The Nintendo Gigaleak preserves a separate SimCity SNES art workspace under other/NEWS/テープリストア/NEWS_04/home/sugiyama/SIM.
Compared with the much broader Mario Kart CAR folder, SIM is smaller and more focused.
It is almost entirely menu, panel, input, and city-interface material, with a distinctive .SFX side format that appears to preserve tool metadata from the asset pipeline itself.
At the top level the SIM branch contains 85 entries: 84 files and one subdirectory called is.
Even before looking inside the files, the naming makes the workspace feel tightly scoped around SimCity SNES front-end and in-game UI production:
Unlike the flatter Mario Kart art branch, SIM has both a clean top-level asset set and a second is/ work area that looks like an internal staging or editor-side directory.
The top-level file types break down like this:
| Type | Count | What it preserves |
|---|---|---|
.SCR |
15 | Screen and panel layout data |
.CGX |
17 | Character and tile graphics banks |
.COL |
3 | Palette data |
.OBJ |
3 | Object or object-text placement data |
.SFX |
9 | Tool-side metadata files with embedded version strings |
.BAK |
37 | Backup revisions of the current asset files |
This is not a random pile of graphics exports. It is a consistent UI-production branch with layout, graphics, palette, object, and tool-side files all surviving together.
This page is based on Gigaleak NEWS_04 home/sugiyama/SIM/, part of Tadashi Sugiyama’s wider multi-project workspace.
The project identification is straightforward from the filenames:
SELECT-SCENARIOMAP-SELECTTOWNLEVELINPUTSAVEDThose names line up very cleanly with the Super Famicom SimCity interface rather than with gameplay scenery, world art, or character animation.
S-CG-CAD version stringSIM that appears to preserve a second staging or editor-side work areaThis workspace preserves several different layers of SimCity SNES’s visual production at once:
.SFX files paired with major UI familiesis/ work area with many related but non-identical assetsThat combination makes SIM more than just a texture dump.
It preserves both the visible UI assets and a surprisingly clear trail of how Nintendo’s SimCity art tools were organizing them.
The SIM branch uses the same broad asset vocabulary seen in other NEWS_04 art folders, but in a much more focused UI-only mix.
All top-level .SCR files are exactly 8,960 bytes, which strongly suggests a fixed editor-side layout format rather than arbitrary bitmap exports.
A few openings make that clear:
| File | First 16-bit words |
|---|---|
SELECT-SCENARIO.SCR |
0x0020, 0x0021, 0x0022, 0x0023 … |
MAP-SELECT.SCR |
0xFFFF, 0x0800, 0x0801, 0x0802 … |
TOWN.SCR |
0x0601, 0x0601, 0x0601, 0x0601 … |
BG2bit.SCR |
0x0080, 0x0081, 0x0082, 0x0083 … |
That pattern fits very well with structured 16-bit layout entries rather than raw graphics.
The reuse is also revealing.
SELECT-SCENARIO.SCR and INPUT1.SCR share the same opening run before diverging at byte 257, while MAP-SELECT.SCR and INPUT2-KEY.SCR share the same opening pattern before diverging at byte 197.
That makes the input screens look less like unrelated one-offs and more like screens built from shared menu templates.
The three top-level .OBJ files are all 13,568 bytes:
INPUT.OBJMOJI.OBJSCENARIO.OBJThey are clearly not graphics banks. Their first words break into repeated small numeric groups much more like object or placement records than tile art.
MOJI.OBJ and SCENARIO.OBJ are especially interesting because they sit beside text-facing assets like KANJI.CGX and OBJ-MOJI.CGX.
That makes them look like object-form text or UI layout data rather than generic sprite lists.
The graphics and palette side is also very regular:
| Type | Sizes seen | Reading |
|---|---|---|
.COL |
1,024 bytes |
palette tables |
.CGX |
17,664 and 34,048 bytes |
fixed-size tile graphics banks |
The smaller 17,664 byte banks tend to show up in utility or low-level graphics sets like BG2bit.CGX and KAGE-G.CGX, while the larger 34,048 byte banks dominate the visible UI layer through INPUT-BG.CGX, KANJI.CGX, OBJ-MOJI.CGX, icon_p.CGX, iconpd.CGX, optishd.CGX, and SAVED.CGX.
