DCFFVII Research Thread

Adding to the list of obscure DoC mechanics: The sum of your stage mission rankings determine how many Death Penalty Bullets you start out with in Chapter 12.

I only just now noticed this because of watching speedruns and comparing with my personal results. I always start out with 500 DPBs (Death Penalty Bullets), while speedrunners typically start out with 210-240 DPBs.

Rank - Points
S - 20
A - 10
B - 6
C - 4
D - 2

A chapter may at maximum have four stage mission (though some have fewer) so your maximum possible score per chapter is 20*4 = 80.

Grabbing my list of the global flags (which I used to test out what determines your starting DPB count) here is the score table from my old DCFFVII International save file. Decimal values used here.

Flag - Purpose - Score​
41 - Chapter 1: Sum of Stage Mission Rankings 80
42 - Chapter 2: Sum of Stage Mission Rankings 20
43 - Chapter 3: Sum of Stage Mission Rankings 40
44 - Chapter 4: Sum of Stage Mission Rankings 50
45 - Omega Reports Collected 63
46 - G Reports Collected 07
47 - Chapter 5: Sum of Stage Mission Rankings 80
48 - Chapter 6: Sum of Stage Mission Rankings 26
49 - Chapter 8-1: Sum of Stage Mission Rankings 46
50 - Chapter 8-2: Sum of Stage Mission Rankings 80
51 - Chapter 9: Sum of Stage Mission Rankings 30
52 - Chapter 10: Sum of Stage Mission Rankings 60
53 - Chapter 11: Sum of Stage Mission Rankings 80
54 - Unknown 00
55 - Unknown 00

Unfortunately the Omega Reports- and G Reports collected have no effect on DPB count, same with the unknown (possibly unused) flags #54 and #55. Would have been cool if the Omega Reports were more than just a collectible and a lore source.

So in my playthrough I have mostly S rankings, with some A and B rankings sprinkled in there.
The sum of my score:
80 + 20 + 40 + 50 + 80 + 26 + 46 + 80 + 30 + 60 + 80 = 592 Death Penalty Bullets

Because the maximum allowed is 500, the 592 is downsized to 500. In other words a great score is far from required in order to have your DPB count maxed out. The game check for all this happens at the end of Chapter 11 after you defeat Weiss.

Even speedrunners are in no trouble when they start out with just over 200 bullets since their goal inevitably leads to ammo being conserved. Additionally, when using a Limit Breaker bullets are not consumed for a brief period of time.

Despite its low impact on gameplay I greatly enjoy this discovery. Long have I wished for there to be some additional purpose to the stage missions and now I found that indeed all these quests and sidequests build up to something: Your DPB count.
Last edited:
Arachnero & the Phalanxpheres


How does this Chapter 11 boss fight function? Let's have a look.

Target - HP
Arachnero - 24000 HP
Phalanxphere - 1000 HP
Shield - 3500 HP

The three defeat conditions for Arachnero
  • Attack the giant arachnid's back or Nero himself until 12000 HP has been depleted.
  • Destroy three of the six legs. Each leg is worth 4000 HP.
  • Deplete Arachnero's HP to 0 or very close to 0. (Some nuances of this third condition still unknown.)
The battle can only end when Arachnero is attached to the topmost, giant levitating rock. In other words the battle can't end during a later potential phase when Arachnero climbs down.

Arachnero's death animation consists of Nero subtly writhing while any remaining legs are clinging to the top levitating rock. Damage numbers will be displayed- and Arachnero's HP bar still depleted before the death animation occurs, even if one of the three defeat conditions have been fulfilled.

The fact that damage may still be inflicted even when Arachnero is essentially defeated, along with the fact that all you need is to deplete 50% of Arachnero's maximum HP, means that players will often experience the battle ending at widely different values for Arachnero's remaining HP.

If you really want to see the battle end with Arachnero's visual HP bar being fully depleted, one way of doing so is to wound all six legs but never destroying more than two of them. Arachnero's total HP will now be super-low. A few shots to the main body will end the fight and let you see a variation where the graphical HP bar is completely empty or almost empty.

General battle- and enemy description

The fight begins with a shield protecting Arachnero, making it impossible to inflict damage until the shield is destroyed.

Four homing laser rods known as "Phalanxpheres", encased in crystal, are flying around Arachnero. One at a time they may set their aim and charge up a laser attack that has great reach. While charging the laser, the Phalanxphere's protective crystal shield will go away thus making it vulnerable to attack. A few seconds after the laser has been fired, the Phalanxphere's crystal will be restored and it becomes invulnerable again.

Randomness appear to determine when one of the four Phalanxpheres decides to exit its flying pattern to charge up an attack.

Phalanxpheres are responsible for reforming the shield when it has been destroyed. If none of the Phalanxpheres are currently in laser attack mode (this includes the moments before charging the laser up until after the laser has been shot) then they will reform the destroyed shield.

The size of Arachnero's shield is determined by how many Phalanxpheres were around to form it. The battle begins with the biggest possible shield which corresponds to the size that these oddly named homing laser rods will reform the shield when all four of them are present.

Shield Size 4: Four Phalanxpheres

Shield Size 3: Three Phalanxpheres

Shield Size 2: Two Phalanxpheres

Shield Size 1: One Phalanxphere

Each of these four shields has 3500 HP and the same defense, meaning each size is as sturdy as the other.

From sizes 3 and lower it becomes possible to shoot Arachnero's legs even when the shield is up! Good positioning and good angles may still be required in order for Vincent's shots to land but it is easily doable. The smaller the shield size the easier it is to hit the legs.

