If you appreciate the work done within the wiki, please consider supporting The Cutting Room Floor on Patreon. Thanks for all your support!
This article has a talk page!

Development:Super Mario World (SNES)/Background Graphics and Tilemaps

From The Cutting Room Floor
Jump to navigation Jump to search

This is a sub-page of Development:Super Mario World (SNES).

To do:
  • Make sure that all file dates are documented using Japan's timezone (GMT+9).
  • Verify that all tilemap files in the leak have been documented.
  • For broken looking tilemaps, see if there isn't a graphics file (or simple shift of) that will make the tilemap look correct.
  • If something fits better in another section, put it there.
This page is loooong...
Consider grouping related content into additional subpages to ease readability.
Cacti speak Japanese.
...But what does it mean?
This game has text or audio that needs to be translated. If you are fluent with this language, please read our translation guidelines and then submit a translation!
Notes: See Castle Cutscene and End of Game section for Japanese text.

The contents of the leak can be found in NEWS\テープリストア\NEWS_11\hino\z-mario-4\. Graphics data is stored in .CGX files, palette data in .COL files, and tilemap data is stored in .SCR files.


Overworld Tilemaps

Very Early Overworld

From the earliest leaked files, Super Mario World had plans for a very different overworld than what was finally created.

Overworld Map Graphics


01 April 1989 22:19


23 June 1989 16:19

SMW Early icons.png SMWleak Map rom.png

These are the earliest map tiles used for marking levels, land, and Mario and Luigi's starting position. These are very reminiscent of Super Mario Bros. 3.

Earliest Overworld Map

Some of the earliest tilemap data found within the leaks contain mockups of possible overworlds for Super Mario World.


01 April 1989 22:19


18 April 1989 19:31


19 April 1989 17:40

Map smwbeta.png SMWleak Map-B.png SMWleak Map-C.png

While map.SCR has graphics provided for it in the leak, map-B and map-C do not. From the basic pattern of the tiles, both of them look like they were going to be in the shape of a mushroom, with a winding path that Mario would traverse from the bottom to the top. Map-C also appears to be much more crudely drawn than map-B. There is also a map-A file within the leak, but that map looks to be almost identical to map.SCR, except for the corrupted ocean tiles on the top and bottom of the map.

Mode 7 Overworld

Some of the early tilemaps reference SNES background mode 7 in the title, which allowed for stretches and rotations to be performed to one background layer. This means the developers may have originally intended for the overworld to be zoomed in and out or perspective to be applied on it like in A Link to the Past. Because of mode 7's limitation of only one background layer, the overworld panels and paths are on the same layer as the actual terrain itself. As the final game uses mode 1 for the overworld, most of the terrain is placed on layer 2, while paths and level tiles are placed on layer 1.


25 April 1989 21:26


06 June 1989 16:42


06 June 1989 17:35


06 June 1989 17:40

SMW Map-mode7.png SMW Map-mode7-B.png SMW Map-mode7-A.png SMW Map-rom.png

The developers experimented with multiple ways of laying out the pathways. The first version contained two paths from the pipe near the bottom left of the main island Mario presumably would enter through, with one pathway apparently having a dead end, and the other leading to a pipe up to Bowser's Castle. Map-mode7-B.SCR is almost the same overworld as in map-mode7.SCR but parts of the map have been raised, and a new singular pathway exists around the world. Map-mode7-A.SCR has the same shape as map-mode7-B.SCR but from each numerically defined tile there is a branching pathway with three alphabetically labelled tiles. Map-rom.SCR contains the map devoid of any level tiles and pathways.

Intermediate Overworld

By virtue of its name and the fact that four rows of tiles along the bottom match with the bottom four rows of tiles in the "mode 7" set of overworlds, two tilemaps, chizu-2.SCR and map-UNIT-1.SCR contain overworld data dated between mode 7 and the following overworld. However, the graphics aren't included in the leak and any existing map graphics render garbage. Still, from palettes a general design can be sculpted.

29 June 1989 14:08

SMWleak chizu-2.png

Mushroom Overworld

Various early tilemaps exist with the general shape of a central mushroom overworld, with differing sets of unknown graphics. kinoko.SCR contains the earliest mushroom-shaped overworld in the leaks. It is likely that one of the tilemaps uses the graphics from 1989 build screenshots of the game.

13 July 1989 14:01

SMWleak Kinoko.png SMWProto Map.jpg

Using the map graphics from X-I-P-master.CGX as a base, one can rearrange its tiles to complement what these SCR files expect to find. While many tiles were clean matches, some had to be improvised to fit the SCR layout (namely some cliffsides had to be manually mirrored, and the creation of some specific grass, cliff, and water tiles). In v-ram-kuppa-big-block.CGX there appears to be early waterfall animation tiles for kinoko.SCR. With some tweaking of the palette, a quick mockup can be made. Further discussion can be found on the talk page.


13 July 1989 14:01


07 September 1989 10:21


21 September 1989 14:06

Kinoko waterfall mockup X-I-P-master.CGX

17 December 1990 15:48

Improvised Tiles V-ram-kuppa-big-block.CGX

3 July 1990 16:53

SMWleak Kinoko scr 1.png SMWleak Kinoko-TEST scr 1.png SMWleak Map-bg-2-0.png SMWKinoko Waterfall.gif SMWleak X-I-P-master.png SMWleak KINOKOTESTNOTASSET.png SMWleak V-ram-kuppa-big-block.CGX.png

map-bg-2-new.SCR expands the mushroom overworld by moving the mushroom island itself to the bottom left, and adding an additional island that leads to a large mainland. The final map graphics and palette files are almost compatible with this overworld.

25 September 1989 17:25

SMWleak Map-bg-2-new.png

This new overworld design is expanded on with four more known tilesets, adding paths and various pieces of decoration around the map, as well as minor terrain changes.


20 October 1989 10:30


26 October 1989 09:49


19 December 1989 10:34


15 March 1990 14:53

SMWleak Map-bg-2-taki.png SMWleak Map-bg-2-end.png SMWleak X-map-bg-2-end.png SMWleak X-map-bg-2-first.png

From the various states of completeness of the maps, it may have been that the latter two maps were intended to be the original overworld before any levels were completed and events shown, and that the former two would be completed overworlds when all event tiles had been revealed.

Test Overworld

map-bg-2-chika.SCR is a notable variation of the mushroom overworld for having the word "TEST" written at the top of the map. The bottom four rows of 8x8 tiles expose the fact that this map was placed over top of the mushroom overworld file. The filename suggests that this was an early test for Vanilla Dome or another underground sub-map ("chika" is Japanese for "underground".)

23 October 1989 10:02

SMWleak Map-bg-2-chika.png

Layer 1 Tilemap Data

Layer 1 data for the overworld handles information regarding level pathways, level tiles, pipes, and some additional decorations. Level tiles from the graphics file v-ram-map-kuppa-shima.CGX and level path graphics from v-ram-map.CGX can be combined in order to create a useful tilemap to look at old layer 1 data used for the mushroom overworld designs.

The oldest layer 1 tilemap, map-bg-1-0.SCR, is applicable to the very first mushroom overworld, when it is laid over 1989 screenshots.

21 September 1989 16:49

SMWleak Map-bg-1-0.pngSMWleak Map layer 1 overlay.png

Two more sets of layer 1 tilemaps correspond to the two versions of the completed mushroom overworld.


16 October 1989 17:42


26 October 1989 09:48

SMWleak Map-bg-1-end-sfx.png SMWleak Map-bg-1.png

The layer 2 data can be overlaid with the layer 1 path data to get additional detail about decorations and how Mario would traverse the overworld.

map-bg-2-taki.SCR overlay

20 October 1989 10:30

map-bg-2-end.SCR overlay

26 October 1989 09:49

SMWleak Taki overlay.png SMWleak End overlay.png
  • At this time in development, consistent with screenshots of older 1989 builds of the game, castles did not yet exist in the game, and the castle sign graphics were instead used to house graphics for other pieces of layer 1, like horizontal bridges, cactus and tree decorations, cave entrance tiles, and the tower from Super Mario Bros. 3.
  • There's also a peculiar number of Yoshi houses, with five on the left overworld and four on the right overworld. It is not unreasonable to assume that the one in the center of the main mushroom island should be the first one there, but perhaps others on the map were placeholders for another type of tile (maybe switch palaces?).
  • The first overworld does not use multiple palettes for the level paths, unlike the second one, which indicate what type of path it is (walking, wading through water, climbing up ladder). In fact, some of the layer 1 tiles used to create the diagonal paths through the water don't exist in the final game.
  • The second overworld also uses four sets of 16x16 water tiles on layer 1 (indicated by X-blocks). The final game only uses three of these water tiles on layer 1, used to cover up the fish sprite that jumps from the water. Perhaps this is an indication of more fish originally or possibly more water sprites planned.

