The Deep Sea Research Center

Story mode. Written for a first playthrough and kept free of spoilers — it will not tell you what happens here, only what the game is asking of you.

A structure in the far south-west of the map with no marker on it. You land the airship directly on top of it.

This is optional. It is also the hardest content in the game outside the final dungeon, and it holds two of the best things in it. Come back later if you are not ready — it stays open right through to the end.

The room with the pulsing core

The first room has a blue core that pulses. Move while it is dark. Stop while it is lit. Moving under the light triggers fights at a brutal rate.

And here is the part worth reading twice: the ability that stops random encounters does not work in this room. Two sources say so explicitly. Whatever you have been relying on all game, you are not relying on it here — timing the pulses is the only method.

Three questions

At the top you are asked three questions.

The first two are straightforward: you are not here to fight, and you will not beg.

The third one is a trap in the interface. The correct answer is not on the screen. There is a blank entry below the visible options — one source says press down three times to find it.

A dragon fight follows each answer. Several sources warn that the second of those dragons gets a free attack on you before you can act, which is where unprepared parties die. Heal between questions.

The descent is a budget

This is the part everyone describes wrongly, so here it is plainly:

There is no step counter. Nobody is counting your footsteps and nothing drains as you walk.

A terminal on the first level gives you twenty units of steam pressure. Doors cost four, two or one to open. The machine at the very bottom needs ten units left to work.

So it is a budget problem. Spend carefully, arrive with at least ten.

Two things nobody tells you and both are reassuring:

• You can move to a cheaper part of the level but never back to a dearer one. That is why a careless early spend cannot be walked off.
• The computer on the first level resets the whole puzzle. If you botch it, you are not stuck — go back up and start the budget again from twenty.

The offer near the bottom

At the last door, if you have a certain character with you and the right amount of pressure left, he offers to force it open for nothing.

This is not free. Taking it switches on fifteen unavoidable fights on the way down, and closes off a side room containing a draw point and some enemies you will not meet anywhere else.

Both routes end with exactly ten units, so this is not a budget decision — it is a choice about what you want to see. If you want the fights, take the offer. If you want the side room, pay the four.

What is at the bottom

A very large fight. It has no elemental weakness of any kind, so there is no clever answer — reduce its defences, buff yourself, and hit it.

Draw from it before it dies. There is a Guardian Force inside it, it is the strongest one in the game, and its draw list carries it at every level so there is no requirement beyond having Draw equipped and remembering to use it.

The strongest spell in the game is on the same list, and there are only two enemies in the whole game you can draw it from — this one and a fight in the final dungeon. Nobody writes that down. Draw until you are full.

One warning before you finish

Every fixed encounter in this dungeon stops existing the moment the last fight ends. Four sources agree.

If you came here for materials — and the fights on the way down are the practical source of a component for one of the ultimate weapons — farm them first. This is the most commonly cited regret about this place, and it is entirely avoidable.

One thing no guide mentions, because it is not visible from inside a playthrough: that farm pays four to eight times more once your party is above level 30. Same enemy, same room, and the quantity it hands over steps up at that line.

Game Notes

The step counter that does not exist.

Ask around about this dungeon and somebody will tell you it counts your steps and you have to move efficiently. It is one of those pieces of folklore that everyone half-remembers. It is also completely wrong: not one of the eight guides that document the descent describes a step count, because there is not one. There is a pool of twenty units and doors that cost money.

I find the mistake more interesting than the mechanic. Why does a pressure budget turn into a step counter in people’s memories?

I think because of the room upstairs. The first thing this dungeon teaches you is that moving is dangerous — the core pulses, you walk under the light, you get jumped. Ten minutes of that installs the idea that this place is about movement discipline. Then you go downstairs into a resource puzzle, and the lesson you already learned quietly rewrites what you are looking at.

Which is a good reminder that the wrong model does not have to come from bad information. It can come from correct information about the previous room.

Completionist mode. Full spoilers, in play order, with every permanent loss called out before you reach it. 12 entries below are marked where our sources disagree with each other — we have not picked a winner for you.

Spoilers from here.

The headline: the most argued HP figure in the game was never a disagreement

Ultima Weapon’s HP is published three ways:

All three are the same curve.

51,100 is level 1. 61,000 is level 10. 160,000 is level 100. The flat 100,000 sits on the curve somewhere in the high forties.