When the biggest shield, size 4, is up there are at least two legs that shots can land on without hearing the *PING* sound from the shield, but no damage will actually be registered.

Once two Phalanxpheres have been defeated, Arachnero will summon a hail of meteors. Both Vincent's melee attack and bullets can destroy these flaming projectiles. These meteors will be summoned at regular intervals until the remaining Phalanxpheres have been defeated. In the Japanese guide book these are known as "Meteor Bullets" [メテオ弾].

When all Phalanxpheres are gone and the shield is down, Arachnero will climb down closer to Vincent's elevation. Arachnero will remain in this position for only a few seconds before jumping back up to the biggest levitating rock. In this moment six balls of fire (known as "Magma Bullets" [マグマ弾] in the Japanese guidebook) will emerge from the lava below and home in on Vincent. The magma bullets can be destroyed with Thunder magic (Fire and Blizzard won't work) and melee attacks. Melee attacks will require some practice, largely because these magma bullets tend to hit Vincent earlier than expected, but Thunder magic can make quick work of these lava projectiles.

After Arachnero climbs down then up again a second time, summoning another wave of magma bullets, four new Phalanxpheres will appear and the biggest shield is restored. And so the battle cycle is back where we started, except for any damage we may have inflicted on Arachnero in the meantime.

Defeating Arachnero:
The Speedrun Strategy

(Any%, Normal Mode, Post-JORG)

Speedrunners at World Record pace getting to this battle will have Vincent on Level 5 and with the following equipped to Weapon Slot 1:
Powered Cerberus​
Long Barrel Gamma​
Power Booster Gamma​

Based on this setup and after much research, I determined what I think optimizes this battle to a consistently fast time.

Because the shield is only reformed when none of the Phalanxpheres are busy in laser attack mode, you will only want to destroy the shield just as a Phalanxphere starts setting its sight on Vincent and charging its laser. With the current setup, five shots will destroy the shield. The battle should thus always begin with firing four shots on the shield. If a Phalanxphere has already started targetting Vincent, then it's okay to destroy the shield with your fifth shot. If no Phalanxphere is targetting Vincent, wait until this happens or else the shield will get instantly restored after you've destroyed it.

The laser fired by a Phalanxphere has a long reach and the Phalanxphere will rotate slightly to home in on Vincent's current position. Thankfully the biggest possible rotational angle of the Phalanxpheres is limited, which can be used to your advantage when avoiding the laser. I have not figured out the exact targetting AI for the laser attack, but I find them easy enough to avoid. Worst case scenario for a speedrunner they will be hit by two lasers during the length of the battle, get KO'd and thus have their Phoenix Down status expended.

Once the shield has been destroyed with good timing, what remains is to shoot away at Arachnero's main body. Any speedrunner will not want to waste time with shooting the Phalanxpheres since this will not help in the goal of defeating Arachnero, plus the lasers aren't too much of a hazzle.

Current speedrunners have their sight scope configuration set to "Automatic". For the Arachnero battle this is a disaster, as auto-lock will often target the Phalanxpheres and make you miss the shield and Arachnero. The Semi-Automatic setting is the only one I can recommend to a speedrunner who is going through this battle.

Speedrun strategy in summary:
- Have the scope sight set to Semi-Automatic
- Timing the destruction of Arachnero's shield
- Avoid the Phalanxphere's lasers
- Attack Arachnero's main body relentlessly when the shield is down

A runner can expect to either get a two-cycle or a three-cycle battle with this strategy. Each cycle represents when the shield is down and you have a chance to deal damage to Arachnero. Here are two demonstration videos.

- Three-cycle, Automatic (showing how the Automatic sight will mess things up)

- Two-cycle, Semi-Automatic (embedded below)

Observe that in the two-cycle video, the timing at which the Phalanxpheres start charging at Vincent means that this becomes essentially a button-mashing boss battle. What's important to keep in mind is that not every battle may turn out like this, with RNG perhaps making it so the Phalanxpheres are too slow to start targetting Vincent. Unless a runner is counting their shots on the shield and being aware of the Phalanxpheres, they may destroy a shield too early, meaning the shield will be reformed and a big time loss will manifest.

The strategy is thusly a well-planned button masher. :monster:

Defeating Arachnero:
The All S Rankings Strategy,
alternatively the Three-Leg Strategy

Many a times when dreaming about what it would be like to actually complete my All S Rankings playthrough, I have wondered how in the world I would do the Arachnero battle. This is because of the two ranking categories "Accuracy" and "Critical Hits", with the former in particular being reason for worry.

Shooting an object like a cargo box counts as a missed shot and the same is true for when shooting Arachnero's shield. Ergo, this boss battle could easily become a catastrophe for a run where you want a high Accuracy score. Magic does not affect your accuracy, but magic is especially weak in the Japanese original and so destroying the shield with only magic is not a reasonable alternative.

These worries are quickly amended however when you know that Arachnero can be defeated by destroying three of its legs and that the legs become exposed in shield sizes 3 and below. Ergo all you'll need is to destroy one Phalanxphere and destroy the shield once so that you get the size-3 shield. This limits the amount of shots you need to land on the shield which would otherwise swallow up your accuracy score.

By great fortune, each successful shot on a leg counts as a critical hit. Shooting Nero's head will also yield critical hits, but due to how Nero wiggles back and forth it can be challenging to land a shot. Plus any damage performed on Nero or Arachnero's giant back won't make the battle go by faster in our strategy where only leg damage matters.