Lastly, the layer 1 tilemap data for the "TEST" overworld is included in the leaks.


23 October 1989 10:01

map-bg-2-chika.SCR overlay

23 October 1989 10:02

SMWleak Map-bg-1-chika.png SMWleak Chika overlay.png

For a "TEST" overworld, there is an interesting connection to be made here. The cave entrances on the main mushroom map appear to overlap exactly with the cave entrances that are in the "TEST" overworld!

SMWleak Chika taki overlay.png

Perhaps the "TEST" overworld was once intended to be an underworld to link up the caves on the main map or merely a testing of warps.

Layer 2 Event Data

MAPTABLE.SCR and MAPPANEL.SCR lay out 12x12 sections of layer 2 tiles that are laid out on the overworld map as part of events that trigger upon completion of a level. These correspond with the older set of mushroom overworld graphics which we only have a few.


07 September 1989 10:20


06 September 1989 21:47


MAPTABLE-6.SCR lays out large event tiles for the known graphics variant of the mushroom overworld, perhaps giving a link in certain places between the overworld in its undetailed state to the state with all the paths revealed.

23 March 1990 09:46

SMWleak MAPPANEL-6.png


This peculiar outlier of a map file bridges the gap between the mushroom overworld and the overworld in the final game. Who knows what this gridlocked mess could have been used for?

16 April 1990 21:16

SMWleak X-map-bg-3.png

The name of the file doesn't make any sense either, as it suggests this would be used for layer 3. In previous tilemaps and in the final game, background layer 2 is used to place the main terrain pieces of the overworld map, while layer 3 is used for the overworld border and status bar.

Blank Map Tables

Map table files B, C, and D are formatted like the large event tiles tilemap, but are all identical in the fact they are completely empty.


28 May 1990 11:01

SMWleak MAPPANEL-6-B.png

Final Overworld Data

There are a few tilemaps that correspond to the overworld that appears in the final game. While most actually share no differences between it and the final version, a few give insight to the developers' plans for Dinosaur Land after the mushroom overworld was presumably scrapped.


This map holds early Layer 1 data for the submap (the section of the overworld that contains Yoshi's Island, Valley of Bowser, Vanilla Dome, Special World, Forest of Illusion, Star World), though only the data for Yoshi's Island, Valley of Bowser, and Vanilla Dome exist. Graphics for a locked key entrance and a much larger Bowser's Castle also exists in the tilemap data, but as it looks out of place, it was probably a test. From the look of the extraneous Bowser's Castle, it was once meant to be a much larger castle with Back Door as part of the main structure and not an offshoot.

07 September 1990 13:10

SMWleak X-map-bg-1-chika-shiro.png

Some observations give key development insights:

  • This overworld apparently predates the addition of castles, as the positions where castles are in the final game only have fortresses. They also lack the usage of the numbered signs to indicate which castle it is.
  • In Yoshi's Island, there is a level tile between the position of Yoshi's Island 1 and the Yellow Switch Palace. From the palette data and position, this level was meant to be a water level (like Yoshi's Island 4) on the way up Kappa Mountain. Additionally, the level tile in Yoshi's Island 4 is not indicated to be a water level based on the palette of the tile.
  • In the Donut Plains section of Valley of Bowser, Donut Secret 2 is located two tiles below where it is in the final game. In the final game, the exit of the level is directly left of the pipe, making a straight connection that would be consistent with the positioning on the overworld. In this version, there is a curved path from the level tile to get to the pipe.
  • In the main section of Valley of Bowser, all level tiles to the right of Valley of Bowser 3/4 are one tile farther to the left compared to the final version, making the path from Valley of Bowser 3 and the ghost house (and Valley of Bowser 4 and the castle) one tile long instead of two.
  • The warp into Valley of Bowser is placed below Valley of Bowser 1 with a straight path aligning it; in the final version, the player enters Valley of Bowser upwards and swings left to approach Valley of Bowser 1 from the right.
  • The Star Warp from the final game in Valley of Bowser is replaced with a pipe that is one tile higher than the Front Door tile, and also isn't linked to Front Door. In Vanilla Dome, Vanilla Secret 1 is completely missing a secret exit that would lead to a star warp and the level tile isn't even marked as such.
  • For Chocolate secret, there is no path between that level and either pipe, conflicting with the warp tile that is along the path. This either could have been a path not created yet, or that hardcoded paths were once used instead.
  • In Vanilla Dome, there is no secret exit attached to Vanilla Dome 2 that would lead to a Red Switch Palace, and in Valley of Bowser there is no trace of the unused Red Switch Palace in the final game. If switch palaces weren't in the game at this point in development, then what is on top of Kappa Mountain?

Unfortunately, the existence of the big dot level tiles give no further information to solving the mystery of the big dots.


This tilemap contains only Bowser's Castle, and the bottom four tiles of Layer 1 data, which only show two paths from Forest of Illusion. The Bowser's Castle in this version uses palette and tile data that remains in the final game.

12 June 1990 21:44

SMWleak map-bg-1-chika-kuppa.png


This tilemap has two copies of the paths in star world, with one in the top left and one in the final position in the bottom right.

SMWleak majikant-bg-1-michi.png

The star warps don't use star tiles, and use regular level tiles like the regular levels in star world. Also, the paths that enter tiles are different from the final version. This version uses tiles that enter at a 45-degree angle from the side. This type of path actually breaks if implemented in the final game, and the final game refers to a different niche tile for this purpose.


This tilemap shows animation frames of the entrance to Valley of Bowser rising out of the water. Compared to the final's whopping 11 frames of animation, this one has a much simpler five.

02 July 1990 10:11

SMWleak kuppa-shima.png

Final Dinosaur Land

Files X-map-bg-1.SCR and X-map-bg-1-chika.SCR correspond to the layer 1 data of the main map and submap, respectively. There don't appear to be any changes between this version and that of the final game, and the dates of the files are after the Japanese release of the game.

Files X-map-bg-2.SCR and X-map-bg-2-chika.SCR correspond to the layer 1 data of the main map and submap, respectively. While bg-2.SCR is identical to the final main map, bg-2-chika.SCR has some of the edges of each world map covered by solid tiles, which is not how it is in the final game.


09 September 1990 15:00

SMWleak map-bg-2-chika.png

Final Event Tiles

MAPTABLE-6-A.SCR and MAPTABLE-2.SCR contain the 12x12 layer 2 event tiles and 2x2 layer 2 event tiles, respectively that correspond to the final game.


07 September 1990 20:44


07 September 1990 11:06

SMWleak MAPPANEL-6-A.png SMWleak MAPPANEL-2.png

Map Tables

X-map-unit.SCR lays out 16x16 Layer 1 tiles, including path tiles, level tiles, warp tiles, and other decorative tiles on layer 1. The graphics included in the leak include two significant sets that can be used for this tilemap: v-ram-map.CGX and v-ram-map-kuppa-shima.CGX.

with v-ram-map.CGX

08 September 1990 02:41

with v-ram-map-kuppa-shima.CGX

28 June 1990 15:13

SMWleak MAPTABLE-2-map.png SMWleak MAPTABLE-2-kuppa-shima.png

With the map table supplied with graphics data from v-ram-map.CGX, outlines of the pathways can be seen. With v-ram-map-kuppa-shima.CGX, much of the older level warp and level tiles that don't look correct even with the current graphics now display, revealing much about the types of level Mario could encounter on the overworld during development.

Status Bar and Title

Various tilemaps and graphics files feature layouts of the status bar, map overlay, title screen, and more.

Status Bar and "Mario Start" Screen


24 October 1989 15:20

SMWleak Mater.png

One of the tilemaps includes a very early Mario Start screen, quite reminiscent to the format of Super Mario Bros.. The "Mario Start!!" text uses the traditional 1x1 font rather than the taller font used in the final game, as well as two exclamation points. It also features in the center the current world and level number, showing that there was no level naming system in place rather than world-level.

This version of the screen also contains the status bar and item box. The player name text in the top left corner uses standard text rather than a custom-drawn scripted Mario/Luigi text. The level time is displayed between the lives counter and the item box, and does not use the custom-drawn three-wide graphic for the "Time Text". The points and coin counter are displayed on the same line to the right of the item box, instead of one on top of the other. Because of the width of the coin counter, the points counter only has space for six digits rather than the seven of the final game.