That gives the whole branch a very editor-friendly shape:
.SCR for layout.OBJ for object-side placement.COL for color tables.CGX for tile and text graphicsThe most distinctive technical artifact in SIM is the .SFX format.
All nine .SFX files are exactly 2,048 bytes, and unlike the earlier stub page’s guesswork, they do not look like vague runtime “effects” files.
They begin with clear ASCII headers:
| File | Header text |
|---|---|
SELECT-SCENARIO.SFX |
NAK1989 S-CG-CADVer1.21 900611 |
MAP-SELECT.SFX |
NAK1989 S-CG-CADVer1.21 900611 |
SCENARIO.SFX |
NAK1989 S-CG-CADVer1.21 900611 |
SELECT.SFX |
NAK1989 S-CG-CADVer1.21 900611 |
LEVEL.SFX |
NAK1989 S-CG-CADVer1.21 900611 |
INPUT.SFX |
NAK1989 S-CG-CADVer1.13 900424 |
BG2bit.SFX |
NAK1989 S-CG-CADVer1.13 900424 |
TOWN.SFX |
NAK1989 S-CG-CADVer1.23 901226 |
That is one of the clearest pipeline clues in the whole NEWS_04 restore.
These files preserve the signature of a dedicated S-CG-CAD tool, complete with version numbers and build dates embedded directly in the sidecar files.
The wider NEWS_04 leak backs that interpretation up.
Outside SIM, Nintendo’s workstation folders also preserve .CAD_SRD directories and CAD.sfx_main.LST manifests that point at /usr/local/srd/cad/sfc/sfx_main.hex, which makes the S-CG-CAD naming look like a real SRD toolchain component rather than a one-off SimCity quirk.
The payloads are not identical either.
The first 48 bytes stay the same within a tool version family, but the rest of each file hashes differently.
So the .SFX files look like per-screen metadata produced by the same graphics tool, not like generic empty headers stamped onto unrelated files.
The version spread is useful too:
Ver1.13 appears on INPUT and BG2bitVer1.21 appears on most of the main menu familiesVer1.23 appears on TOWNThat suggests the SIM branch captures more than one moment in the UI toolchain rather than one single export pass.
The body of the .SFX files is also informative.
After the shared header, most of the screens only carry a small amount of nonzero data:
| File | Nonzero bytes after header | Reading |
|---|---|---|
BG2bit.SFX |
37 | compact metadata block |
SELECT-SCENARIO.SFX |
38 | compact metadata block |
SCENARIO.SFX |
39 | compact metadata block |
SELECT.SFX |
39 | compact metadata block |
INPUT-KEY.SFX |
39 | compact metadata block |
INPUT.SFX |
40 | compact metadata block |
MAP-SELECT.SFX |
42 | compact metadata block |
LEVEL.SFX |
44 | compact metadata block |
TOWN.SFX |
918 | much denser metadata payload |
That makes TOWN.SFX stand out immediately.
It is not just a newer header on the same kind of tiny sidecar.
It appears to preserve a much richer payload, which may reflect a more complex city-status screen or a later, more featureful version of the same authoring tool.
Even the compact files still differ from each other beyond the header. So the safest reading is that these are screen-specific tool records, not one shared stub copied across the whole branch.
TOWN.SFX deserves special attention because it behaves differently from the rest of the set.
The first thing it shares with the other files is the same overall structure:
48-byte ASCII tool header0x100But the density changes dramatically after that point.
| Detail | TOWN.SFX |
Typical Ver1.21 file |
|---|---|---|
| Size | 2,048 bytes |
2,048 bytes |
| Header | S-CG-CADVer1.23 901226 |
S-CG-CADVer1.21 900611 |
| Nonzero bytes after header | 918 |
roughly 38 to 44 |
| Nonzero ranges after header | 67 |
around 30 to 34 |
So TOWN.SFX is not just a minor variant.