Thanks to the shots that will be landed on Arachnero's legs we will gain back any Accuracy score we lost on the shield, while also getting those critical hits that we wouldn't get if we had went for just spamming Arachnero's main body with bullets.

Shooting down Meteor Bullets also counts as a "missed" shot, same as shooting the shield. Of course that fact never enters into the equation since we consciously avoid the Meteor Bullets ever spawning, by making sure that we never destroy more than one Phalanxphere.

One trick to get your accuracy and critical hits through the roof would be to shoot the two legs that are partially exposed even when the size-4 shield is up. Shooting these legs (possible to reach by shooting from the opposite side of the stage where the battle starts) won't count as damage but they'll still count as valid hits AND critical hits on top of that. So you could theoretically shoot until you run out of ammo and get insane scores for both Accuracy and Critical Hits this way.

The one thing you WON'T get from this strategy is any increments to your killchain score. Destroying Arachnero's legs and shooting down the Phalanxpheres won't add to your chain. The only options available are...
- Repeatedly shooting down the shield
- Shooting and/or melee-ing the Meteor Bullets
- Melee-ing and/or using Thunder magic on the Magma Bullets

So if we end up desperate for any increments to our killchain then the two latter options are available, awkward though they may be.

Defeating Arachnero:
The Shield-Killchain Strategy

Yes, you pretty much know this one already based on what you've read. :monster:

If you have a weapon that is fast and powerful enough, you can destroy the shield so fast that the Phalanxpheres will continuously reform the shield while you destroy it again to increase your killchain and thus the feedback loop escalates.

There is always a chance that a Phalanxphere will instantly decide to home in on Vincent, thus quitting any shield reformation and making it impossible to add to your killchain (unless Meteor Bullets are currently falling towards you). Using cheats I was able to instantly defeat the shield with one shot when it reformed, but when I got to killchain 55 randomness decided that a Phalanxphere would start targetting Vincent and so the cycle ended right there, even with cheats.


The slower and weaker your weapon the more you'll need randomness to delay the Phalanxpheres targetting Vincent in order to get a chance to destroy the shield enough times for a great killchain. That's why this strategy is only really recommended with the Ultima Weapon, even though it is sort of viable with the Powered Cerberus.

Fun facts: The Japanese guide book, "DIRGE of CERBERUS - FINAL FANTASY VII - OFFICIAL COMPLETE GUIDE", never goes over many fine details of this boss battle like for example the three-leg defeat condition for Arachnero. It does however on page 215 recommend the shield-killchain strategy for this boss battle.



Google translate:
Let's actually use the Ultima Weapon​
If you use the Ultima Weapon, you can easily defeat the boss.​
Here, we will introduce the strategic method that can be achieved only with Ultima Weapon,​
using Nero Rafurea as an example. First, with the Master M Barrel and​
the Power Booster γ or β attached to the Ultima Weapon,​
attack the barrier put on Nero Rafurea. Then you can easily break barriers that usually​
do not break easily. Moreover, if you break the barrier, the slave​
will stop attacking you in order to put a barrier, so you can defeat the boss​
without giving the enemy a chance to fight back. It can be said​
that it is a unique strategy of Ultima Weapon with high POW and SPD.​
The chain gauge can also be accumulated if the barrier is broken.​
If the attack power rises,​
it is possible to destroy the barrier with a single blow.​

Arachnero, or "Nero Rafurea", as seen on page 213:

Last edited:
Gorgonero and the Frame-Perfect Trick Shot

After defeating Arachnero you'll battle Nero's final phase: Gorgonero.

Vincent will be limited to moving on a small floating platform, unable to move to any of the other levitating rocks.

Gorgonero will teleport in and out of the arena, randomly spawning in one of eight possible locations. Here as pointed out in the Japanese guidebook:


If Nero spawns on the same platform as Vincent, Nero will engage in a melee attack on Vincent unless you cancel him out with a shot or Vincent's own melee attack.

If Nero spawns on any of the other platforms he will briefly shoot at Vincent and then de-spawn of his own accord or after having been hit by Vincent's bullets.

Your only chance to damage Gorgonero is that brief window when he spawns on a platform and attacks Vincent. Gorgonero quickly becomes invulnerable again after getting hit, meaning that with most weapon setups you'll only get one shot of damage per cycle.

The Ultima Weapon's high firing rate however allows for you to easily land two or three shots on Gorgonero.

Even without a killchain and with Vincent at a low-level you have a chance to get a 1-cycle battle if you can land multiple critical hits in a row.

The Powered Cerberus is too slow to land multiple hits in the same fashion as the Ultima Weapon unless you are also equipped with one of the short barrels. Equipping a short barrel is not ideal however because of their low power and range.

But there is a way for even slow-firing weapons to land multiple hits on Gorgonero in one cycle and that is with the Frame-Perfect Trick Shot.

CAVEAT: I do not have precise frame-advancing tools nor have I studied the assembly code for this scenario. My conclusions are based on over 250 recordings attempts and over 100 attempts on top of that, all using save states and having Dirge of Cerberus play at a very low speed. The trick may be frame-perfect or it may only be near-frame-perfect.

If you can land a bullet on Gorgonero EXACTLY as his hit box generates, then damage will be dealt but Gorgonero's battle AI will act as though you had not just landed a shot. Gorgonero will continue with his attack animation, hit box still up and open for an additional shot. This works both for when Gorgonero is shooting at Vincent and when he spawns on the same platform as Vincent for a melee attack.