On top of including Layer 3 background graphics for castle levels, this tilemap includes two copies of an older status bar design.

18 June 1990 20:03

SMWleak 2bit-shiro-tate.png

This version does use the stylized "Mario" text with uncentered lives counter. Compared to the last image, the time and coin counter has swapped places, and the points counter is still six digits. This is almost identical to the earlier magazine scans, except that the timer is white here instead of yellow.


This tilemap shows a version of the status bar that is almost the same as the final game.

06 September 1990 23:51

SMWleak 2bit-mater.png

Like in early screenshots, the lives counter is not centered under the player's name. The time counter itself is white instead of yellow, and the coins and bonus points counter is one tile closer to their respective values than in the final version. Also, there's a few tiles under the status bar for some reason. The points counter and coin counters are in opposite places, and the points counter only seems to display six digits instead of seven.

Title Screen

An assortment of tilemaps and graphics data yields insight into the development of the title screen, title screen graphics, and menu layouts.


This early tilemap features two panels, dated July 1989. The left panel, for some very strange reason, contains tilemap data that, when the graphics are properly deciphered from chr-stock-F.CGX, show the boss platform for Iggy and Larry! The right panel, when deciphered, should yield the title graphic.

14 July 1989 15:40

SMWleak title-obj.png


This tilemap advances on the previous one with four panels. The second panel now properly uses the graphics for the title screen, and the palette data suggests this should be the version of the title screen with a flat blue palette instead of the colorful final one, however the title screen frame is missing. The third panel, based on its pattern, shows that this is actually the zoom-out of the mushroom island seen in a 1989 build screenshot of the game, though no graphics for this exist in the leak.

17 July 1989 09:33

What is that? A corrupted background with a lot of letters?

SMWProto Title.jpg

For some reason, the palette for the 1989 date is different than the rest of the text on the line. Like the early screenshot, the Nintendo name is not given its own unique graphic and instead uses the regular font. The trademark symbol is also floating near the "Super" rather than to the right of "World."

title-new.SCR improves the inconsistency of the Copyright Nintendo line by giving the copyright symbol its own distinct palette from the other text. This version now uses the color scheme of the final game.

04 June 1990 10:09

Not again!


This tilemap is very similar to the final game, showing four different states of the title screen.

07 November 1991 16:34

SMWleak 2bit-title-new.png

The first screen shows something similar to the file select for the Nintendo Super System, which doesn't quite look correct in the Japanese graphics set as it only came out in the United States. The second, third, and fourth panels show the file select, file deletion, and player select screens respectively, all of which are accurate to the final game save the selection cursor being green in the middle two screens. Also, the trademark symbol is still out of place!

2bit-title-new-NES.SCR is the United States localization version of the title screen frames, however, this version seems to have more unused information.

13 May 1991 14:38

SMWleak 2bit-title-new-NES.png

The tilemap features a proper frame for the first image, accurate to the final game. The second set uses both the Japanese version of the logo, with the American trademark symbol, but also uses the "Super Mario Bros. 4" subtitle, as well as the "AND" symbol for Trademark and Copyright! The third frame uses the Japanese logo without subtitle, with the Japanese trademark symbol in the correct position. The fourth screen only uses a special version of the graphics set, v-ram-2bit-NES-no-shadow.CGX, that has the outline of the American version, but as the name suggests, no shadow at all. Including the "Super Mario Bros. 4" subtitle, this last frame uses the proper zone text intended for the Super System arcade.

Overworld Frame


This tilemap shows four different possible situations that a player might encounter. The status bar at the top is consistent with the layout of the bar in the 1989 early build screenshots, with the world counter displayed in large blue text without unique level names, as well as coin and score counters being present.

30 August 1990 15:31

SMWleak 2bit-map-back.png

The layout of the lives transfer screen is different than that in the final game. In the final game, the life counter is displayed next to where the sprite for Mario and Luigi are, and their names displayed well below the sprite.

SMWfinal lives-transfer.png

Additionally, the save prompt in the second frame has three options, and very peculiarly there's a file deletion menu that would have been accessible from the overworld.


10 December 1990 14:14

SMWleak 2bit-map.png

This tileset comes from later in the development of the game, with the file select screen apparently no longer present from the overworld, and the first panel only showing the player lives counter and the full level name. The save prompt is down to two options (save and continue, continue without save), and the third panel, which was before filled with strings of text saying "Mario" now has the continue/quit options from the game over screen.

2bit-map-NES.SCR is the International translation version of 2bit-map.SCR, though for some reason the old status bar design persists for some of the layouts in the tilemap.

10 December 1990 15:10

SMWleak 2bit-map-NES.png

The second save prompt, which is "Continue Without Save" in the final International version of the game, is written as "Continue With No Save" in this prerelease tilemap.


Uninterestingly, this tilemap as the name suggests merely outlines the dimensions of the save box window prompt for the International version of the game.

Foreground Tilesets

Roof Slope

"Yane saka", which roughly translates to "Roof slope" is one of the first tilemaps devoted to a foreground theme. The VRAM is located in v-ram-yane.CGX, the CGRAM data in yane-saka.COL, and the tilemap in yane-unit.SCR. A yane-saka.CGX exists in the files, but doesn't apply to this tilemap.


24 October 1989 15:23

SMWleak yane-saka.png

This tileset looks similar to the current mushroom platform tileset, however this version is orange and and is remarkably different in the design of the platform/background itself.


This tilemap lays out a couple of underground sections. To the top left is a "level-sized" preview featuring the layer 3 background for the underground. To the right is a basic laying-out of the graphics assets in block form.

19 June 1990 21:16

SMWleak chika-back.png

Interestingly, the tilemap uses a very repetitive set of tiles for the vertical walls on the left section. The correct tiles (at least the configuration used in the final game) are on the right side.


This tilemap lays out various sizes for the crates in the Sunken Ghost Ship, the launcher for the Torpedo Ted, and the foreground seaweed object that goes unused in the final game.

08 February 1990 14:40

SMWleak gost-ship-contenta-unit.png


This tilemap lays out tiles to be used in the mushroom tileset, including stacked mushroom platforms, the moving rope conveyor belts, and various line guide configurations.

13 February 1990 11:42

SMWleak sora-unit-sen.png


This tilemap lays out the tiles for the large plants seen in levels like Vanilla Dome 4 and Groovy.

11 January 1990 15:35


The corresponding palette graphics make the spots on the side of the blocks a much darker blue than in the final. Also, despite the yellow palette being correct in the 16x16 tile selections in the top of the tilemap, the example plant uses an incorrect, dark yellow palette instead.


This tilemap lays out the graphics for the tapestry seen in Door 3 of Bowser's Castle. Unfortunately, there's not much relation between the layout here and that of the final game.

06 December 1989 16:45

SMWleak X-shiro-maku-unit.png

Diagonal Pipe

naname-dokan-unit.SCR tilemap compiles 16x16 tiles and puts them together to create the diagonal pipe object.

15 August 1990 15:55

SMWleak naname-dokan-unit.png



This tilemap compiles the old bush textures, among other decorations in two layouts.

24 January 1990 17:12

SMWleak kusa.png


This tilemap generates the large and small bushes using the graphics set in the final game.

15 August 1990 16:18

SMWleak kusa-new.png


This tilemap lays out the graphics for the Layer 2 crusher objects in both upwards and downwards variants.

16 July 1990 16:48

SMWleak shiro-bg-bosun.png


This tilemap contains 16x16 tiles for 16x16 background layer tiles, as well as foreground bushes and other multi-block decorations.

06 December 1989 16:23

SMWleak bg-unit-0.png


This tilemap shows 16x16 foreground blocks and components, such as brick blocks, note blocks, midway points, and goal tapes (including the bottom of the goal tape!). A number of graphics sets work in this tilemap, and therefore multiple sets of blocks that changed later in development can be displayed here.

07 June 1990 09:52

SMWleak bg-unit-1.png


Just two tiles, an orange platform and a solid red lava tile.

27 July 1989 16:24

SMWleak shiro-sel2-unit.png


This tilemap lays out the individual 16x16 tiles for the underground tileset, while sample slopes and other platforms are displayed on the right. It's also identical to X-saka-unit-kihon.SCR.

06 December 1989 16:04

SMWleak saka-unit-kihon.png


This tilemap is like saka-unit-kihon, but with more slopes tests.

03 November 1989 17:39

SMWleak saka-unit-sanple.png


This tilemap lays out individual 16x16 tiles, but lays out sample platforms, including diagonal platforms, in the right half.

