DCFFVII Research Thread

Ch12-1-3 and Ch12-1-4 have been cleared. The goals I set out for myself with Ch12-1 have been met.
  • 100% accuracy rate (AR)
  • Destroy every single target
  • Receive zero damage
  • Achieve maximum possible killchain count
I actually ended up undoing my big-brain moment from the end of Ch12-1-2. I switch there to Weapon 3 where I have Blizzard Lv3 equipped so that I may instantly start off Ch12-1-3 with a Blizzard Lv3 spell. But it turned out more advantageous to begin the tunnel-flight with Death Penalty and then use Blizzard Lv1 throughout this segment. It stuns me just how often I was able to defeat all 11 targets and even get the full 10-chain.

What follows is a room where a total of 12 Omega Buds will spawn, though only three will roam the place at any given time. Even with fast gunning it is entirely luck-based whether or not you get hit by the lasers. Your odds of not getting hit are pretty good though.

The final area of Ch12-1-3 has you ascending some organic stairs while facing 11 Omega Leaf enemies in JORG, Hard Mode. This is where I acquired my final Critical Hit required for S rank (plus one extra critical hit just for good measure). To my surprise I actually did learn how to strike critical hits on the Omega Leaf enemy with a decent amount of consistency, though I'd go so far as to call it the enemy that has the most difficult-to-hit weak spot.

To aid my aiming I unequipped Rapid Fire. The game doesn't always signal this, but the Rapid Fire in fact worsen all your range-based accuracy stats: Short, Medium and Long. Similarly, the game doesn't outright tell you that the Sniper Scope improves your Medium- and Long distance accuracy. In the Japanese game these ranges are referred to as LOW, MED and HIGH.

Your accuracy range is ranked in-game from lowest to best as C, B, A and S. If you are moving from one rank to the other when you change equipment then yes that does indicate to the player the equipment's effect on range-based accuracy. Nevertheless, the item descriptions to Rapid Fire and Sniper Scope don't inform you of these effects and the effect may easily go unnoticed unless you, like me, have Cheat Engine to show you the numbers.

Here is what the Dirge of Cerberus Complete Guide says about range-based accuracy on page 40.




※ 距離の目安は、カーゴの一辺が1.2m、ドラム缶の高さが1.5M

Google Translate said:
HIGH: Hit accuracy at a distance of 38m or more.

MED: Hit accuracy at medium distances from 12m to 37.99m.

LOW: Accuracy at close range (0-11.99m).

* As a reference for the distance, each side of the cargo crate is 1.2 m and the height of the drum can is 1.5 m.

Because the Omega Leaf I aimed for in Ch12-1-3 was probably in the medium-distance, I equipped M Adjuster. Similar to how in hindsight I should have upgraded my S Adjuster to S Adjuster Gamma when Chapter 10 ended, so too should I have upgraded M Adjuster to M Adjuster Gamma at an earlier point. Still, the current range-accuracy stats were good enough.


The red ball is the Omega Leaf's weak point but the reason I usually failed with critical hits (apart from having had Rapid Fire equipped before) is because I was aiming towards the center of the red ball: Just above the hole that shoots projectiles and just below the grey pin on top. The area LOOKS very open for critical hits but it is in fact NOT.


You must aim for the sides of the red ball, marked above, far away from any of the grey shielding. If you shoot between the cannon and the top pin, your hit will usually register as 0 damage (shield is struck) or normal damage. On a rare occasion the game may register it as critical damage but it quite simply is not a reliable target.

In my final recording I ended up taking a lot of extra time with this Omega Leaf just to make sure I didn't mess up and even then I accidentally landed one normal hit because I was one pixel too close to the center of the red ball. Thankfully I did amass two critical hits in the end plus the extra ten seconds spent here just means more time for subtitled commentary (and listening to Longing).

Ch12-2, "A Finale Chaotic" is all that remains......

I do not know what to aim for here and how much time I will spend on this final battle. We shall see. But the finale of the All S Rank challenge will definitely be uploaded before January 1st, 2021.

The Transformers theme by Hans Zimmer definitely reflects the epicness I'm feeling right now.