Nobody was wrong. Nobody was even slightly wrong. Five sources published the endpoints of the curve, one published an intermediate anchor without saying which level it belonged to, and one published a single observation with no level attached. Twenty-five years of players comparing those numbers and concluding somebody must be lying, and the three figures are three points on one line.

This is the cleanest demonstration of this project’s central claim that I expect to find. If there is one page to point at when explaining why this guide exists, it is this one.

Zero EXP and 100 AP — the highest single-fight AP award anywhere in the guide. It is immune to Gravity, carries the guaranteed-card flag, and has no elemental weakness of any kind: all eight elements at exactly 100%, which one database already publishes correctly. Its mug is one Three Stars and its drop is a hundred Ultima Stone, in every slot at every band.

Bahamut

10,800 at level 1, 18,000 at level 10, 34,000 at level 30, 90,000 at level 100. Settled, matching four sources exactly.

One wiki publishes 9,150 at level 100, which is about a ninetieth of the real figure. That page is describing the summon’s own stat block, not the boss — a different creature sharing a name, which by now is the most familiar shape in this entire project.

Zero EXP and 40 AP. Immune to Gravity. Carries the guaranteed-card flag, so its card is not a roll.

Correction: four of its published weaknesses do not exist

One database publishes weaknesses to Fire, Ice, Water and Holy, and immunities to Thunder, Earth and Wind.

The elemental table holds three entries and only three:

So: the two immunities that were reported are real. The Earth immunity is not. The four weaknesses are not — there is no elemental answer to this fight at all. And a Poison resistance nobody reports does exist.

That is four fabricated weaknesses, one fabricated immunity, and one unreported resistance in a single published table.

Loot

One source lists the same item as both the mug and the drop; one database lists it as the drop only. Both are satisfied — the record carries a Hyper Wrist in every mug slot and every drop slot, at every level band, so it genuinely is both and the level you fight it at changes nothing.

Its draw list is agreed by three sources and is Flare, Curaga, Full-life and Dispel, the same four in all three bands. Two sources urge stocking Full-life to a hundred here, which is sound — waiting for a higher band improves nothing, because there is no higher list.

The three questions

The first two answers refuse the fight and refuse to beg. The third correct answer is a blank entry below the visible options — one source says three presses of down.

Three sources give one wording for that hidden answer and one gives another. Two sources say a wrong answer warps you back to the entrance to redo the whole core approach; two say you simply fight and are asked again. An even split, and the consequence is very different.

Multiple sources flag the second dragon fight as an unavoidable back attack, which is where unprepared parties die.

The descent is a budget, not a step count

This mechanic is widely misdescribed, including in the premise this page was researched against, so plainly:

There is no step counter and no step limit anywhere in this dungeon. All eight sources that document the descent describe the same thing: a terminal on level one dispenses twenty units of Reserve Steam Pressure, doors cost four, two or one to open, and the machine at the bottom needs ten remaining to activate.

Two sources describe the underlying structure: a nine-sector grid in which door prices are fixed by sector, and you may move to a cheaper sector but never back to a dearer one. That one-directional rule is why a careless early spend cannot be recovered by backtracking.

And a botched run is recoverable. Two sources state the first-level computer resets the entire puzzle, returning you to twenty. This is worth its own entry because the dungeon is usually written up as unforgiving and the sources that say so are not the ones that mention the reset.

One disputed figure: four sources cost the side-room door at four units with a seven-unit refund inside; one wiki passage costs it at two, contradicting another passage on the same wiki.

The free door is not free

At the last door, with a particular character in the active party and between ten and thirteen units remaining, he opens it for nothing.

Both routes arrive at exactly ten units remaining. So this is not a budget decision at all — it is a content fork:

One source alone claims declining removes the offer permanently; seven are silent.

Correction: the weapon farm is eight times better above level 30

Tri-Face, in the first fixed area, is the practical source of Curse Spike. No published source mentions level in connection with it. The quantity is banded.

Mug yields Curse Spike at every band. The drop table is listed commonest slot first.

So the same farm, in the same room, against the same enemy, is between four and eight times more productive above level 30 — and every guide that describes it describes it at whatever level its author happened to be.

The top band is not just more of the same table. At 30 the two rare drop slots stop being Curse Spike and become Star Fragment ×3, while Curse Spike moves up into a common slot at ×8. So crossing 30 does not only multiply the yield, it changes what the rare roll pays out — which is the same shape as the Ruby Dragon’s drop table four areas back, where crossing a boundary took an item off the list entirely.