24 January 1990 17:14

SMWleak saka-unit-chijo.png


This tilemap lays out tiles for the very special cloud tileset, including intersecting slopes (though some are broken). Too bad this tileset didn't get more playtime in the final game!

03 November 1989 17:24

SMWleak saka-unit-kumo.png


Another set of slope tiles laid out, but with its own cool section on the top left.

19 July 1990 10:29

SMWleak saka-sanple-moniter.png


Another set of slope tiles, but this one fits within the top half of the tilemap.

14 September 1989 11:22

SMWleak saka-sanple.png


A tilemap featuring both 16x16 foreground and background tiles in the top half, and a sample level layout on the bottom half. As the name suggests, this tilemap would use mushroom/rope platform graphics, but when supplied with relevant graphics for it, it looks completely broken. The foreground tiles that look to be walls or platforms refer to sprite graphics instead of foreground tiles.

24 October 1989 15:21

SMWleak sora-unit.png


This tilemap shows 16x16 foreground tiles for the moving escalator in castle levels. Compared with the final, each step is actually very large.

24 August 1989 14:46

SMWleak kaidan-saka.png


This tilemap appears to lay out sloped lava, based on the tiles in the graphics it refers to, though when given a graphics file with sloped lava, it appears broken.

24 August 1989 10:42

SMWleak takisaka-unit.png


A set of slopes for the underground theme, including upside-down ceiling slopes. However, it's broken when supplied with relevant underground graphics.

28 June 1989 11:19

SMWleak chika-saka.png


Like the tilemap above, but includes a layout of 16x16 tiles in the top half of the tilemap.

28 June 1989 11:18

SMWleak chika-saka-unit.png


The following four tilesets lay out individual 16x16 tiles from a generic tileset, which if given graphics from unit-fail-master.CGX, create grids that seem compatible for printing?


12 September 1989 18:23

SMWleak unit-fail-muster-saka1.png


12 September 1989 18:26

SMWleak unit-fail-muster-saka2.png


12 September 1989 19:44

SMWleak unit-fail-muster-saka3.png


13 April 1990 14:30

SMWleak unit-fail-muster.png

Background Tilesets


The "Kabe" tileset comprises a unique set of backgrounds that don't appear in the final game. VRAM data is located in v-ram-kabe.CGX, CGRAM data in kabe.COL, and tilemap data both in kabe.SCR and kabe-unit.SCR. v-ram-kabe-2.CGX is similar to the other VRAM file, though with lighter shading on the edges of the blocks used in the background that look worse.


13 October 1989 17:49


02 October 1989 16:03

SMWleak kabe.png SMWleak kabe-unit.png

The background on the left is less detailed than the background on the right, but occupies the entire tilemap space, while the background on the right reserves some space for laying out possible 16x16 tiles.


sora-master.SCR is a tilemap of the background seen in Yoshi's Island 1, among other levels. There aren't any differences between this version and the final.
X-sora.SCR includes a shifted version compatible for importing into the game.


26 October 1989 21:51


27 November 1989 16:37

SMWleak sora-master.png SMWleak X-sora.png

Background tilemaps in Super Mario World are only 432 pixels tall (missing 10 rows of 8x8 tiles from the bottom), as this is the maximum vertical scrolling range of the camera in a normal horizontal level.


yama.SCR is a tilemap for a completely unused background.

16 March 1990 22:01

SMWleak yama.png

When given graphics by v-ram-yama.CGX, the mountains and other background structures have faces and even eyebrows! The clouds for this tilemap also appear to be much more aggressively shaded, while the vertical hills are clustered strangely.


taki-master.SCR contains an unused background featuring a waterfall and tall polygonal structures.

11 November 1989 19:26

SMWleak taki-master.png


chika-master.SCR and X-chika.SCR are tilemaps corresponding to the underground background in the final game with no changes from this version to the final.


19 December 1989 11:49


19 December 1989 11:51

SMWleak chika-master.png SMWleak X-chika.png

Interestingly, this tilemap has its own "unit" background, chika-unit.SCR, but it's nothing like the actual underground background. With 16x16 tiles laid out in the top half, the bottom half has a background of a different palette, referring to a different set of animated tiles for the background, and a hint of takisaka-unit.SCR on the bottom.

07 August 1989 14:55

SMWleak chika-unit.png

2bit-chika.SCR is a Layer 3 underground tilemap, very similar to the kinoko set of tilemaps. This one, which is much similar to the final game, is far less detailed than the mushroom set and doesn't have mushrooms.

04 July 1990 13:14

SMWleak 2bit-chika.png

The right side compiles the layer 3 tiles into different size rock formations that could have been used to create a more detailed look to the background.

Ghost House

gost-house-yoko-master.SCR and X-gost-yoko.SCR are tilemap for the ghost house. It's again an identical tilemap to the final game.


28 December 1989 10:16


28 December 1989 10:20

SMWleak gost-house-yoko-master.png SMWleak X-gost-yoko.png

gost-house-yoko-old.SCR shows an older background design similar to the one above, but is brighter and has different windows, though it doesn't appear to be properly viewable with the associated ghost house graphics file.

19 December 1989 11:11

SMWleak gost-house-yoko-old.png

X-gost-house.SCR shows a peculiar alternative configuration of the ghost house background, much different from the final!

19 December 1989 16:26

SMWleak X-gost-house.png

X-gost-house-old.SCR shows an even older background design, with peeling wallpaper and many beams. It doesn't use the typical background graphics for the ghost house, instead using those found in v-ram-kabe.CGX.

19 December 1989 11:08

SMWleak X-gost-house-old.png

X-gost-yoko-old.SCR is another old background design that uses v-ram-kabe.CGX. However, this version is shifted for import, and therefore doesn't look correct in the tilemap state.

19 December 1989 11:08

SMWleak X-gost-yoko-old.png


forest-unit.SCR tilemap includes a section of 16x16 tiles for laying out objects given the graphics. Below it is a section of the background, though the palettes cause some of the mushrooms to show up blue. Based on the graphics used to create this image, this background would have been intended for Layer 3 as opposed to Layer 2. forest.SCR lays out the full background from forest-unit in a slightly different order to make it easier to import into the game engine.


12 October 1989 22:02


13 October 1989 09:37

SMWleak forest-unit.png SMWleak forest.png

forest-muster.SCR is another forest-type background, again identical to the version of this background in the final game. X-forest.SCR includes a shifted version compatible for importing into the game.


25 August 1990 22:22


25 August 1990 23:40

SMWleak forest-muster.png SMWleak X-forest.png


kinoko-master.SCR and X-kinoko.SCR are tilemaps that is identical to the forest, except for its associated palette which makes the background tiles a dark gray color, perhaps making this an early version of the underground Layer 3 background. The final version lacks the mushrooms, and is generally far less detailed.


31 January 1990 16:39


31 January 1990 16:38


31 January 1990 16:38

SMWleak kinoko-unit.png SMWleak kinoko-master.png SMWleak X-kinoko.png


shiro-unit.SCR, shiro-muster.SCR, and X-shiro.SCR are castle background tilemaps like the ones in Iggy's Castle, with the latter two laying out much of the graphics tileset.


23 October 1989 20:44


27 November 1989 18:31


27 November 1989 19:22

SMWleak shiro-unit.png SMWleak shiro-muster.png SMWleak X-shiro.png

shiro-2-master.SCR and X-shiro-2.SCR are tilemaps identical to the final game in castles like room 1 of Morton's Castle.


16 February 1990 09:53


16 February 1990 10:05

SMWleak shiro-2-master.png SMWleak X-shiro-2.png

2bit-shiro-tate.SCR is the layer 3 castle background, like in the vertical section of Morton's Castle. Compared to the final version, this one is slightly more detailed and more irregular than in the final version.


18 June 1990 20:03

Final Version
SMWleak 2bit-shiro-tate.png SMWfinal layer3-bg-castle.png

This tilemap includes another early version of the status bar! The changes in that version are covered in its respective section.

shiro-unit-2.SCR and shiro-2.SCR are two background tilemaps that look broken when given castle graphics.


14 July 1989 16:22


15 July 1989 22:22

SMWleak shiro-unit-2.png SMWleak shiro-2.png


sea-unit.SCR lays out all the tiles in the graphics set for the underwater background. sea-master.SCR and X-sea are identical background tilemaps to the water background.


23 October 1989 11:58


27 March 1990 21:23


27 March 1990 21:30

SMWleak sea-unit.png SMWleak sea-master.png SMWleak X-sea.png


rock-unit.SCR lays out all the tiles to be used as part of the rock backgrounds.