One limitation with this trick shot is that you are only allowed to deal damage from one single bullet at just the right frame (or few frames). Vincent's Cerberus fires three bullets per shot. Each bullet that lands on an enemy is registered one at a time. What needs to happen in the frame-perfect trick shot is that the first two bullets land on Gorgonero when he is still invulnerable, then have the third bullet land when Gorgonero's hit box allows him to be damaged. If the shot lands any later than that, Gorgonero's AI will kick in where he registers a shot has landed on him, which will lead to the cycle ending and Gorgonero teleporting away.

Can speedrunners use this trick to defeat Gorgonero faster? Can this battle which usually consists of two cycles be diminished to one cycle?


In this video we have mimicked GocuComeBack's setup as it is in his world record.

Vincent Level 5​
Powered Cerberus​
Nova L Barrel​
Power Booster Gamma​

Each critical bullet hit deals 793 damage. With the trick shot we get to land one bullet's worth of critical damage and with the second shot we land three bullets worth of three critical hits.

The damage dealt is thusly 793 + 793*3 = 3172. With Gorgonero's HP being 3000, this damage is more than enough.

It is very important however that your equipment is not downgraded from this setup. In gumv's personal best, which landed him the bronze trophy in Dirge Any% (Normal Mode) speedrunning, he has the Long Barrel Gamma equipped in the Gorgonero fight. With this the damage dealt would be too low to 1-cycle Gorgonero.

Damage dealt: 739 + 739*3 = 2956. Only 44 more damage and Gorgonero would be defeated. So close!

Even when using an emulator I found that my success rate with the trick shot was roughly 1 out of 30 attempts. This diminishes the likelihood that we'll see this trick performed in a speedrun that isn't tool-assisted. I hope though I'll one day get to see the trick shot performed in a runner's personal best.
Last edited:
GocuComeBack has achieved a new world record in the Any% category of Dirge of Cerberus speedrunning, at last achieving a time below 1 hour 37 minutes.

At the time of writing, the leaderboard for DCFF7 is yet to be updated with the new record, but there you can see GocuComeBack's previous WR of 01:37:08 and which was achieved just over a year ago.

In the new record Gocu stopped the timer at 01:36:52 and later re-calculated the time as 01:36:48. The moderators will decide what the true final time stamp is (assuming Gocu has submitted the run for review).

Gocu's gameplay is solid and he maneuvers the menus quickly. The Chapter 12 segment with the Crystal Feelers is a dreaded segment because of the heavy RNG behind which Crystal Feeler opens and when they open. Gocu got massively unlucky with the final CF taking 32 seconds to open. In other words, with good RNG Gocu is quite close to achieving a time below 01:36.

I am honored that Gocu is making use of optimizations suggested by me. One example is an idea I showed to him less than 24 hours before this WR was made.

Normally runners wait for the Rusty Keys to spawn, meaning just below thirty seconds is spent doing essentially nothing. I demonstrated that the runner can use this time either to pick up a Limit Breaker or a Hi-Potion (assuming the latter had not been picked up earlier) and do so with zero time loss.

The way this changes Gocu's actual route is that he picks up the Limit Breaker here instead of one Limit Breaker in Chapter 4 just before you enter the basement of WRO HQ. This saves roughly 3-4 seconds because the LB in Chapter 4 sidetracks you from the straight path. One downside is that this means you no longer pick up the Handgun Bullets that are adjacent to the Limit Breaker but in the end Gocu has determined that not picking up the bullets is worth the time save.
Last edited:
Incidentally, all of the FMVs from Dirge of Cerberus have been upscaled with the use of AI Gigapixel and uploaded over at the channel Beyond Midgar. :monster:

Tangentially, on the topic of fanmade content, I simply have to share this tumblr page: Little-Vincent or "crochet Vincent". Follow the adorable crochet Vincent on his adventures throughout the world.


Going back to regular content, I listed the EXP gain from each enemy in Dirge, specifically in Normal mode of the International version. My motivator was to provide speedrunners with more knowledge so they can make more calculated decisions. Inbetween most chapters a speedrunner will convert EXP to gil and thus each EXP gain matters, given that a speedrunner's wallet will always be quite limited in any given run.

- DCFVII: EXP gain from enemies, document version 1.0

Enemies either give a set amount of EXP when defeated or they continuously grant EXP as you deplete their HP. Only very few enemies will grant EXP by both these means.

While researching this I discovered two surprises.

Getting EXP from bosses even when they are invulnerable
At any point when the boss is invulnerable you can still gain EXP from them by landing melee attacks or Thunder magic shots.
This is true whether the invulnerability is when the battle starts, mid-battle or even after the boss's death animation.
Bullets (all available types), Fire and Blizzard magic, as well as Galian Beast's fire projectiles, will NOT contribute to EXP in these moments.

In speedrunning this is moderately useful since more EXP means more gil. I have not yet calculated the exact gains but I estimate that in certain battles the speedrunner is guaranteed a few hundred extra gil in a given boss battle by making sure to use Galian Beast's melee attack on bosses during their invulnerable state(s). So the bosses in Chapter 3, Chapter 5 and Chapter 9 will by far be the most fruitful EXP/gil-farming instances since current strategies already involve transforming into Galian Beast.

One exception with Thunder magic is Azul the Cerulean in Chapter 4 since he is immune to Thunder.
No extra EXP can thus be gained from this version of Azul by using Thunder on him when he is invulnerable.

If the player has MP to spare, they may even use Thunder magic on Rosso in Chapter 8 BEFORE even triggering the battle and thus still gain EXP this way. Thunder Lv3 works best with its long range. Thunder Lv2 just barely reaches but you have to stand very close to the cutscene-trigger in order to reach Rosso with the magic shot.

