up is Cool, Cool Mountain:

Here, due to the obvious difficulties of making a tall diorama compared to a flat one, more concessions have been made regarding the relative positions of the course elements to each other. Compare the course from the game:

There is a significant shortening of the distance between the lowest level (with the Penguin mother) and the next-highest level connected to it via the ski lift. This causes the floating platform with the Bob-omb Buddy to take up almost the entirety of the space between the lift and the long bridge, while in reality, it is tiny and only takes up a fraction of that space.

I suspect the tree models for this diorama were reused from one of the others and covered with white powder, explaining why they do not match the conifer style of the trees in the actual course.

Here is Big Boo's Haunt:

The diorama is missing the outside of the mansion and the underground, which is understandable due to their relative simplicity.  One advantage the diorama has is being able to represent a whole floor of the house at once by design; whereas in-game, viewing a single floor at a time is impossible due to how the rooms are loaded. Since most rooms' lower and upper parts are connected, they all render together, making it impossible to achieve the same perspective seen here, even with a level editor. 

In fact, the closest thing to this we have is the floor plan from Super Mario 64 DS:

This is largely unhelpful, but even from this we can see that the basic floor plans of each room are correct in the diorama. Strangely, the only major change that I can find is that the top right room on the second floor has four coffins in the diorama compared to six in the game (which is unfortunately not visible on the Super Mario 64 DS map, either).

Next, Hazy Maze Cave. For this, I would like to first show what the course looks like in-game to demonstrate its spatial complexity:

As you can see, the cave has three layers, including some rooms that span two layers, and corridors that change elevation. To create a diorama that represents all this would be nearly impossible, which is why the designers opted for a simplified version:

To see how accurate this is, we need to view the course in-game from an overhead rectilinear perspective, like this:

We can see that the layout of the entire cave is extremely simplified, with the exception of the elevator room in the lower right, which is reasonably accurate and was clearly the main focus of the designers. The hidden passages leading from the Hazy Maze in the upper right corner to the Navigating the Toxic Maze and A-Maze-Ing Emergency Exit stars do not appear hidden at all and instead seem to be connected directly to the maze.

Also note that a small diorama of the underground lake, complete with a Dorrie figure, is found in an inset on the left.

Next, perhaps one of the most accurate dioramas, Lethal Lava Land:

Due to the extremely flat nature of the course, it is comparatively easy to create a diorama of it without the compromises required from more complex courses. Here is it in-game:

Please note that due to limitations of the editor, the Bowser puzzle in the center left and a few other platforms are not displayed. However, even with this, it is clear that most of the platforms are present and in their correct places. The only discepancy large enough is the platform with the three Bullies in the top right corner being much closer to the edge of the stage in the diorama than in the course. Note also that the number of diamond patterns on the sinking platforms is almost entirely accurate, with only the one left of the Big Bully platform having 8 in the scan and 10 in-game.

Next, Shifting Sand Land:

Due to the importance of the pyramid, it is much bigger in the diorama than it is in-game. However, the other features seem to be remarkably accurate.

Unfortunately, it was not possible to take a satisfactory screenshot of the course in the editor due to the models for the whirlwinds blocking the view, so here is the camera being moved in-game. As before, you can see the limitations of the game engine by the top left pillar no longer displaying properly, and the same going for the left side of the Tox Box maze, as well as the shape of the oasis. Still, even with these imperfect images, it is visible that no major objects are missing from the diorama.

The diorama for Dire, Dire Docks is very simplified to the point that providing an in-game comparison is meaningless:

The most prominent part of the first part of the course, the whirlpool that sucks Mario in, is completely absent, although the screenshot on the left side of the page does point towards a spot on the diorama with nothing in particular, showing the whirlpool.

In the upper right corner is a smaller diorama of the submarine room, which is also largely useless for the purposes of a guide due to missing the entire upper layer of floating platforms and moving poles on the ceiling, although I must point out the inventiveness of simulating the water surface with what appears to be a thin sheet of metal going through the middle of the submarine, attached to the bottom of the dock on the right. The submarine halves are likely glued to it. The jet stream is also missing from the seafloor in that room, like the whirlpool.