24 October 1989 15:22

SMWleak rock-unit.png

The following two rock backgrounds are identical in layout to those in the final game, both with their own "X-" versions.


28 March 1990 18:56


01 June 1990 16:18

SMWleak rock-master.png SMWleak rock-mountain-master.png

01 June 1990 11:37


01 June 1990 16:15

SMWleak X-rock.png SMWleak X-rock-mountain.png


Four more rock/cloud type backgrounds exist in the tilemap with "sel3" in the title, each with their own "X-BG" version.


08 October 1990 19:21


26 August 1990 11:29


08 October 1990 19:35


27 March 1990 22:46

SMWleak sel3-1-master.png SMWleak sel3-2-master.png SMWleak sel3-3-master.png SMWleak sel3-4-master.png

28 March 1990 17:31


28 March 1990 17:31


09 May 1990 19:10


28 March 1990 17:32

SMWleak X-BG-sel3-1.png SMWleak X-BG-sel3-2.png SMWleak X-BG-sel3-3.png SMWleak X-BG-sel3-4.png

The first background is only used in Funky in the final game. Background three has a slightly different cloud layout when compared to the final game, while background four apparently still uses the old "donut clouds" instead of the clouds with eyes like in the final game, though the "X-" version has the correct clouds.

Vertical Mountain

X-yama-tate.SCR contains the mountain background similar to sel3-4-master.SCR but with the correct cloud configuration.

09 May 1990 19:14

SMWleak X-yama-tate.png

Interestingly, this tilemap is actually the full 512 pixels tall, with extra clouds in the sky filling up the extra space.

Ghost Ship

These ghost ship tilemaps are identical in configuration to final game, but with shading that's a little brighter.


07 February 1990 17:38


05 February 1990 14:49

SMWleak gost-ship.png SMWleak X-gost-ship.png

ghost-ship-2bit.SCR and 2bit-ghost-ship.SCR are two identical layer 3 tilemaps for fish scrolling across the screen, except for the fact this isn't used in the ghost ship level, and is only used in other water levels. The layout here is identical to the final game.

05 February 1990 15:54

SMWleak ghost-ship-2bit.png


night-unit.SCR contains 16x16 tiles used in night-muster.SCR and X-night.SCR, again identical to the final game.


25 September 1989 10:53


24 October 1989 21:13


24 October 1989 21:13

SMWleak night-unit.png SMWleak night-muster.png SMWleak X-night.png

Mode 7

mode-7-kaiten-boss-bg.SCR simply contains a layout of the graphics for the rotating background for the Reznor fight.

03 September 1990 14:36

SMWleak mode-7-kaiten-boss-bg.png

mode7-kuppa.SCR contains a layout of the multi-screen background and foreground of the boss fight room in Morton's Castle.

22 September 1989 15:12

SMWleak mode7-kuppa.png

Because of Mode 7's limitation of one background layer, which is reserved for the boss himself, sprite objects are used to create the background. However, for the ceilings and floors and the area above and below those respectively, the mode is switched to regular mode 1, which allows for the platforms and ceiling to be regular foreground tiles.

Early Level Designs

Ghost House Tilesets

There's a particularly significant number of tilemaps within this dump of internal data devoted to Ghost Houses as well as the Ghost Ship, giving layouts of the blocks, structures, backgrounds, and even some levels!


An early mockup of a ghost house level, with individual block designs to the right and multiblock mockups below it.

02 October 1989 13:20

SMWleak gost-house-sel2-unit.png


Another level mockup that makes use of pipes, question blocks, brick blocks, and note blocks on top of the ghost house tileset items. The level design here looks very crude, and may have been intended to give a sample of the tileset rather than be implemented as a final level.

05 December 1989 21:47

SMWleak X-gost-house-sel-2-sanple.png


A design that is similar to sel2-unit.SCR, but has more platforms that are vertically stacked and look to be vertically suspended.

18 December 1989 22:54

SMWleak gost-house-sel2-unit-tate.png


This tilemap features multiple vertically-restrictive tunnels reminiscent of the beginning of Forest Ghost House. Below the two sets of tunnels that is made out of the modern ghost house tiles, there is a another tunnel with walls made out of the Super Mario Bros. 3 fortress tiles, though the background of the tunnel uses the ghost house graphics (with a gray palette).

13 July 1990 10:13

SMWleak gost-house-danjon-BG.png


This tilemap includes level layouts for both the entrance and exits to the ghost houses.

06 April 1990 08:48

SMWleak gost-house-GOAL.png

As it turns out, the exit level is exactly the same as the "Cloudy Ghost House Exit" level that goes unused in the final game! Additionally, the "three bushes" level that is unused in the final ROM turns out to simply be the layer 2 background of the ghost house exit!


A broken level design featuring an older style background in the top left quadrant, and foreground over top that background in the lower right quadrant. However, the ghost house graphics for the foreground are broken when given the final version's graphics.

13 November 1989 21:58

SMWleak gost-house-sanple.png

Bonus Rooms

Various tilemaps have "bonus" in the name, and reveal various ideas for bonus rooms in Super Mario World. The graphics file v-ram-bonus.CGX and the palette file bonus.COL can be used to render these bonus room tilemaps.


07 September 1989 14:37

SMWleak bonus.png

This the earliest tilemap file that exists showcasing two bonus room designs, both using the gray blocks used in Super Mario Bros 3.'s castle levels. The design on the left is reminiscent of the final bonus room, but only contains three rows of coin blocks instead of five and has doors instead of pipes. The design on the right has a vertical pipe presumably used to enter the room with three question blocks. There doesn't appear to be an easy way to exit the room, so perhaps the level would have ended at this point after hitting the question blocks.


09 September 1989 19:54

SMWleak bonus-back.png

This tilemap contains four different bonus rooms laid out. The two rooms on the left side contain pipes that can be used to exit the level, and others contain a platform in the middle of the room. The background graphics for each design are distinct, but not apparently correctly renderable with the dedicated graphics files for bonus rooms. It is the first tilemap to use the final striped bonus room graphics; however, the corners on the bottom left of the room are not correctly rendered, and the palette is slightly darker than in the final game.


bonus-1.SCR and X-bonus-1.SCR both appear to have identical layouts when it comes to the actual room layout itself.

20 October 1989 14:44

SMWleak X-bonus-1.png

The left room simply contains a pipe on the top left and bottom right, with the rest empty. The room on the right has the same pipe configuration but nine question blocks in a square pattern, which is very similar to the bonus game that occurs in the final game upon reaching 100 bonus stars.


bonus-2.SCR and X-bonus-2.SCR contain room designs that are almost identical to those in the final game, with the latter having an issue on the bottom of the platforms that is corrected in the former.


31 July 1990 11:53


20 October 1989 14:45

SMWleak bonus-2.png SMWleak X-bonus-2.png

The primary difference between these rooms and the final, besides the differences in graphics, is that the entrance pipe is set in the wall so that Mario enters flush with the left side of the room, instead of the pipe coming out of the wall. This room is also taller than in the final version, with a sixth row of question blocks included (the bottom row must have been removed for the final, as the first set of question blocks begins aligned to the left). The final exit pipe is unaltered from this version, however.


This tilemap lays out various blocks of different sizes made from the castle tileset, including various gaps.

11 August 1990 16:35

SMWleak shiro-big-block.png

shiro-big-block-enkei.SCR is a variation of shiro-big-block, moving the whole tilemap up a few tiles, and adding a various assortment of six different tiles (undisplayable in the castle graphics set) inside the central gap.

17 June 1990 20:26

SMWleak shiro-big-block-enkei.png


This tilemap shows various 16x16 tiles for the grab cage that shows up in levels like Iggy's Castle and Morton's Castle, animation frames for the rotating cage piece, and a sample layout for a level, showcasing rotating cage pieces, and cage geometry.

09 April 1990 17:05

SMWleak bg-ami.png

For some reason, an identical copy that does not use the same tiles as the original exists on the bottom of the tilemap.


This level is castle-themed and serves as a test for line guides. Interestingly, the line guides are ended with regular wood blocks, instead of a special tile like in the final game. If this were imported into the final game, any object along the line guide would run off at the end. Also, spikes on the right side of the ceiling are upwards pointed for some reason.

16 July 1990 14:09

SMWleak shiro-sen-etc.png


This forest level tilemap shows a combination of foliage, berries, tree trunks, and platforms.

07 June 1990 10:13

SMWleak hayashi.png

The foliage layout to the middle right is the same layout as the foliage object in the final game, though none of the actual layouts in the image appear to use this layout.

Cutscene Graphics

Tilemaps labelled with "demo" in the name indicate cutscenes that Mario isn't able to move in.