My intuition tells me that this behavior of getting extra EXP (by kicking the enemy when they are down!!! :wacky:) was not intended by the developers but honestly I can't prove this intuition of mine.

Destructible glass tiles grant EXP
Chapters 1, 3, 5, 9 and 10 have glass windows (or in the case of Chapter 9, glass casings around crystals/materia) that will grant EXP when destroyed.

Using single-bullet projectiles always lead to only 1 EXP being gained. Fire and Blizzard can destroy glass tiles but Thunder (and melee attacks) can't. Using the three-bullet shots from a Cerberus gun you can gain 1, 2 or 3 EXP. The further away you are when shooting, the more likely it is that you'll gain 3 EXP from the shot. I have no idea why the glass tiles behave so weirdly but it clearly is related to how the tiles register each of the three bullets that come from a Cerberus shot.

I was quite surprised by this EXP gain since glass tiles don't even count as "valid" targets, in the sense that hitting them with bullets will count as a "missed shot" and thus decrease your accuracy score. This mechanic is not useful or economical, since each bullet in the store typically cost 10 gil. So each single EXP you gain (1 EXP = 10 gil) is automatically lost if you intend to use that extra EXP/gil for buying bullets. Still, neat to know this obscure fact.
Last edited:
Watch out Hades. I'm back for you and your three-headed beast.

JORG - Hard Mode - Vincent Lv1 - Chapter 9: An Empire in Ruins


Chapter 9 - All S Rankings Video Part 1

Chapter 9 - All S Rankings Video Part 2 *re-upload*

Activate English subtitles for video commentary.


Compare the results above with the minimum requirements:
Targets Destroyed - 120 or more
Accuracy Rate - 75% or higher
Damage Sustained - 4999 or less
Critical Hits - 60 or more
Killchains - 80 or more
Items Used - 5 or fewer
Limit Break Used - 5 times or more
Mako Collected - 90% or more
Times KO'd - 0
Time Expired - Less than 40 minutes


Stage Missions:
- Collect top secret Deepground data files! : 5 out of 5 Top Secret Files
- Destroy the surveillance mines! : 90 points or more

I recorded the first part three years ago. My nerves were killing me and after only a few attempts with segment 9-4 I realized I could not go on. Upon my return three years later I spent three days preparing by testing out strategies on emulator, using save states to spare me from much needless stress. After that, all it took was 2 1/2 hours of gameplay on console to achieve my desired result for the 9-4 segment.

In all my work in 2016 I only ever played on console. It is definitely more hardcore to perfect your gameplay that way but it is also definitely smarter to make use of emulator tools to perfect your gameplay. For reasons of health I have to play smart, not harder.

S rank in Critical Hits is definitely the hardest category to achieve here, unless you cheese for Critical Hits by using weak guns on the Black Widow's belly or by using Azul as a bullet sponge. I decided to not cheese my results and instead see how close I would end up getting to the minimum requirement. My final result of 66 Critical Hits may look shameful, but my performance against the Aerial Mines in the elevators make up for it.

Truly, segment 9-4 (video 2/2) is the star of the show and demonstrates high performance gameplay coupled with good planning. If you want to test your skill with the manual scope to the fullest then the elevators in segment 9-4 is a great place to do so, in particular if you're aiming for a high score. Shooting down an Aerial Mine grants 1 point, but if an Aerial Mine is destroyed by another exploding Aerial Mine, you get 3 points. So knowing when to shoot, and landing the shots, is essential. Remember also that Vincent has a way lower firing rate in JORG, so it's not like you can easily correct a mistake by shooting frequently.

For the boss fight I decide to play it safe and intentionally go for no-damage. You can tell how I intentionally only jump for the Mako Points when I am confident that Arch Azul has slowed down and won't catch up with me afterwards. This means I end up with slightly less MP at the end than I wanted, but the masochist in me wants to see if I'll be forced to use an Ether or not in Chapter 10 because of this. So far the only time I have used an item was in Chapter 8-1. Otherwise every chapter has been a zero-item run.

EDIT: I have noticed that video part 2 is extremely fuzzy. I will try to see if rendering the video differently solves the issue, but I can't guarantee anything since my recording setup is not 100% identical to what it was three years ago.

EDIT #2: I've deduced the problem. My new recordings have their source frames at resolution 654x480 instead of the usual 640x480. My video editor doesn't have a matching renderer. I may have to get creative with this one.

EDIT #3: So I re-recorded the source video file as a 640x480 video as it was playing. This leads to a way better result on YouTube and the only thing I'm "sacrificing" are most of the black borders. Will soon link a re-upload of part 2.

EDIT #4: There we go.
Last edited:
It is only natural to fear a GoD (God of Dirge)

The stage mission "Destroy the surveillance mines!" is mostly the same between versions. The biggest difference is in what I call "phase 3" of elevator #2. In the Japanese original this is when eight Aerial Mines emerge. If you leave them alone they will settle in stationary position and even remain after the elevator has reached the bottom. The Aerial Mines position themselves roughly like this.


Meanwhile in all later releases, a whopping thirteen Aerial Mines emerge and surround the moving platform.

Thanks to the increase in Aerial Mines it is way easier to reach the cap of 99 points in the stage mission. Destroying eight Aerial Mines in a row grants 1+3*7 = 22 points. Destroying thirteen Aerial Mines in a row grants 1+3*12= 37 points.

If the score wasn't capped at 99, I would have gotten 108 points in my JORG playthrough. Doing the same thing in post-JORG nets you 123 points. If you know what you are doing you won't even have to defeat the 4*2 Aerial Mines in phase 4 of the second elevator, since you'll already have reached the 99 cap in the post-JORG game.