The challenge that began on August 2016 is now over as of December 2020.​

JORG - Hard Mode - Vincent Lv1 - Chapter 12-1: Omega and Chaos


Activate English subtitles for video commentary.


Compare the results above with the minimum requirements:
Targets Destroyed - 75 or more
Accuracy Rate - 80% or higher
Damage Sustained - 2799 or less
Critical Hits - 10 or more
Killchains - 50 or more
Items Used - 0
Limit Break Used - 0
Mako Collected - 0% or more
Times KO'd - 0
Time Expired - Less than 15 minutes​

JORG - Hard Mode - Vincent Lv1 - Chapter 12-2: A Finale Chaotic


Activate English subtitles for video commentary.


Compare the results above with the minimum requirements:
Targets Destroyed - 1360 or more
Accuracy Rate - 67% or higher
Damage Sustained - 67749 or less
Critical Hits - 725 or more
Killchains - 790 or more
Items Used - 49 or fewer
Limit Break Used - 55 times or more
Mako Collected - 91% or more
Times KO'd - 0
Time Expired - Less than 6 hours, 31 minutes 40 seconds​



Ninja Potato
You've done a man's job.
but are you really a man?

Great job dude, your dedication to this thing over the years has been astonishing. I really respect all the time and effort put in to completing this challenge. You probably know more about Dirge than anyone at this point. Bravo!
Published a frontpage article yesterday about the achievement.

Four and a half years ago, after having played through multiple versions of Dirge of Cerberus multiple times, the idea was seeded in my head of completing the main story campaign with perfect grades in every single category. That idea is now a reality and you can watch the complete video series showcasing this achievement in the YouTube playlist linked below.

Dirge of Cerberus ( JP ) – Hard Mode – All S Rankings Run

When a stage is completed in Dirge of Cerberus the player is ranked, from worst to best, with the following letters:
D, C, B, A & S. The player is ranked in ten categories, then given their average ranking for that chapter, followed by the rankings based on how they performed in the quests known as Stage Missions. After Omega Weiss is defeated in the game’s final chapter the player is then ranked based on their performance throughout the whole game.

The playlist contains 4 hours 20 minutes 10 seconds of footage. If watching the entirety feels overwhelming then I recommend jumping straight to the final video which includes a 4 1/2 minute montage of the run. There is no special reward from achieving S rank in every category other than the satisfaction of having a really good Dirge of Cerberus save on your PS2 memory card.

One big question may be occupying the reader’s mind right now.


The answer being quite simple: Dirge of Cerberus is my hyper-fixation. I enjoy pushing the game to its limits and I love learning more about the game. Adding to the sense of exploration and discovery is the high likelihood that nobody has ever achieved S rank in every single category throughout the game before now. Certainly, nobody else has ever uploaded a recording of such an achievement.

Now follows a detailed summary of the particulars of the way I decided to go about the All S Rank run.

Game version played: Original Japanese release
The scores required for S rank is generally more harsh here than in the later retail editions. Vincent moves slower, fires slower and lacks a double-jump which also adds to the challenge in a way that the revised game does not.

Difficulty mode: Hard, no auto-targetting
The highest difficulty setting in the original release. You do not have the option of auto-aim so you better learn that manual targetting! In the post-original releases you can shift between manual- and auto aim at any time. I consider Hard mode in the original release to be harder than playing on the added Ex Hard mode in the later editions because of how comparatively handicapped you are as a player in the initial retail edition.

Segmented run
The game was played one checkpoint at a time with each chapter segment played multiple times until a satisfactory result was achieved and recorded. Sometimes I go the extra mile of overdoing the scores and looking really good while doing it, other times I settle on a more casual-looking result that is good enough for S rank. I believe that a segmented run is the only reasonable way to achieve top grades in every category: You’d have to reset dozens and dozens of time to perfectly finish Chapter 1 in one sitting and even after that point you could easily mess up the run.

PS2 Controller, analog sticks
Dirge of Cerberus supports mouse and keyboard. Aiming with precision is by far easier when using a mouse and is my recommended way of playing the game. Essentially, playing with a PS2 controller and having to aim with the analog stick is yet another handicap that makes this playthrough more hardcore.