Castle Cutscenes

Intro Room

shiro-demo.SCR is a tilemap showing the intro room when entering a castle. There are two versions, with the second one having a wider entrance that isn't in the final game.

14 February 1990 10:43

SMWleak shiro-demo.png

The corresponding palettes to the tilemap make the castle much bluer than how it looks in the final game. Additionally, the bush graphics are glitched, because they actually use an older version of the bush! With the correct graphics it looks like this:

SMWleak shiro-demo-fix.png

Early Cutscene Text

To do:
add a screenshot of the second castle cutscene from the final game to the comparison

The file 2bit-shiro-end-demo.SCR holds early Japanese text data to be displayed during the end of castle cutscene. Compared with the final version in Super Mario World, this early cutscene is very generic withholding references to the overworld. However, it does bear some similarities to the cutscene text that appears when clearing the second castle in the final game.


30 August 1990 13:47

Final Version
SMWleak 2bit-shiro-end-demo.png SMWcastle1cutsceneJ.png
And so Mario defeated the first Koopaling. But! The game has just begun! Who knows what kind of trap could be lying ahead? Is Peach safe!? Having defeated the Koopaling in the castle of Yo'ster Island, Mario rescued his friend who had been turned into an egg, and left for Donut Plains.

Cutscene Level Graphics

shiro-end-demo.SCR has four tilemaps for the four different backgrounds in the cutscene. Other than palette selection differences, there's an erroneous tile on the bottom right panel.

01 September 1990 18:34

SMWleak shiro-end-demo.png

shiro-end-demo-castle.SCR contains the castle graphics used with the cutscene, with one intact and one in crumbles.

01 August 1990 21:51

SMWleak shiro-end-demo-castle.png

shiro-end.SCR is another tilemap for the end of castle cutscene, but the palettes and graphics are broken when supplied with graphics files used to generate the other castle cutscenes.

26 July 1990 20:26

SMWleak shiro-end.png

Ghost Houses


This tilemap has data for the intro room when Mario enters a ghost house, along with 16x16 tiles made from the corresponding graphics file.

26 February 1990 20:54

SMWleak gost-house-demo.png

Based on the 16x16 tiles to the top right, the background clouds apparently had a lot of complexity associated with them, rather than just the simple short clouds seen in the final game. Again, the bush is broken here, fixable by using the older bush graphics.

SMWleak gost-house-demo-fix.png


This tilemap uses the final version of the bush, but also includes the exit area of the ghost house on the bottom!

04 April 1990 10:50

SMWleak X-gost-demo.png

Unlike the final game, the exit area on the bottom of the tilemap has both decorative bushes and clouds in the sky. In fact, the clouds in the sky are in exactly the same configuration as the "Cloudy Ghost House Exit" level that goes unused in the final game!


This tilemap holds the data for the layer 3 "cloudy" effect that shows up when Mario is entering the ghost house. Another file, X-gost-demo-2bit.SCR also has this, but both files are identical.

06 June 1990 10:27

SMWleak 2bit-ghost-demo.png

Ghost Ship

The presence of "demo" tilemaps seem to indicate that the sunken ghost ship (or perhaps multiple ghost ships!) were intended to have an intro sequence.

SMWleak gost-ship-demo-fix.png

The blocks used to create the hull of the ship in this tileset are not similar at all to the tiles used in the final game for the ship.


This tilemap shows two copies of an intro sequence using the mushroom platform tileset, with large green plants coming from the ground, as well as a broken ghost house intro copy. This seems similar in function to the level used for the "No Yoshi" intro.

08 March 1990 18:17

SMWleak X-sora-demo.png

Yoshi's House

doragon-house-demo.SCR is a tilemap that is probably used for Mario walking into Yoshi's House during the credits sequence. Interestingly, the graphics for the Yoshi sign was overwritten with the English localization, but the berry graphics remain an older pre-release version. Additionally, the berries aren't all the same color.

31 July 1990 11:53

SMWleak doragon-house-demo.png


There are quite a bunch of tilemaps devoted to laying out both the staff roll at the end of the game and the enemy cast list.

End of Game

Japanese Version

Two tilemap exists for the Japanese version that covers the text that appears on screen just after rescuing Princess Peach. Both of these pieces of text differ from the final version!


02 September 1990 21:35


07 September 1990 17:14

Final Version
SMWleak v-ram-mode-7-final-kuppa.png SMWleak ending-obj-moji-new.png SMWfinal end-of-game.png
So then Princess Peach was rescued safely. Mario and co. hurry to Yo'ster Island with Yoshi and his friends where everyone was waiting for them... The same as the final version, but one of the letters in the word 'again' is replaced by a star symbol. With that, this adventure of Mario's is over. He joins in with Yoshi's friends and his fun vacation together with Peach begins again....

The first tilemap includes the entire a copy of the entire graphics bank below the placement of the text, probably to make it easier to lay out the text. It's also four rows long instead of three in later versions. The second tilemap has one character that is a star from the graphics bank.

International Version

ending-OBJ-NES.SCR holds the English version of this screen, and has no changes between this version and the final. However, the graphics data (v-ram-mode-7-final-kuppa-NES.CGX) that supplies this tilemap includes the entire alphabet for both the uppercase and lowercase versions, as well as a few symbols!

ending-OBJ-NES.SCR: 07 December 1990 16:43

v-ram-mode-7-final-kuppa-NES.CGX: 07 December 1990 15:01

SMWleak v-ram-mode-7-final-kuppa-NES.png

Interestingly, the way the text and symbols are laid out perfectly matches the font used for the message boxes and level titles!

Staff Roll

These tilemaps give much insight into the process of creating the staff roll for the credits.


This tilemap seems to merely lay out the role of each staff member for the game. Many of the differences between this first tilemap and the one in the final game are retained in later leaked tilemaps.

17 August 1990 18:15

SMWleak all-staff.png

The differences between the naming conventions in this tilemap and the final are:

  • The words "produser," "editer," and "directer" are spelled incorrectly, with an e as the final vowel instead of an o.
  • Hideki Konno's name is differently romanized in the final version as "Hidequi Konno."
  • Shigefumi Hino is credited as "Graphic Designer" instead of "C.G. Designer" in the final game.
  • Tatsunori Takakura is credited as "Satoru Takakura."
  • Yoshihiro Nomoto is credited as "Otsu Nomoto."
  • The special thanks section is titled "Thanks For" as opposed to "Special Thanks To" in the final game.
  • Takao Shimizu is credited as "Takaochan."


2bit-staff-roll.SCR and 2bit-staff-roll-komoji.SCR, two identical tilemaps, contain the first laid out staff roll graphics.

21 August 1990 11:51

SMWleak 2bit-staff-roll.png

Compared to the final version, this one doesn't use special palettes for the developers' positions on the team, instead using white for everything. Additionally, the small 8x8 text is used for the names instead of 16x8 tall letters in the final game. Additionally, Miyamoto's and Yamauchi's names are present at the beginning of the staff roll instead of at the very end, with Miyamoto's title in red instead of green. Koji Kondo is credited before Shigefumi Hino; this is flipped in the final versions.


22 August 1990 19:31

SMWleak 2bit-staff-roll-ohmoji.png

These credits look quite similar to the final, using palettes for the names and a two-tall for the names. It also contains the entire alphabet written out in both font sizes beyond the last developer's name. Also "Back Ground Programer" has been made plural, though other positions with multiple people involved have not been pluralized and this change was reverted for the final.


2bit-staff-roll-typography.SCR and 2bit-staff-roll-typography-nami.SCR are two very strange tilemaps. Instead of using just the 8x8 font or just the 16x16 font for the developers' names, these tilemaps use both, and which letters are "capitalized" is not even consistent between the two tilemaps. Very strange, indeed.


23 August 1990 00:39


23 August 1990 14:35

SMWleak 2bit-staff-roll-typography.png SMWleak 2bit-staff-roll-typography-nami.png

Besides the differences in capitalization, the nami version has greater spacing.


2bit-staff-roll-new.SCR and 2bit-staff-roll-new-1.SCR take the concept of completely unreadable capitalization and furthers it with yet another two tilesets that do this. The "1" version is the second page, as the first page is offset by a quarter like in the final version, presumably to delay the start of the credits appearing on the screen.

2bit-staff-roll-new.SCR (page 1)

24 August 1990 18:24

2bit-staff-roll-new-1.SCR (page 2)

24 August 1990 18:23

SMWleak 2bit-staff-roll-new.png SMWleak 2bit-staff-roll-new-1.png

Information about the "Produser" and "Executive Produser" have been moved to the end of the credits like in the final game, but Miyamoto's name is included a second time between Koji Kondo and Shigefumi Hino, because he was just that important to Super Mario World.


