11-label-comparison.md

11 — Superfrog against The Speris Legacy

The reason to do this disc. Superfrog and The Speris Legacy are the two Team 17 titles in the set, and they are the third same-publisher pair after Guildhall (Legends / Gloom, which agree on nothing) and Core Design (Dragonstone / Banshee / Universe, which agree on three things that all turn out to be wider than the label).

Open item 14 asks whether build practice belongs to the studio, the label, or neither. Two predictions were written before either was measured, deliberately pointing in opposite directions, so that the disc could not confirm a story whichever way it fell.

The two discs

	The Speris Legacy	Superfrog
Developer	Binary Emotions	Team 17 (in-house)
Publisher	Team 17	Team 17
Master cut	1996-01-10 20:47:52	1994-08-24 12:17:15
Tracks	1 data, no audio	1 data, no audio
Image	2,455 sectors, volume declares 2,303	4,194 sectors, volume declares 4,164
Files / directories	47 / 10	221 / 22
Disc occupancy	0.74 %	1.26 %
Games on the disc	1	3 (one game, two demos)

The comparison table

Every row measured on both discs, not inferred.

	Speris	Superfrog	agree?
Cruncher	Imploder IMP!, 35/47 files, 52.8 %	Imploder ATN!, 96/221 files, 43.4 %	yes
Nesting depth	1	2	no
.TM block	22,152 B of cdtv.device	2,048 B, the standard three SHA-1s	no
Preparer field	empty name, tool signature only	Kenny Grant - ISOCD 1.04 …	no
Trailing unclaimed run	32	32	yes
Mastering tool	ISOCD 1.04	ISOCD 1.04	yes
SetPatch	—	40.16 (14.2.94), 13,484 B	—
freeanim.library	opened	opened, by 3 binaries	yes
FreeAnim as a file	no	no	yes
Save system	nonvolatile.library, plus floppy save-disk code	none — library ships, nothing opens it	no
Akiko	not touched	not touched (0/0/0/0)	yes
Colour depth	AGA, 24-bit palettes, six bitplanes	five bitplanes, 32 colours, BPLCON3 at defaults, FMODE never written	no
Palette calls	—	neither LoadRGB4 nor LoadRGB32 in any game binary	—
Music format	12 ProTracker modules	26 The Player 4.1A modules	no
Music inside executables	yes, 4 of the 12	yes, 8 of the 26	yes
Text encoding	30-character line records	ISO 8859-1 plain-text scripts, 4 languages	no
Copy protection	manual code table, codes computed	none	no
Bytes on disc / resident	4,514,540 / 8,543,154	4,102,748 / 13,336,690 (game only)	—
Expansion factor	1.89×	3.25×	no
Floppy ancestry	conversion of the floppy release	three-floppy layout recoverable from the binary	yes

What the two predictions were, and how they landed

P3 — the cruncher follows the label. Written before any magic was read:

After three Core Design discs the only surviving same-label regularity is the cruncher: Core Design does RNC ProPack 1 three times out of three, against five different crunchers in the other studios. Speris uses the Imploder. Prediction: Superfrog compresses with the Imploder.

Correct. And note what makes it more than a coincidence of sample size: the two Team 17 discs were built by different studios — Binary Emotions and Team 17 in-house — two years apart, in different genres. The Core Design run is one studio three times. This is one label, two studios. The reading "the cruncher follows the label" now spans two labels instead of one, and one of them has cross-studio evidence.

The corroboration is stronger than magic-matching alone: the decruncher transcribed from Speris's own shipped 68000 routine (tools/speris_imp.py, reused here byte-for-byte with no modification) decodes all 96 of Superfrog's files exactly, in-place pointer landing on zero every time. Same packer, same default tables, same 38-byte tail.

P10 — the .TM follows the tool. Written at the same time, pointing the other way:

Twelve of thirteen CD32-era discs share three SHA-1s; the thirteenth is Speris. Prediction: Superfrog matches the twelve — i.e. the .TM follows the TOOL, not the label.

Correct. Superfrog's block is c5ffcef2… / 8d841151… / 690aae24…, the standard trio.

The answer to open item 14

The question's shape is wrong, and this disc is what shows it.

"Does build practice belong to the studio, the label, or neither" presumes a single owner. It does not have one. Different artefacts have different owners, and the owner is readable per artefact:

artefact	owner	why, on this disc
the cruncher	the label	two Team 17 discs, two studios, two years, same packer; five other studios, five other crunchers
the .TM block	the mastering tool	twelve discs, ten labels, identical; the one Team 17 disc cut differently differs
the preparer field	the operator	a third name pattern here, and he is a different person from the game's authors
the trailing run	the operator	232 for D J Pocock on 5 of 5, 32 for everyone else on 11 of 11
colour depth, save system, music format	the studio	Speris and Superfrog disagree on all three

And this disc supplies the mechanism, which is why the split is not arbitrary. The preparer field names Kenny Grant, and s/T17Menu's $VER: names Kenny G as the author of the front-end menu. The person who cut the master is a Team 17 person who wrote the compilation shell — and he is not the person who wrote Superfrog. So the artefacts produced by the mastering step (.TM block, preparer, trailing run, ISO layout) and the artefacts produced by the game build (cruncher, display setup, save system, music format) were literally made by different hands, on this disc, verifiably.

That is why they can disagree, and it is why "studio or label" was never going to have one answer.

The rows where they disagree, and what they say

The three disagreements that are not about mastering are all studio signatures, and they line up with the fact that the two games were written by different teams:

• Colour. Speris is AGA with 24-bit palettes and six bitplanes. Superfrog is five bitplanes and 32 colours, with BPLCON3 written only to safe defaults and FMODE never written at all — a display an ECS machine can show, on a disc whose volume is named SuperFrog_AGA. The 1993 game was not rewritten for AGA; it was recompiled for CD32.
• Saving. Speris uses nonvolatile.library and keeps the floppy save-disk code. Superfrog ships nonvolatile.library and no binary on the disc opens it, and no binary loads $00B80030 either. Two independent measurements, same answer: Superfrog on CD32 does not save.
• Music. Speris is ProTracker. Superfrog is The Player 4.1A — a packed replayer format that a M.K. scan cannot see, and a fifth answer to section 8's "where is the music" question (08-audio.md).

One row that agrees for a reason worth keeping

freeanim.library is opened on both discs, and on neither is there a FreeAnim file. The checklist's FreeAnim file lead — the 3,492-byte file with SHA-1 449c610071ace58d… on Liberation, Universe, Dragonstone and Gloom — is a separate thing from the ROM library of nearly the same name, and this pair is a clean illustration: two discs that use the library, zero that carry the file.