Vincent always on Level 1
When a stage ends the player may either use EXP to level up or they may convert that EXP to gil. Levelling up increases your stats and primarily adds a safety net by giving Vincent higher max HP. Remaining on Lv1 is a sign of confidence that you know how to avoid taking a lot of damage.

NO Phoenix Down/Auto-Life
The lifeline of being able to resurrect after a fatal mistake, and enjoying brief invincibility after having your HP fully restored, adds a huge comfort and most players will always have Auto-Life at the ready. Due to aiming for stylish results and receiving little damage I decided that activating the Auto-Life status was beneath the honor of the All S Rank run.

All cutscenes skipped
The stage timer ticks even while cutscenes are playing. The conditions for S rank in the “Time Expired” is especially harsh in the first four chapters, meaning the only reasonable way to play is to skip all cutscenes. Beyond this point I skip cutscenes because I’m only interested in the gameplay portions of the session. Any cutscenes included in the video playlist, with the exception of Omega Weiss’s defeat, were added during the video editing process.
To make the experience more pleasant for whoever is watching the playlist I have also edited out some of the loading times. Similarly, I’ve sped up the footage in places that consist mostly of Vincent running for long stretches.

Hopefully my All S Rankings run will become a common reference for whenever people want to show what expert-level play can look like in Dirge of Cerberus. If you want to follow along on my continuous journeys into Dirge of Cerberus, do check out my research thread on the forum and show your support with a like or two!
Playing the original release is also made harder because it doesn't store chain-damage for Fire- and Blizzard spells for more than five seconds. I swear, there were so many times in Chapter 10 and some instances in Chapter 11 when my hands felt tied because of the chain-storage bug. Thankful that proper chain-storage still works for Thunder magic though.

It pleases me that the final amount of footage is 4 hours 20 minutes 10 seconds. 420 BLAZE IT!

I've shared the news on social media and with my close friends. I personally won't be reaching out to any major sites, twitter accounts or news outlets. The amount of self-promotion I've conducted is already obnoxious and I don't want to annoy anyone with further cries for attention. Let my work be found either by luck or by promotion done by friends and fans.

Due to the game's mediocre and contentious status, I fully expect that my All S Rank playlist will remain a super-specific niche. Then again that is always the kind of work I do.

Even if I take a long break, I shall always return to Dirge of Cerberus in one form or another. There simply is too much left to explore.
Last edited:
On December 19, when I achieved the final cumulative S rank results, there was one thing left that bothered me: The fact that the game could not save my final stats from clearing Ch12-2.

Checkpoint data is saved in the eleven scene0XX files (numbered scene000 up to scene010). Each scene0XX file is composed of six equally sized regions, each region corresponding to a checkpoint and containing data for when you load up that part of the game.

scene008, Region 5 : Ch11-1
scene008, Region 6 : Ch11-2
scene009, Region 1 : Ch11-3
scene009, Region 2 : Ch11-4
scene009, Region 3 : Ch11-5
scene009, Region 4 : Ch11-6
scene009, Region 5 : Ch12-1-1
scene009, Region 6 : Ch12-1-2
scene010, Region 1 : Ch12-1-3
scene010, Region 2 : Ch12-1-4
scene010, Region 3 : Ch12-2, A Finale Chaotic
scene010, Region 4 : Ch12-3, Memory Fragments [Original Event Viewer]
scene010, Region 5: Unused
scene010, Region 6: Unused

In the original Japanese release, clearing the game for the first time unlocks the final checkpoint which is that version's gateway to the Event Viewer. Data is saved into scene010 Region 4 that corresponds to when you cleared the game, how you performed during the final battle etc.


Even though post-original versions lack this extra checkpoint in the main game menu, data is still saved into scene010 Region 4 when you clear the game. Indeed, it is here that the game remembers what your current inventory was and what equipment you had on your three weapon slots. This is what the game draws from when you start up the Extra Missions. Your only way to change your starting equipment in the Extra Missions is to defeat Omega Weiss while wearing different equipment. Needlessly cumbersome. They should have just let you change the equipment while browsing the list of Extra Missions.