This version of the credits is nearly the final Japanese version with some spelling mistakes and alignment issues still persisting.

2bit-staff-roll-new-big.SCR (page 1)

03 September 1990 13:13

2bit-staff-roll-new-1-big.SCR (page 2)

03 September 1990 13:12

SMWleak 2bit-staff-roll-new-big.png SMWleak 2bit-staff-roll-new-1-big.png

International Localization

Two versions of the credits exist for the purpose of the English localization of the game.

2bit-staff-roll-NES-0.SCR and 2bit-staff-roll-NES-1.SCR show many of the changes to the developers' titles that occur between the Japanese and International versions, as well of the addition of localizer Dayv Brooks (misspelled Dayv Brocks!).

2bit-staff-roll-NES-0.SCR (page 1)

17 December 1990 19:51

2bit-staff-roll-NES-1.SCR (page 2)

17 December 1990 19:51

SMWleak 2bit-staff-roll-NES-0.png SMWleak 2bit-staff-roll-NES-1.png

As in earlier version of the Japanese credits, Hideki Konno's name is still romanized differently, and "Produser" is still spelled incorrectly. Koji Kondo and Shigefumi Hino's names are still flipped. This is the only version of the tilemap where Miyamoto's title is written in green.

2bit-staff-roll-NES-0-LINE.SCR and 2bit-staff-roll-NES-1-LINE.SCR attempt to center the names of the developers by lining up the spaces between first name and last name vertically.

2bit-staff-roll-NES-0-LINE.SCR (page 1)

07 December 1990 16:42

2bit-staff-roll-NES-1-LINE.SCR (page 2)

13 December 1990 14:06

SMWleak 2bit-staff-roll-NES-0-LINE.png SMWleak 2bit-staff-roll-NES-1-LINE.png

Enemy Cast List

A lot of information about scrapped names and other text are located outside the standard "window" resolution of the SNES, giving insight to possible alternate names for the enemies in both the Japanese and International versions of the game.

Japanese Version

Japan Early Japan Relevant File International
Para Bom Para-Bom 2bit-ending-TEKI-W.SCR*

28 August 1990 22:16

Para Kuri Para-Kuri Para-Goomba
Fishing Jugem Fishing-Jugem Fishin' Lakitu
Jugem Same as final Lakitu
Togezo Same as final Spiny
Bomhei Same as final Bob-omb
Hanachan Same as final Wiggler
Appare Unknown Amazing Flyin' Hammer Brother
Mantogame Super Koopa
P-Pakkun Jumping Piranha Plant
Ponkey Volcano Lotus
Bul Chargin' Chuck
K.K. Same as final 2bit-ending-TEKI-1D-1.SCR

26 August 1990 12:56

Sumo Brother
Sanbo Same as final Pokey
Choropoo Same as final Monty Mole
Killer Same as final Bullet Bill
Dorabon Same as final 2bit-ending-TEKI-big-killer.SCR

28 August 1990 22:20

Indy Same as final Mega Mole
Magnum Killer Same as final Banzai Bill
Raita Same as final 2bit-ending-TEKI-nakagon.SCR

28 August 1990 20:33

Chibi Raita Same as final Dino-Torch
Noko Noko Same as final Koopas
Togemet Same as final 2bit-ending-TEKI-chika.SCR

28 August 1990 20:54

Spike Top
Basa Basa Same as final Swoopers
Met Same as final Buzzy Beetle
Unbaba Same as final Blargg
Buku Buku Buku-Buku 2bit-ending-TEKI-sea.SCR

28 August 1990 21:23

Fugumannen Same as final Porcu-Puffer
Unira Same as final Urchin
Gooska Same as final Rip Van Fish
Torpedo Same as final Torpedo Ted
Telesa Same as final 2bit-ending-TEKI-ghost.SCR

26 August 1990 20:14

"Boo" Buddies
Spook Fishing Telesa Fishin' Boo
Atomic Telesa Same as final The Big "Boo"
Telesauls Same as final Eeries
Karon Same as final 2bit-ending-TEKI-shiro.SCR

26 August 1990 22:42

Dry Bones
Keseran Same as final Lil Sparky
Hone Met Same as final Bony Beetle
Pasaran Same as final Hothead
Dosun Same as final Thwomp
Coton Pochon Thwimp
Guru Guru Gulu-Gulu 2bit-ending-TEKI-shiro-2.SCR

28 August 1990 22:03

Ball 'N' Chain
Gari Gari Gari-Gari Grinder
Fish Bone Same as final Fishbone
Buibui Same as final 2bit-ending-TEKI-kaiten-boss.SCR

27 August 1990 11:21

Meka Koopa Mechakoops 2bit-ending-TEKI-meka-kuppa.SCR

14 December 1990 11:16

Morton Same as final 2bit-ending-TEKI-kuppa-family.SCR

30 August 1990 20:55

Morton Koopa Jr.
Lemmy Same as final Lemmy Koopa
Iggy Same as final Iggy Koopa
Ludwig Same as final Ludwig Von Koopa
Larry Same as final Larry Koopa
Wendy Same as final Wendy O. Koopa
Roy Same as final Roy Koopa
Koopa Kuppa Bowser

* Also found in 2bit-ending-chijo-W.SCR.

† The tilemap for this should be 2bit-ending-TEKI-1D-0.SCR, however it has been erroneously overwritten with data for the English localization.

Within the various tilemap files mentioned above for use laying out the Japanese credits, enemy names are also written outside the viewable window of the SNES. While many are on this list, there are some versions of the enemies' names that are similar, and others that don't show up at all. In no particular order:

  • P-Hat
  • Fire Dragon
  • Manta
  • Bat Boy
  • Magnam-Killer
  • Soukun
  • Munya Munya
  • Chiku Chiku
  • Water Killer
  • Iggy Koopa
  • Morton Kuppa Jr.
  • Kuppa Wendy

English Localization

The early english localization is mostly the same as the final, the most notable differences being the Koopalings lacking their last names and the names of the "Boo" Buddies and The Big "Boo", which were originally puns on famous musician Bo Diddley and former professional baseball/American football player Bo Jackson, respectively. "Boo" Diddly was also the name given to the boos in the english Super Mario Bros. 3 manual.

International Early International Relevant File
Para-Bomb Same as final 2bit-ending-TEKI-W-NES.SCR

13 December 1990 16:13

Para-Goomba Same as final
Fishin' Lakitu Same as final
Lakitu Same as final
Spiny Same as final
Bob-omb Same as final
Wiggler Same as final
Amazing Flyin' Hammer Brother Same as final 2bit-ending-TEKI-1D-0-NES.SCR

13 December 1990 15:16

Super Koopa Same as final
Jumping Piranha Plant Same as final
Volcano Lotus Same as final
Chargin' Chuck Same as final
Sumo Brother Same as final 2bit-ending-TEKI-1D-1-NES.SCR

13 December 1990 15:01

Pokey Same as final
Monty Mole Same as final
Bullet Bill Same as final
Rex Same as final 2bit-ending-TEKI-big-killer-NES.SCR

13 December 1990 16:35

Mega Mole Same as final
Banzai Bill Same as final
Dino-Rhino Same as final 2bit-ending-TEKI-nakagon-NES.SCR

13 December 1990 16:55

Dino-Torch Same as final
Koopas Same as final
Spike Top Same as final 2bit-ending-TEKI-chika-NES.SCR

14 December 1990 11:00

Swoopers Same as final
Buzzy Beetle Same as final
Blargg Same as final
Blurps Same as final 2bit-ending-TEKI-sea-NES.SCR

14 December 1990 10:11

Porcu-Puffer Same as final
Urchin Same as final
Rip Van Fish Same as final
Torpedo Ted Same as final
"Boo" Buddies "Boo" Diddly 2bit-ending-TEKI-ghost-NES.SCR

13 December 1990 17:20

Fishin' Boo Same as final
The Big "Boo" "Boo" Jackson
Eeries Same as final
Dry Bones Same as final 2bit-ending-TEKI-shiro-NES.SCR

14 December 1990 11:33

Lil Sparky Same as final
Bony Beetle Same as final
Hothead Same as final
Thwomp Same as final
Thwimp Same as final
Ball 'N' Chain Ball "N" Chain 2bit-ending-TEKI-shiro-2-NES.SCR

14 December 1990 11:47

Grinder Same as final
Fishbone Same as final
Reznor Unknown Unknown
Mechakoopas Same as final 2bit-ending-TEKI-meka-kuppa-NES.SCR

14 December 1990 11:19

Morton Koopa Jr. Morton 2bit-ending-TEKI-kuppa-family-NES.SCR

02 April 1991 14:21

Lemmy Koopa Lemmy
Iggy Koopa Iggy
Ludwig Von Koopa Ludwig
Larry Koopa Larry
Wendy O. Koopa Wendy
Roy Koopa Roy
Bowser Same as final

Foreground Scene

The foreground scene in the enemy cast list covers the layer 1 level tiles that the enemies being shown stand on during the credits sequence scrolling by. There are a few minor differences between the scenes in the tilemaps and the respective scene in the final game, as well as a lot of data outside the standard resolution of the SNES on the tilemap.

ending-TEKI-W.SCR uses a green pipe instead of the white pipe in the final.