Going to the castle's next floor, Snowman's Land:

The angry face on the snowman that seems to have been added for no discernible reason also suggests this was based on a different, earlier form of the course, as nothing else in the guide has been altered in such a drastic manner.

Here is the course seen in the editor:

Interestingly, the icy rock in the middle of the area has been modeled as grass, or a similar green texture, instead. As before in Cool, Cool Mountain, the trees are reused from the other scenes. With the focus on recreating every single tree, it is unclear why the tree on top of the snowman's head, arguably the most visible tree in the course, is missing.

Here is Wet-Dry World:

While there are differences in geometry, the part that most differentiates the diorama from the course to me is the color of the walls. While the course has light grey tiles with relatively low contrast, the diorama has yellow, almost bathroom-like tiles with a highly visible grid, giving the course the look of, perhaps, a traditional swimming pool.

The Arrow Lifts on the left of the area, while so small as to be hardly visible in the course itself, are much larger in the diorama. Outside of making the colors of some of the platforms more vibrant, the main difference is again, as with the Cool, Cool Mountain diorama, a reduction in height between the tiers of elevation. The top area of the outer wall, which in the course has a very visibly different diamond pattern, is neither sloped nor patterned in the scan, which is peculiar given how much care went into recreating the pattern on the Lethal Lava Land diorama.

Tall, Tall Mountain gets two photos of its diorama due to containing too many points of interest to fit them all on one side:

What is particularly interesting is the water on the bottom. This is clearly taken from the course's skybox, which depicts a sea with distant islands (the most common skybox in the game, also used in Peach's Castle Grounds, Bob-omb Battlefield, Dire, Dire Docks and Tiny-Huge Island). However, this is the only time the skybox is incorporated into the diorama.

Interestingly, here, it was not the height of the course that was reduced, but the width. The actual course is much less "pointy" than the diorama. The addition of the gradient going from bottom to top in the guide is very useful to differentiate between the elevations, as the mountain in-game has the same rock texture throughout, which camouflages the paths against the rock.

Interestingly, looking at the left side of the diorama, one of the Lakitu Bros is present, to show that the bridge is narrow and needs extra attention to the camera.

Tiny-Huge Island is next:

This diorama should strike anyone who has played the game as odd due to the change of the rock wall in the bottom area to an artificial brick wall. Compare how different it looks from the actual course:

While the rock wall has a sharp top edge and essentially ends on the left side, leaving only a small rock as a terminator, the brick wall in the diorama has a flat top that looks as though Mario could walk on it, as well as having a sizable portion of it remaining to the left of the opening next to the Piranha Plant.

Taking this design into account, it appears to make slightly more sense than the game's finished design. The small, door-shaped opening in the rock wall is less logical than the same opening in a brick wall, plus, although of course realism has never been a priority for Mario game terrain, the fact that the rock wall is perfectly straight and ends in a sharp edge seems rather artificial, and would have looked more fitting the way it is depicted in the diorama.

There are also a few texture replacements on the sides of ledges, and the island on the right beach is much lower and has no pole on it. 

Finally, due to the Tick Tock Clock map being missing, we are left with Rainbow Ride:

It appears somewhat cluttered due to the metal sticks holding up the platforms, but is actually reasonably accurate minus the fact that it is missing the entire lower area of the course:

The pole going down on the bottom of the screenshot is not present, likely due to this being a physical diorama, the platforms in the missing section would have to be lower than the rest, which would only be possible by raising the existing platforms. Likely, the metal supports would not have been able to hold the objects if they were made to be twice as long.

Outside of spacing out the rainbows around the airship to avoid clutter, the diorama is accurate. Of special note is the maze on the left, which is an almost identical reconstruction of the maze in-game, which is impressive given how simplified other maze-like structures like the Hazy Maze in Hazy Maze Cave are. 

This concludes the showcase. I hope that this comparison was able to show how much care went into the construction of the models, and what kind of concessions need to be made when translating 3D video game worlds into a physical space.

Thank you very much for reading.