The problem with the original Japanese release is that, once again due to poor programming, scene010 Region 4 is only updated ONCE: The very first time that you clear the game. After this, the time stamp, the displayed difficulty mode etc is locked. Defeat Omega Weiss again in Ch12-2 and the memory card will not perform a single update. Because I was doing the All S Rank challenge on a save where I had previously completed the game, I was doomed to not have my final progress saved.

Fortunately, today, I found the solution and I can now clear the game while still having my All S Rank results properly saved into scene010.


When you clear the game for the first time, offset 0xA80 in the file cfg000 (configuration file for Dirge settings, memory capsule acquisition etc) has its 0th bit set to 1. In the post-original releases, this flag is responsible for unlocking the Event Viewer chapter menus and for unlocking the very first Extra Mission. Meanwhile in the original release all it does is denote that you've cleared the game.

By setting this flag back to 0 the game will once again be able to save into scene010 Region 4 when you defeat the final boss. The flag is then set back to 1 and you'll be locked in the same way again unless you edit the save file.

One thing to keep in mind is that data from cfg000 is copied onto the scene0XX files when you play and reach a corresponding checkpoint. So for any given scene0XX file the flag may be updated into the following offsets.

scene0XX, Region 1: Offset 0x48
scene0XX, Region 2: Offset 0x990
scene0XX, Region 3: Offset 0x12D8
scene0XX, Region 4: Offset 0x1C20
scene0XX, Region 5: Offset 0x2568
scene0XX, Region 6: Offset 0x2EB0

This is relevant because if the game notices that even a single active scene0XX region has the flag set to 1, then that is the state the game will operate in. Meaning I had to set the flag to 0 not only in cfg000, but in multiple spots in all eleven scene0XX files. Extra work for me but it's a pretty decent failsafe: Even if values in cfg000 somehow changed, the game can draw from progress remembered when you cleared the scene0XX files and thus your progress won't be lost.

Once the deed was done using McDirge v0.4, I could confirm that setting the flag to 0 worked!

I spent so much of December 19 trying to edit my save file and fixing the "scene010 Region 4"-bug I was honestly slightly bitter even after the All S Rank challenge was done because, strictly speaking, my final performance had not been saved. Now I can safely say that genuine examples of All S Rank data has properly been saved into the memory card without me having to manually change things like the final time stamp, targets destroyed, damage received, shots fired, shots landed etc.

My statement that the game doesn't reward you for achieving S rank in every category was actually a lie (in the sense that I spoke as if the truth was 100% known but I wasn't actually sure) based on a reasonable assumption until just now. I can breathe a sigh of relief and say that since the achievement has been properly saved onto my memory card...there is indeed no reward for performing an All S Rank run. :monster: Nothing at all in the game changes.

At least...that is definitely true for the original Japanese release. I can't speak for the later editions but I can reasonably assume that there are no rewards or special markers that pop up to signify an All S Rank achievement.
Last edited:
Though updating McDirge is first on my agenda, my next gameplay challenge is definitely to speedrun the Extra Missions. I have no idea when I'll get started with this speedrunning project. An updated version of McDirge would come in handy not just for changing your starting equipment for the missions but also for quickly printing out your current Extra Mission record times.

One thing the game should have included is a "total playtime" for all the missions combined. Simply having the display there would add incentive to see how low you could go with the total time, seeing just how many seconds you can shave off the missions.

I would exclude three particular missions from this total playtime, as these always take a fixed amount of time to clear.
Mission #10: Missilebreaker Melee [Fixed at 3 minutes]​
Mission #25: Missilebreaker Deathmatch [Fixed at 3 minutes]​
Mission #27: Vincent the Beast [Fixed at 5 minutes]​

If you look at your completion times in the cfg000 file, the completion time for these missions remain set at value 0 even after you've cleared them.

The way I play the Extra Missions I am competent but relaxed, meaning I am yet to actually speedrun most of them. My total mission playtime in my main DCFFVII International save is 5 hours 22 minutes 46 seconds. Pretty sure I could shave it down to below 4 hours if I put my mind to it.