When I made my recording I thought that the only way to destroy all eight Aerial Mines in a row was with that particular timing right when the first Aerial Mine comes around the pillar. Some additional research on the mines when they've settled in mid-air shows however that a maxed out chain-explosion is possible: It depends on the angle from which you are firing.

The general rule is this: Firing on an Aerial Mine from an angle where the mine is mostly pushed out of the circle of airborne mines means it is much more likely that it will explode alone, or only move so much that it will trigger an explosion on its closest neighbour. The domino effect will be small and contained.
Firing on the mine so it is pushed closer to its neighbour will almost guarantee that its neighbour explodes AND seemingly nudge the neighbour so much that it is able to trigger an explosion with the next neighbour and so the domino effect is in full swing.

I say "seemingly" and "general rule" since the collision data of the Aerial Mines will not always act in your favor and it's not like you are getting any clear visual feedback that you are pushing an Aerial Mine closer or farther from its neighbour(s). My conclusions are based on repeated attempts with the use of save states, but because the collision data isn't visualized we can't always make sense of why a full chain-explosion happened or not. The dream would be to have a build of Dirge where we could visualize the collision data, like this build for Ocarina of Time.

By adding five more mines in the post-JORG game it is ensured that the whimsical collision behavior of the Aerial Mines won't prevent a complete chain-explosion. You also make it easier to guarantee that the DG soldiers are defeated during the process as well. As always though it is important to keep out of the radius of the explosions.

The only way to mess this up is if you fire on the first few Aerial Mines too close to their spawning location.


^The yellow square is the spawning point.

Since not all mines instantly spawn, you could easily commit the mistake of destroying the first few mines just as they appear which will lead to the explosions finishing early and then the subsequent mines spawning safely and unharmed, thus lowering your potential score. This is why I shoot the Aerial Mine later in phase 3 in post-JORG compared to JORG. If I use JORG-timing, not all mines will have time to spawn before the explosions are over. There are just too many mines in the queue.

The fact of spawn-timing is also why I wait for the mines to complete a rotation in phase 1 and phase 2 at the second elevator.

The post-JORG game created a new problem though: The sudden displacement of exploding Aerial Mines onto the platform where Vincent stands. By the chaotic, random behavior of the exploding mines they will sometimes be thrown on top of the platform and thus the explosion radius will be way closer to Vincent, sometimes unfairly so.



This is why even careful players may sometimes find themselves harmed or even KO'd in this segment. They maneuver carefully to avoid the explosions but then a mine gets displaced suddenly and you may have no chance to avoid it. I have not seen this behavior in the original Japanese release.

My guess is that this sudden displacement of the mines happen because there are simply too many objects interacting with one another. So in solving the JORG issue of a complete chain-explosion being needlessly difficult to achieve, they created the problem of the player possibly getting an unfair KO or even Game Over.

There is one more version difference. In phase 2 of elevator #1 there is an Aerial Mine that will pop right out of nowhere if you are playing the Japanese original.

Frame 1

Frame 2

So if you explode the left Aerial Mine too early, it will not chain together with the adjacent Aerial Mine that suddenly pops out of nowhere. This was fixed in later releases so that the formerly invisible Aerial Mine is now visible from very early on. It will be stationary and the three other Aerial Mines of the phase will seem to spawn out of it, until the time comes for the mine to be paired up with another and move in sync.
While playing ahead on emulator (to figure out strategies and test that everything will work for S rank) I found there is a certain spot in the game that you aren't allowed to pass. Because of this you can't clear Chapter 10.


No matter how high you jump you aren't allowed to proceed from this point. The game insists that you are jumping against a solid wall when visually you see an open path.

At first I got REALLY scared. "Have I corrupted my save file?" and I began to wonder if my S rank run was ruined.

But I faced the same issue even when accessing the area on wildly different save files. Apparently the PCSX2 emulator simply isn't properly emulating this part of Dirge. Fascinating.

I have definitely played through this segment on emulator with the post-JORG game but it's entirely possible this is my first time playing this section on emulator with the JORG version.

Still. Now I'm frightened that this problem will manifest on console, even though by all accounts it SHOULDN'T. *bites fingernails*
To ensure anyone who might be worried otherwise: Yes, the grind for Chapter 10 is going on.

After over 100 attempts there have been only two times when I got close to a desired result in Ch10-1. The various ranking categories isn't the problem. The challenge is to perform Ch10-1 in a way that looks good. Doing this requires both fast reflexes and a lot of luck. The grind could easily consist of a few hundred more attempts.

Spoiler: Chapter 10 marks the first time that I equip the S Adjuster for potentially legit strategic purposes. This accessory increases the accuracy of short-range shots.

For reference, here is the short range bullet spread of Powered Cerberus + Nova L Barrel without the S Adjuster.


And here is the bullet spread WITH the S Adjuster equipped.

This is the sacrifice you make by using the Nova L Barrel. Your stats for short-range accuracy are awful and your mid-range accuracy is only decent. As it happens, most of the shots I fire are either at short range or at medium range.

When accuracy is perfect over a particular distance, there will be zero bullet spread. Indeed, even when using Cerberus which fires three bullets it will be adjusted so that all fired bullets converge on the same, single point.

If one's own aiming is imprecise, then having bad short-range accuracy stats can save you just as much as it can doom you. You might hit with a badly aimed shot BECAUSE the randomized bullet spread was so wide, or you may miss a well-aimed shot BECAUSE of the bullet spread.