13 December 1990 16:03

SMWleak ending-TEKI-W.png

ending-TEKI-1D-0.SCR uses a green pipe instead of an orange pipe, and also uses a different foreground palette and background color.

13 December 1990 14:04

SMWleak ending-TEKI-1D-0.png

ending-TEKI-1D-1.SCR uses a different palette rather than the default green of the final game. The platform made of turn blocks in the top left is replaced with used brown blocks in the final game, and the holes that the Monty Moles leave is not present in the final game either.

13 December 1990 15:01

SMWleak ending-TEKI-1D-1.png

The following four scenes are identical to that of the final game. Below the main foreground scene is a copy of the foreground of ending-TEKI-1D-1 without any blocks removed!


13 December 1990 16:35


13 December 1990 16:52


14 December 1990 11:00


14 December 1990 10:11

SMWleak ending-TEKI-big-killer.png SMWleak ending-TEKI-nakagon.png SMWleak ending-TEKI-chika.png SMWleak ending-TEKI-sea.png

ending-TEKI-ghost.SCR contains tileset data outside the SNES window that is the same as earlier tileset layouts for the ghost house.

SMWleak ending-TEKI-ghost.png

ending-TEKI-shiro.SCR and ending-TEKI-shiro-2.SCR are two castle tilemaps. The second tilemap uses a different palette from the final game, alternatively using the same palette as in the first one.


14 December 1990 11:33


14 December 1990 11:48

SMWleak ending-TEKI-shiro.png SMWleak ending-TEKI-shiro-2.png

ending-TEKI-kaiten-boss.SCR and ending-TEKI-meka-kuppa.SCR are tilemaps for Reznor and Mechakoopa respectively. The associated palette files make the lava much more red, and the palette for the ground tiles in the Mechakoopa design are again different from the final, with a lighter color scheme used. The yellow platform for Reznor also doesn't have any railing tiles in this version.


27 August 1990 11:21


14 December 1990 11:17

SMWleak ending-TEKI-kaiten-boss.png SMWleak ending-TEKI-meka-kuppa.png

Miscellaneous Content

This section features various tilemaps that don't quite fit the sections above.


As the title suggests, this tilemap has a big block, five in fact, with one at the top left and four in a row at the bottom. The name and associated graphics file may suggest this platform may have been used in mode 7 with rotation effects.

02 July 1990 10:03

SMWleak mode-7-big-block.png


This tilemap creates a forward facing image of Bowser's face out of 8x8 tiles. Based on the title "kaiten" this may have been rotated with mode 7.

03 July 1990 15:48

SMWleak mode-7-kaiten-kuppa-block.png


No graphics exist for this tilemap, but based on tile property patterns, a general silhouette of the item, and the filename (檻, ori, means cage) it turns out this is the tilemap for the unused flying cage that still exists in the final game!


20 November 1989 19:00

Final Game
SMWleak ori.png SMWCage.png

Applying the graphics bank that covers Layer 3 objects in the tilemap viewer will give the glitched appearance that is still reproducible in the final game. While the bars for the cage apparently haven't been overwritten, the walls and whatever is holding up the cage has been overwritten by pieces of layer 3 backgrounds and other objects.


There's a set of baseball-themed graphics and tilemaps. This was likely not intended for Super Mario World in any way, and there's no baseball game on the SNES that looks like this, leaving the reason for these graphics' existence a mystery.


03 October 1990 16:32


01 October 1990 17:19

SMWleak baseball-CGX.png SMWleak baseball-SCR.png


This tilemap comprises graphics for an absolutely massive Banzai Bill that would have taken up almost the entire screen! From the sheer complexity of the graphics, it may have been intended to be on Layer 3 like other features, such as the moving tides from Mondo. The idea of a screen-sized Banzai Bill would eventually be incorporated in New Super Mario Bros. Wii as the King Bill enemy.

31 July 1990 11:28

SMWleak killer.png


This tilemap has one central Bowser in his clown car in the center of the screen. The clown car features no propeller on the bottom, however.

12 April 1991 17:56

SMWleak bowser.png


This tilemap has several animation frames of Bowser in his clown car.

28 June 1990 19:19

SMWleak kuppa.png

The bottom right two frames are unused in the final game, as the clown car makes a sad face when Bowser is hit. Also, the clown cars don't have any propellers.


This tilemap lays out a series of unknown egg(?)-shaped things that cover the top right quadrant.

20 December 1989 20:47

SMWleak chika-ten.png


This tilemap has what appears to be either dripping lava or some sort of Layer 3 stalactites in different parts. Because of the layout of the tilemap, the left side renders properly when v-ram-chika.CGX is supplied, and the right side renders properly when v-ram-chika-2.CGX is supplied.


11 January 1990 14:18

with v-ram-chika.CGX

03 September 1990 23:12

with v-ram-chika-2.CGX

20 June 1990 09:30

SMWleak chika-obj-bg-chika.png SMWleak chika-obj-bg-chika-2.png


This tilemap looks the same in spirit to the other chika tilemaps above, although the tilemap refers to blank files when given the standard underground graphics. The blank region in the bottom right probably makes it a background remapped for import into the game like other background tilemaps.

19 July 1989 16:54

SMWleak chika-sel3.png


This is the layer 3 water that is used in levels like Yoshi's Island 4 and Mondo.

03 August 1990 13:54

SMWleak 2bit-sea.png

Layer 3 Crushers

There are three Layer 3 crusher tilesets.


This tilemap contains the only correct-looking crusher, although the one on the right is using the wrong graphics from the graphics bank.

04 July 1990 10:55

SMWleak 2bit-kui.png


This tilemap has the incorrect copy of the crusher from 2bit-kui.SCR duplicated on the left side too.

17 November 1989 22:01

SMWleak ashi.png


Both of the incorrect versions of the crusher use two different patterns of graphics, neither of which are correct, along with the incorrect palette. Also, the bottom of the crusher isn't wider than the rest of the structure.

17 November 1989 18:05

SMWleak kui.png


A tilemap entirely filled by tile F8, which in standard level tilesets is a solid transparent color.


A tilemap entirely filled by tile 17D, but with rotations. Given its name, it seems like v-ram-bul-kuppa.CGX should work, but it doesn't.

01 November 1989 17:15

SMWleak X-bul-kuppa.png


This tilemap by the name should lay out slope tiles for the castle tileset, but instead doesn't lay out much in the center section of the screen when supplied with castle graphics. The patterned rectangles refer to sprite graphics and not tile graphics.

23 June 1989 13:07

SMWleak shiro-saka.png


A tilemap with a blank background and dots which should presumably be raindrops, though no graphics fit the tiles referred to by the tilemap.

30 September 1989 22:43

SMWleak map-rain.png


A strange tilemap laying out large rectangles made out of tiles, graphics unknown.

29 August 1990 17:01

SMWleak 1-CLR.png


A tilemap for some kind of menu. When supplied with relevant Japanese text graphics, the tilemap displays garbage text.

14 December 1989 16:13

SMWleak select-1Q-1R.png


A patterned background (?) and some formation in the bottom left, with 16x16 tiles strewn about. This tilemap refers to graphics in the position of sprite graphics when a standard graphic set is given. The actual graphics file used for this is unknown.

16 August 1989 19:13

SMWleak check.png


A tilemap file apparently used to lay out graphics for the player. Normally, tilemaps aren't used for sprite layouts as object files can accomplish the same task more efficiently given the enhanced display abilities for objects on the SNES.

05 April 1989 17:07

SMWleak mario1-C.png


A tilemap displaying the Koopalings. Based on the uniform layout of each of the Koopalings, all seven may have been intended to display similarly, rather than the diverse body shapes in the final game. A corrected graphics set for this is unknown.

21 September 1989 14:24

SMWleak kokuppa.png