There is no speedrun leaderboard for the Extra Missions. The only leaderboard is for the main game. By all likelihood I will be the one to set up an organized collection of known record times, whenever the day comes and I actually start speedrunning them myself. Some of the fastest times I've seen remain those by fiel2c, with many of their recorded times being FAR better than my own and even here there is massive room for improvement.

@Makoeyes987 How about beating fiel2c's Stronghold Impervious time? :monster: Looks to be about 17m8s. One day I shall beat this.
Last edited:
*Post was heavily edited after more accurate data sets had been acquired.*

While I was working on the Crystal Feelers section I had to measure which was faster: Running/gliding as Chaos Vincent or performing the dash.

In all versions, Chaos Vincent's maximum running/gliding speed is 100 units/second. The distance between two opposite Crystal Feelers is 800 units (the map is 1000 units across), so gliding from one CF to the other at full speed takes 8 seconds. When pressing the dash button your speed immediately moves up to 130 units and stays there for a good while before Chaos Vincent then halts to a complete stop.

Traversing uninterrupted from one Crystal Feeler to another, both being on opposite sides of the map, these were the frame counts on emulator (60 frames per second). Consistent data was acquired by locking Chaos Vincent so he could only move along one axis.

Japanese Original test
Gliding/Running only: 480 frames
Jumping only: 490 frames
Dashing followed by jumping: 452 frames
Dashing only: 410 frames

Dashing wins by 1.17 seconds. Your average speed from dashing ends up being roughly 117.13 units/second. The jumping-only tests confirm that even though you retain your normal running max speed of 100 for a great deal of the time, there is a brief slow-down from performing a jump. As expected, there is no benefit from jumping after a completed dash.

In the post-original releases you can perform a dash-cancel by jumping or doing melee. This means you don't have to calculate as carefully that your dash animation completes just as you reach the Crystal Feeler. If you're playing the original release and you happen to dash up right against a Crystal Feeler, the time it takes to complete the animation may lose you more time than if you had just ran up to the crystal and struck it.

The post-original versions of Chaos Vincent allow for way more animation interruptions/cancels in general. Not only can you interrupt the dash but you can also (after a few short frames) interrupt the gun-reload animation by jumping! This is definitely not a thing in the original version. There were plenty of times while fighting Omega Weiss when I couldn't evade the needle-shockwave attack because I had been careless and caused Chaos to start a reload animation. Oddly enough the delay for interrupting a rifle reload as Chaos is still very long.

Seeing as you can interrupt the dash in the later versions, does this mean you can travel even faster? No.

DCFFVII International test
Gliding/Running only: 480 frames
Jumping only: 490 frames
Dash-cancelling with jumps: 452 frames
Dashing only: 412 frames

The results for dash-interruptions astonish me. They are exactly the same as if the dash had never been cancelled at all. Clearly the slow-down effects from jumping are just that dramatic. Performing solely dashes remains the winner.

I found dashing to be surprisingly effective during the Omega Weiss battle. If you are simply gliding along in a straight line while the Omega missiles are homing in on you, it is mostly down to luck whether you can avoid them exploding on you and inflicting damage. When I switched to dashing the missiles never hit me. I suspect it also helps to perform sharp turns when doing the dash. This makes sense as Omega missiles can be seen slowing down when they approach a stationary Vincent. Perhaps the sudden dashing-and-stopping aids in confusing the Omega missiles thus increasing the distance between you and them.

I only wish dashing was also effective in avoiding the Omega Bud lasers. :monster: Unfortunately dashing does nothing at all to help against the lasers.
Last edited:
Bug: Infinite Galian Beast, Chapter 3

Something I found while doing a "Use only Galian Beast mode" playthrough in JORG. It exists in all versions and the only example I have found of it is in Chapter 3, "Silent Edge".

Effects of glitch in Japanese Original:
- MP depletion stops entirely, meaning you can retain Galian Beast (GB) form throughout the rest of Chapter 3.
- The glitch stops soon as you transform back into human form or when you clear Chapter 3. MP behaves normally after this point.