Were I playing with a mouse I believe the positive value of the S Adjuster would be far more definitive. I began the S rank playthrough with a normal PS2 controller and I intend to continue that way for the sake of consistency. I wish to show how S rank is possible (and can look somewhat decent) even with the limitations of the analog stick.

So why am I switching to the S Adjuster? Because in Ch10-1 there are two fast-moving Aerial Mines that I each need to hit with all three bullets of a single shot, especially so on the first Aerial Mine before my killchain has increased by one more point. If my shot lands on the Aerial Mine I want to at least have increased the probability that ALL the bullets of that successful shot land. A wide bullet spread will, by my intuition, decrease the odds of striking both Aerial Mines down in quick succession. Maybe I'm fooling myself but this is the strategy I'm going with. All three bullets MUST land on the target.

And so the grind continues.
^In retort, I will admit that I now completed Chapter 10-1 in a satisfactory fashion but that I'm leaving in two embarrassing mistakes in there. :monster:

Mistake #1: I lose some MP because I use an Ether too early.
Mistake #2: I accidentally switch to a weapon where my Defense stat is lower, when my intention was to maximize my defense.

Both mistakes happened because my pulse was so high that my thought process was inhibited. The pressure was ON! But imperfection is a part of creation, so I'm allowing myself to be satisfied with this run. Now awaits the remaining four segments of Chapter 10.

For anyone looking for a Dirge of Cerberus review with plenty more positive energy than negative, I highly recommend this video by Avalanche Reviews.

He highlights that the gameplay is superior when played with a mouse. I believe his positive experience also stems from the fact that he's using an excellent upscaler: The XRGB-mini Framemeister. Somehow, even though he's playing on console, he is able to play the game upscaled to 1080p and the quality rivals that of PS2 emulation.

From my very first time playing Dirge, I observed how the graphics looked very fuzzy and washed out. Enemies blend into backgrounds, making it harder to spot them and to aim your shots. When upscaled to higher quality, or emulated, the game is easier to play. When enemy models are sharp and in clear contrast against the environments you can spot foes way faster and you can feel how your manually aimed shots land more effortlessly. Add in mouse support and you can gain the most out of the manual scope in particular.

I also believe that playing the game on a larger television helps, if only because the visual input is widened thus giving you larger targets in-game. When the game was released in 2006 it was more common for households to have smaller televisions. I doubt there were tools back then to play in higher quality than the intended 480p but I could be wrong.

For these reasons I believe that Dirge of Cerberus would benefit more from an HD Remaster than many other games from the PS2 era.
Chapter 10-2 has been cleared in a pleasing manner. Ahead of me now are the absolutely MONSTROUS segments 10-3 and 10-4.
The gun turret section in 10-3 will test my patience and the Defeat-100-Enemies challenge in 10-4 will surely lead me down many frustrating Game Overs. The long-dreaded challenges await. *John Wick mode intensifies*

With how I've played 10-1 and 10-2 my Accuracy Rate score is now 91%. Keeping an especially high AR score is necessary because the gun turret section in 10-3 will easily drop your score to below 60% even if you are trying to play carefully. Because the Chapter 10 requirement for S rank in AR is 55% or higher, the inevitable drop in AR is a disaster that must be prepared for. In post-JORG the minimum Ch10 requirement for S rank in AR is 43%.

In Ch10-2 you need to destroy four carts, kill five DG soldiers and destroy the one chopper. Miss one and you won't get S rank in the "Eliminate the pursuers!" Stage Mission.

The unfair part here is that the carts are not "valid" targets, so shooting them with normal bullets counts as a missed shot. Magic doesn't count as a fired "shot", ergo why it won't affect your Accuracy Rate, so the solution would be to destroy the carts with magic, were it not for how weak magic is in JORG and the fact that I need to preserve MP and avoid MP restoratives in my playthrough. I have no choice but to shoot the carts and thus get my AR lowered. What I can do to compensate is to shoot down all the DG soldiers and later use the chopper as a bullet sponge to boost my AR score. The chopper has 10000 HP but will stick around as an active target even after you've depleted its HP, so that's useful for us.

Fun fact: The chopper is special in that you aren't actually "depleting" HP. When damaging the chopper in 10-2, the damage is added together and when you reach 10000 or higher then that means the chopper will be destroyed when you reach the end of the road. This makes the 10-2 Dragonfly very special, since most targets are programmed so that when their current HP reaches 0 then they'll be defeated. Not so with this Dragonfly.

On top of the carts not being "valid" targets and thus decreasing your Accuracy Rate, they also sport the behavior of not adding to your killchain value...usually. The more I dig into these carts the more of a clusterfuck they become.

Destroying the first two carts with normal bullets will NEVER up your killchain. But destroying the third and fourth carts at the appropriate moments during this on-rails segment will add to your killchain. There are no special markers, no indicators, no visible variables other than your own experience playing the section over and over that tells you "if I destroy the cart here then I will get the killchain, but if I destroy it over there I will not boost my chain".



Each cart has 800 HP. As you'd expect, their HP is reduced to 0 when they are destroyed. But when the third and fourth cart are destroyed at a time when they boost your chain, their HP is not reduced! DESPITE the cart model being destroyed and despite the Stage Mission counting the cart as destroyed, its HP never reached 0.

All the aforementioned behavior of the carts is true for both the Japanese Original version and the International version.