It looks like the same file type carrying much more real data.
The early payload still looks familiar.
Like the other .SFX files, it begins with tiny control values such as 0x0001 and 0x0100, which suggests simple flags or counters rather than text or image data.
What changes is that TOWN.SFX keeps going.
Instead of collapsing back into near-empty space, it preserves many more populated ranges across the rest of the file.
The safest interpretation is that TOWN.SFX records a richer screen configuration than the other top-level UI sidecars.
That could mean more windows, more panel regions, more tool annotations, or simply a later tool version writing out more of its internal state.
Either way, TOWN.SFX is currently the best single candidate in the whole branch if you want to reverse the S-CG-CAD metadata format itself.
Looking at the file in 16-byte blocks makes the difference even clearer.
The early part of the payload still behaves like the other sidecars:
0x30, 0x40, and 0x500x1000x0001, 0x0100, and 0x0200But after that, TOWN.SFX develops two much larger active regions:
| Range | Behavior |
|---|---|
0x0200 to 0x02EF |
dense byte ramps and table-like data |
0x0320 to 0x03EF |
second dense region with long graded runs |
0x0400 to 0x05EF |
another structured block with many repeated low values and small control words |
That is very different from the other .SFX files, which mostly go quiet after the first setup fields.
The middle ranges especially look table-driven.
They contain long ordered byte runs like 0x49 0x48 0x24 0x48 ..., then broader descending and ascending sequences through values like 0x6b, 0x66, 0x63, 0x62, 0x74, 0x72, 0x4e, and then long bands in the 0x2b to 0x5c range.
That does not read like text. It looks much more like structured numeric data: screen parameters, lookup tables, curve data, or serialized region definitions from the CAD tool.
The repeated shape of the lower blocks is another useful clue.
The region around 0x0400 starts with:
0x4948, 0x4948, 0x4c4b0x0212, 0x0606, 0x0008, 0x0911Then the region around 0x0500 begins in a very similar way:
0x4948, 0x4948, 0x4c4b, 0x4b4dThat repeated pattern suggests TOWN.SFX may be storing more than one closely related configuration block, perhaps for multiple windows, states, or layered panel definitions inside the town interface.
So while we still cannot name the exact fields, the file is clearly more than “a richer sidecar.” It appears to be a multi-block serialized tool record, probably describing several interlocking pieces of the SimCity town screen rather than one flat parameter list.
The safest way to summarize the file now is as a set of regions rather than one opaque payload:
| Range | Working reading |
|---|---|
0x0000 to 0x002f |
ASCII S-CG-CAD header and version string |
0x0030 to 0x005f |
tiny setup flags, mostly 0x0001 and 0x0100 |
0x0100 to 0x015f |
denser setup area with short control tables |
0x0200 to 0x02ef |
first major dense payload block |
0x0320 to 0x03ef |
second dense payload block with long graded runs |
0x0400 to 0x04ef |
structured lower block, likely one screen-state definition |
0x0500 to 0x05ef |
second lower block with a very similar shape |
That does not prove the exact field names, but it does move the page beyond “this file is denser than the others.”
TOWN.SFX now looks like a multi-region serialized tool record with several linked configuration blocks inside it.
The top-level names cluster into a few clear interface groups.
The front-end entry path is the clearest part of the directory:
SELECT-SCENARIO.SCRSELECT-SCENARIO-2.SCRSELECT-SCENARIO.SFXSCENARIO.OBJSCENARIO.SFXSELECT.SCRSELECT.SFXThis looks like a layered scenario-selection stack rather than one finished menu screen. There are separate layout files, object-side data, and tool sidecars all living under the same naming family.
The next layer looks like city-state and region-facing UI:
MAP-SELECT.SCRMAP-SELECT2.SCRMAP-SELECT.SFXTOWN.SCRTOWN2.SCRTOWN.SFXLEVEL1.SCRLEVEL2.SCRLEVEL.SFXThis grouping is very menu-heavy and panel-heavy. It does not look like gameplay scenery or simulation tile art. It looks like the screens around choosing, viewing, and configuring city scenarios.