Mug and drop are worth separating here. This record’s drop rate is the maximum value — the same reading that makes one Propagator colour a guaranteed drop — while its mug rate is a quarter of it. So above level 30 the kill is the reliable half of this farm and the mug is the gamble, which is the reverse of how the fight is usually written up.

This is the second level-banded loot table this batch has turned up, after the Esthar crystal farm. Together they establish it as a general mechanism rather than a one-off, and it joins level-scaled stats and name collisions as a third way for honest guides to contradict each other.

Eden

Drawn from the last fight rather than found. The list is Regen, Dispel, Ultima and Eden, identical in all three level bands, so there is no level requirement — only the requirement that somebody has Draw equipped and uses it before the fight ends.

Ultima is on the same list, and no source mentions it. Across all 142 enemy records exactly two carry Ultima in a draw list: this one, and the superboss in the final dungeon. There is no third. So the fight everybody comes here to draw a Guardian Force out of is also one of two places in the entire game where the strongest spell can be drawn from an enemy at all — from the same four slots, at any party level.

If you have Draw equipped, take both before it dies.

Four sources agree it is the strongest Guardian Force in the game and that it breaks the damage limit.

Is this the only source? Three sources say it can also be drawn from an enemy in the final dungeon. Two say this fight is the only opportunity. One hedges by calling it the last chance before the final disc, which is consistent with the first three.

Correction: the superboss does not drop a weapon component

No source credits this fight with an ultimate-weapon ingredient, and the data agrees — its mug is one Three Stars and its drop is a hundred Ultima Stone. Both are worth having; neither is a weapon component, and neither is what people come down here hoping for.

The genuine weapon connection in this dungeon is the fixed-encounter farm above, which is exactly why killing the superboss too early is the most commonly cited regret about this place. Four sources agree every fixed encounter here stops existing the moment it dies, and one says the dungeon goes completely silent afterwards.

One source alone reports that the fixed triggers reset every time you leave the area and return, which would make the farm repeatable rather than one-off.

The rest

Four draw points, two hidden. One wiki puts the first on level two; three sources put it on level one. A Triple point is placed in the third excavation area by two sources and on a different floor by one. The side room’s point is reachable only on the route that declines the free door, making it a genuine fork-dependent pickup. Two save points sit near the bottom, one of them hidden.

There is not a single chest or key item in the entire dungeon. No source lists one. Everything here is draw points, boss loot, two Guardian Forces and two cards.

One source warns that equipping the rare-item ability actually lowers your chance at the best drop tier here. No other source mentions it, and it is a claim about how a junction ability interacts with the drop roll — which the enemy records cannot settle on their own. Flagged.

The encounter-suppressing abilities do not work in the core room. Two sources state this explicitly, and it is worth carrying into the fight: whatever you have leaned on all game is not available for the approach.

One community thread reports a progression bug on the PC release where the rubble needed to descend fails to appear after the first superboss; the reported fix is to leave, board the airship, take off, and return.

Game Notes

If this project ever needed one page to justify itself, this is it.

Ultima Weapon’s HP has been argued about for as long as there has been an internet to argue on. 51,100. 61,000. 100,000. 160,000. Four numbers, and the obvious conclusion from inside any one playthrough is that three of them are wrong — somebody misremembered, somebody typed a digit twice, somebody was playing a different version.

They are four points on one line. Every person who published one of those numbers measured correctly and wrote down exactly what they saw. The only thing missing from every single account was the party’s average level, because from inside the game there is no reason whatsoever to think that is a relevant fact about a monster.

And that is the whole thing, really. FFVIII made a design decision — enemies scale — that is perfectly sensible on its own terms and completely invisible from the player’s chair. Then twenty-five years of careful, generous, honest people wrote down what they saw, and the records they produced cannot be reconciled by any amount of care, because the missing variable was never on screen.

The fix is not better guides. The community’s guides are extraordinary; the research behind this page alone pulled from eleven of them and they are mostly right about mostly everything. The fix is a number that was sitting in a file the whole time.

I keep coming back to how cheap the resolution is. Four coefficients per enemy, one quadratic, and a dispute that has outlived several console generations just stops. Nobody needed to play better. Somebody needed to open the file.