Destroying the carts with powerful magic in the INT version I observed the following weirdness... Destroying carts 1, 2 and 3 with Thunder magic will always add a boost to your killchain. Destroying cart #4 with Thunder magic will NEVER boost your killchain. (Fire magic never boosts your chain for any of the carts)

I am reminded here of the weird behavior that melee attacks and Thunder magic can add to your EXP even when a boss is invulnerable. Clearly this also extends to the topic of killchains but it's fascinating that the fourth cart is immune to this special effect. In all these cases when the cart is destroyed by Thunder magic their HP is always reduced to 0, so it doesn't perfectly mimic what happens when you just happen to shoot down the third and fourth cart at an arbitrary moment.

Were I to dig deeper I'm sure I would find more weird behavior. One thing I did not mention earlier is that when you strike certain DG soldiers (the DG "Sergeant" type) in Ch10-1 with the Galian Beast's fire projectile, the hit will ALWAYS count as a critical hit. But only for particular DG sergeants. For others the hit from the fireball will always count as a normal hit. How odd.
Last edited:

Clement Rage

Pro Adventurer
Do you need to do the 100 soldiers mission? I feel like that would break me in your shoes.

Also, I never knew I was destroying the carts, I thought I was just shooting the drivers.

Re those crits, are those the ones that suddenly jump in from nowhere between the cardkey barriers, or is that later?
The condition for S rank in the Stage Mission "Defeat 100 enemies!" is to defeat 80 or more foes. So technically I could just run away after having defeated 80 enemies. But then I would miss out on the gil rewards and the Gold Moogle Doll from defeating the final four enemies. Most importantly though, running away from this challenge would bring dishonor on me, my family, my cow, my PlayStation 2, etc etc. I will bring honor to us all.

The soldiers that jump in from nowhere between the EM barriers is the final part of 10-3, so indeed I haven't gotten there yet. After a bit of testing I have become wiser about the critical hits bug and how it *actually* works.

Summary: JORG-exclusive "Critical Hits" bug with Galian Beast

If your last bullet shot on an enemy was a critical hit, then all fire projectiles from the Galian Beast will land as a critical hit on that particular enemy. This works on both normal enemies and on bosses.


The effect can not be reset by using melee or magic. Only after a bullet lands on an enemy and causes normal damage will the effect be reset. Once reset, all fire projectiles from the Galian Beast will land as a normal hit on that particular enemy.


For enemies that "respawn", like these Deepground Sergeants in 10-1, the critical hits bug will carry over to its respawn enemy counterpart since the game treats this as the same enemy but revived.

The bug was patched out in all post-JORG releases, meaning this exploit can no longer be used to farm critical hits.
Thread post #999! Whoever makes post #1000 in this thread better make it count! :monster:

At long last Chapter 10-3 has been cleared according to my high standards. This was a brutal gravity well, as expected.

I did in fact clear Ch10-3 seven days ago and did so in a manner that was technically good enough. But I wanted an even better result. One with fewer embarrassing mistakes. This has now been achieved. There are still plenty of frustrating missed shots but that's to be expected in a 12-minute segment that is so crowded with enemies and challenges. RNG is a factor in almost all sections so that must also be taken into account.

When I started out practicing the gun turret section I consistently got a final Accuracy Rate score of 48%-53%. Gradually I became better until I consistently landed on an AR score of 60%-63%. With S rank requiring an AR of 55% or higher I confirmed that indeed I will have no trouble with this S rank.

Then while playing through Ch10-3 today, going through the frustrations and wondering if I was ever going to get out of this hell that my high standards were putting me through, I suddenly landed on a whopping 72% AR score! The run was already dead when I found this out, but it absolutely blew my mind. I had NEVER gotten above 64% AR by the end of a run. There was nothing in my gameplay that "felt" like I had improved my performance in the gun turret section, so the sudden jump in performance is still a mystery to me. In the final run I landed on 68% AR and that makes me very happy. Hopefully Ch10-4 can help me improve my score up to 69% or 70%, but I know that with how many shots has been fired so far it will be very difficult to increment my percentage score.
(Checking my save file results on emulator and with Cheat Engine, I can see that I've fired 572 shots and landed 390 of them.)

A big reason that I consume so few restoratives in JORG is because transforming into Galian Beast restores Vincent's HP. Because you can't transform while on the gun turret in Ch10-3 (or for that matter, sit down on the gun turret in Galian Beast form) you will definitely be forced to use restorative items here. Normally I would have to use two restorative items here, but I got such good RNG that I only had to use one restorative item!

For S rank you must use 13 items or fewer in Chapter 10. It is of course the Stage Mission "Defeat 100 enemies!" that will require you to chug restoratives like an addict. Thanks to having only used an Ether in Ch10-1 and a Hi-Potion in Ch10-3, I am allowed to use up to 11 more restoratives! Feels so good to be within such safe margins.

Having used 7 limit breaks so far I only need one more for S rank, but I reckon this will be my first chapter where I actually transform into Galian Beast MORE TIMES than the minimum required for S rank. :D

Ch10-3 is just so complex and has involved so much research, despite it being one of my most replayed segments even before this S rank run. I look forward to writing more about my strategies later. For now I'm going to enjoy my hard-earned victory in peace.

Ch10-4 is the final MAJOR hurdle. Nothing that comes after will be as hard as Ch10-3 or Ch10-4 was. Slowly...we are approaching the end.

Last edited:

The Twilight Mexican

Ex-SeeD-ingly good
Thread post #999! Whoever makes post #1000 in this thread better make it count! :monster:

1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 ... 12 ... 13 ...

How far was I supposed to go with this?

Anyway, that's it. Happy 1000 posts! :monster:

Shademp said:
Slowly...we are approaching the end.

It's been such a long time coming. I imagine the only one nerding out about this more than me is you.
Top Bottom