The input side is one of the most revealing parts of the whole branch:
INPUT-BG.CGXINPUT.OBJINPUT.SFXINPUT-KEY.SFXINPUT1.SCRINPUT2-KEY.SCRINPUT2-PUSH.SCRINPUT-LEVEL.SCRMOJI.OBJOBJ-MOJI.CGXKANJI.CGXThis is the point where the branch stops looking like “menu screens” in a generic sense and starts looking like an actual UI production environment. There are dedicated text graphics, object-form text layouts, background graphics, and multiple input-screen compositions all preserved together.
The text side is especially revealing because it preserves both graphics and placement data.
KANJI.CGX is a large 34,048-byte graphics bank, while OBJ-MOJI.CGX is another 34,048-byte text-facing bank with very different binary structure.
Together they look like the core reusable text and glyph resources for the UI.
The two banks also differ quite a bit in character:
| File | Size | Nonzero bytes | Unique byte values | Reading |
|---|---|---|---|---|
KANJI.CGX |
34,048 |
14,649 |
256 |
broad, dense character inventory |
OBJ-MOJI.CGX |
34,048 |
10,046 |
205 |
more UI-facing or object-text-focused bank |
So even though the files are the same size, they do not look like clones with different names.
KANJI.CGX appears to preserve the wider character set, while OBJ-MOJI.CGX looks more shaped around assembled UI text usage.
The object side is a little stranger:
MOJI.OBJ is 13,568 bytesSCENARIO.OBJ is also 13,568 bytesINPUT.OBJ is 13,568 bytesMOJI.OBJ and SCENARIO.OBJ are not just similar in shape.
They have the same hash and the same opening records, which suggests the same underlying object-layout container was being reused under different UI-facing names.
INPUT.OBJ, by contrast, opens with a different set of repeating 3-word groups and looks like a distinct placement set tailored to the input flow.
That split is clearer if you compare the first few record triples:
| File | First 3-word records | Reading |
|---|---|---|
MOJI.OBJ |
[0x0080, 0xCAB2, 0x3030], [0x0080, 0xCAAA, 0x2030], [0x0080, 0x18C0, 0x8C30] |
text-facing object layout |
SCENARIO.OBJ |
same opening triples as MOJI.OBJ |
reused or near-identical layout container |
INPUT.OBJ |
[0x0000, 0xE0E0, 0x7130], [0x0000, 0xE8E0, 0x6130], [0x0000, 0xF0E0, 0x7030] |
separate input-specific placement structure |
INPUT.OBJ also starts with more zero-led triples than the other two.
That makes it feel less like a generic text-placement file and more like a dedicated object-layout surface for the input workflow itself.
That gives the text layer a nice split:
KANJI.CGX and OBJ-MOJI.CGX as reusable graphics banksMOJI.OBJ / SCENARIO.OBJ / INPUT.OBJ as object-side placement or assembly data for those graphicsSo the branch is not only storing finished menu screens. It still preserves the lower-level machinery needed to build text-bearing UI panels out of reusable graphics and object records.
The remaining graphics banks look like reusable panel or HUD assets rather than one-screen specials:
icon_p.CGX, icon_p-F.CGX, icon_p-G.CGXiconpd.CGX, iconpd-F.CGX, iconpd-G.CGXoptishd.CGX, optishd-F.CGX, optishd-G.CGXSAVED.CGX, SAVED-F.CGXICON-GGG.CGXKAGE-G.CGXThe repeated -F and -G suffixes make these look like state or variant banks rather than separate unrelated graphics sets.
That is exactly the sort of naming you would expect in a menu-heavy simulation game with repeated panel states and icon modes.
The lone is/ directory is not just noise.
It contains 81 more files of its own:
| Type | Count in is/ |
|---|---|
.CGX |
25 |
.COL |
11 |
.SCR |
20 |
.BAK |
24 |
.old |
1 |
This is much more than a tiny support folder. It is effectively a second SIM asset workspace, with many lowercase or shortened names and several files that have no top-level equivalent at all:
menu.CGX, menu.SCR, menu.COLsim.CGXstart.SCR, start.COL, start-ma.SCRsamp.CGX, samp.COL, samp.SCR, samp1.SCRbank.CGX, bank.SCR, bank.COL, bank-ji.SCRcity.SCR, bud.SCR, eva.SCR, waku.SCR, rrr.SCREven where the names line up, the files are usually not identical. For example:
BG2bit.CGX, BG2bit.COL, and BG2bit.SCR all differ from is/bg2bit.*INPUT-BG.CGX differs from is/input-bg.CGXOBJ-MOJI.CGX differs from is/obj-moji.CGXicon_p.CGX differs from is/icon-p.CGXKANJI.CGX is 34,048 bytes, while is/kanji.CGX is 17,664 bytesThat makes is/ look much more like an internal staging or editor-side work area than a plain backup copy of the top level.
The is/ branch also has a stronger internal structure than the top level.
Its families cluster into a few clear groups:
| Family | Example files | Reading |
|---|---|---|
| Core menu banks | menu.CGX, menu.COL, menu.SCR |
generic menu composition set |
| Start and setup | start.SCR, start.COL, start-ma.SCR |
start-screen and entry-state assets |
| Bank or reference sets | bank.CGX, bank.COL, bank.SCR, bank-ji.SCR |
likely editor-side reference or staging screens |
| Map and city support | map-1.CGX, map-map.CGX, mapname.CGX, city.SCR, bud.SCR, eva.SCR |
city and map-support assets |
| Text banks | moji.CGX, mojiA.CGX, kanji.CGX, obj-moji.CGX |
alternate text and object-text resources |
| Icon and object banks | icon-p.CGX, micon.CGX, newobj.CGX, optish.CGX |
reusable UI and object-support sets |
| Samples and tests | samp.CGX, samp.COL, samp.SCR, samp1.SCR |
likely sample or test screens |
That makes is/ feel less like “the same files but lower case” and more like a hands-on work area where generic menu resources, sample screens, and alternate text/icon banks were being tried out alongside the named production assets at the top level.
The is/ directory is also where the deepest text and glyph material lives.
The key families are:
moji.CGX, moji.CGX.BAK, moji.CGX.oldmojiA.CGX, mojiA.CGX.BAKkanji.CGX, kanji.CGX.BAKobj-moji.CGXselect.CGX, select.CGX.BAKsel.CGX, sel.CGX.BAKbank-ji.SCRNIN.SCR and nin.SCR.BAKThat is already more varied than the top level.
Instead of one KANJI bank and one OBJ-MOJI bank, the staging area keeps multiple text-bank variants, dedicated select-screen banks, and at least one screen file whose name directly points at kanji handling.
One of the clearest patterns in is/ is the split between smaller 17,664 byte text banks and larger 34,048 byte UI banks.
| File | Size | Reading |
|---|---|---|
moji.CGX |
17,664 bytes |
compact general text bank |
mojiA.CGX |
17,664 bytes |
alternate compact text bank |
kanji.CGX |
17,664 bytes |
compact kanji-side bank |
obj-moji.CGX |
34,048 bytes |
larger object-text bank |
select.CGX |
34,048 bytes |
large select-screen text/UI bank |
sel.CGX |
34,048 bytes |
alternate select-side text/UI bank |
That looks like a real tier split inside the editor pipeline:
The top level hints at the same idea, but is/ makes it much clearer.
The three smaller text banks are not all doing the same job.
moji.CGX and mojiA.CGX are very close:
17,664 bytes2536,965 for moji and 5,804 for mojiAThat makes them look like two nearby revisions or variants of the same general text-bank family.
kanji.CGX is different much earlier.
It diverges from both moji and mojiA at byte 9, and its opening values are visibly less like the very sparse moji headers.
It also carries fewer unique byte values overall.
So the safest reading is:
moji and mojiA are sibling banks in the same general text tierkanji is a separate bank with a different glyph inventory and a different structural roleThat lines up nicely with the names. The smaller text tier does not just hold “font data” in the abstract. It appears to separate ordinary menu text from a more specialized kanji-side bank.
The revision pattern across the three banks is useful too.
moji.CGX differs from both its .BAK and .old formsmojiA.CGX differs from its .BAKkanji.CGX is identical to kanji.CGX.BAKThat makes kanji look comparatively settled, while the general-purpose moji banks were still moving.
For a UI-heavy project, that makes sense. The broader menu text banks would be more likely to keep changing as screens and labels were revised, while a larger kanji bank could stabilize earlier if it was acting as a more general reusable character source.
The text banks inside is/ also preserve a strong revision story:
moji.CGX differs from moji.CGX.BAK, with the first change at byte 2,361moji.CGX also differs from moji.CGX.old, with the first change at byte 253mojiA.CGX differs from mojiA.CGX.BAK, with the first change at byte 2,377select.CGX differs from select.CGX.BAK, with the first change at byte 16,427sel.CGX differs from sel.CGX.BAK, with the first change at byte 9,737That is a useful spread. Some changes happen quite early in the file, while others land much later, which suggests these were real edits to the bank contents rather than automatic timestamp churn or metadata noise.
kanji.CGX is the main stable exception here.
It is identical to kanji.CGX.BAK, which suggests at least one part of the text pipeline had already settled while other banks were still being revised.
The screen side adds two especially interesting outliers:
bank-ji.SCR at the standard 8,960 bytesNIN.SCR at a much larger 16,640 bytes, with nin.SCR.BAK surviving at 8,192 bytesbank-ji.SCR opens with repeated 0x03FF words, while NIN.SCR opens with repeated 0x02FF words.
Those are unusually rigid patterns even by the standards of the other .SCR files, which makes both files look like dedicated utility or special-purpose text screens rather than ordinary menu layouts.
The naming hints at two different roles:
bank-ji looks like a kanji-related bank or reference screenNIN may be a branding or Nintendo-specific presentation screenEven without a full format decode, they are strong evidence that is/ was carrying not just production screens, but also helper layouts and text-bank support surfaces for the UI toolchain itself.
The is/ folder has 24 .BAK files of its own plus one moji.CGX.old.
That is a nice clue that this was still an actively edited branch, not just a static mirror of the parent directory.
The most visibly revised groups inside is/ include:
bank.*kanji.CGXmoji.CGX and mojiA.CGXselect.CGXicon-*, micon-*, newobj-*, and optish-*sim.CGXThat combination reinforces the same general picture as the top level. This was a UI-production workspace with real iteration history, not a stripped-down art export folder.
The top-level revision trail is smaller than Mario Kart’s, but it is still very revealing.
There are 37 .BAK files at the top level, and they split like this:
| Extension | Count |
|---|---|
.SCR.BAK |
14 |
.CGX.BAK |
13 |
.SFX.BAK |
6 |
.OBJ.BAK |
3 |
.COL.BAK |
1 |
That pattern is useful.
The SIM branch was preserving not only layout and graphics revisions, but also tool-side .SFX revisions.
That strengthens the case that .SFX was part of the active authoring workflow rather than dead baggage.
The main revision clusters are easy to spot:
SELECT-SCENARIO*INPUT*MAP-SELECT*TOWN*LEVEL*icon_p*, iconpd*, and optishd*So the heaviest visible churn was concentrated in exactly the parts of the game where a simulation UI would need lots of tuning: scenario flow, input flow, town display, and reusable panel/icon banks.
The SIM branch is one of the cleanest UI-production artifacts in Sugiyama’s NEWS_04 workspace.
What makes it especially valuable is not just that it preserves SimCity SNES front-end graphics. It also preserves the structure of the tool pipeline around them:
.SFX sidecars with embedded S-CG-CAD tool versionsis/ work area with related but non-identical staging assets.BAK trail across both screens and tool-side filesThat makes this page a much better complement to source-code-oriented leak analysis.
Instead of showing how SimCity SNES was programmed, SIM shows how its screen-heavy interface was actually built, revised, and organized on the art side.