Effects of glitch in post-original versions:
- GB form will remain even after the 43-second mark (which is when you, normally, automatically revert back to human), but you'll no longer be able to jump, melee and shoot fireballs once those 43 seconds have passed. You can still open doors and use cardkeys however.
- The glitch stops as soon as you transform back into human form.

How to trigger "Infinite Galian Beast":
You must be in GB form when the battle against the Pegasus Riders ends in Chapter 2 (Ch2-5). Without exiting the current playthrough you must continue onwards to Chapter 3. Make sure that when you speak to the dying WRO soldier that you are once again in GB mode. It does not matter whether you skip the cutscene or watch it all. When the scene is over, Infinite Galian Beast will have been triggered.

The glitch will never work if you load Chapter 3 from the main game menu. The game needs to have in active memory that you just cleared Chapter 2 when you were in beast form. I have only tested the glitch in a playthrough where a memory card is being used. The cutscene with the dying WRO guy seems to be the only cutscene that can help you transition into Infinite GB.

I have not found that getting KO'd (and then resurrected) affects the glitch in any way nor have I found a way to restore jump/melee/fireballs in the post-JORG versions while still retaining Infinite Galian Beast. EDIT: Creating a tempsave during IGB and loading that save also does not change the bug, neither in original nor in post-original.
Last edited:
Finding the Galian Beast Ch3 glitch feels especially good because pretty much all the Dirge of Cerberus glitches up until now have had no exploit potential. In the original version there is potential to use Infinite GB mode for a speedrun while in post-original it is mostly a curiosity. Even if there is a way to restore jumping/melee/fireballs in post-JORG while still keeping IGB, the overall slower movement of GB may be too much of a sacrifice.

The way I did my "Only Galian Beast" playthrough was to create an Easy Mode save, create a Tempsave almost immediately, then edit my Tempsave data to import the following items to my inventory:
- 9999x Hero Tonic Gold (unused item that grants temporary invincibility)​
- 9999x Elixir​
- 9999x Elixor​
- 9999x Hi-Potion​

This way I could dash through almost the whole game, spamming invincibility status with Hero Tonic Gold and restoring MP with Elixir thus allowing me to stay in Galian Beast mode. So much fun! I love cheating in video games. :wacky:

The idea for the playthrough wasn't born out of the desire to just plow through the game in a casual manner though. I wanted to investigate how the game acts and how data is written into the save files.

In a fluid single-sitting playthrough, start to finish, these are the transitions that force you out of Galian Beast mode.
- The beginning of every chapter.
- Ch1-4: Office fight.
- Ch4-3: Shelke boss fight.
- Ch4-4: Azul boss fight.
- Ch5-1: Sahagin cutscene almost immediately at the start of the stage.
- Ch5-3: Any beast mode from Ch5-1 will have been cancelled.
- Ch5-3 again: The cutscene that gives you the first Omega Report.
- Ch6-3: Transitioning to this area/segment cancels beast mode.
- Ch9-5: Azul boss fight.
- Ch10-3, Ch10-4 and Ch10-5: Transitioning to these areas/segments will cancel beast mode.
- Ch11-4 and Ch11-5: Transitioning to these areas/segments will cancel beast mode.
- For obvious reasons Ch11-6 and Ch12 don't allow GB mode.

I could not find any cutscene where Galian Beast stuck around to make the scene hilarious/awkward. The game is consistent with making sure that cutscenes only use human Vincent, with the obvious exception of scripted transformations.

There is not a single gun turret that Galian Beast can use. You can't progress in Ch10-3 without using a gun turret so you are forced to transform back into a human.

Any given checkpoint from the main menu will NEVER initialize Vincent in beast mode, no matter how much you used Galian Beast during the fluid playthrough. This is somewhat humorous because the scene0XX files actually do save data that indicate Vincent is in beast mode and many of them even save the "p001" string that denotes the Galian Beast model. So all these addresses that indicate Vincent's beast status effectively become read-only values, never writing into the final checkpoint load-up.

The model strings (p000 = Vincent Valentine (human), p001 = Galian Beast) were actually what sparked this playthrough to begin with. Among the scene0XX file regions, 26 out of the 56 used regions will include the current character model string. Cait Sith's model string, n004, is never saved. Every scene0XX that includes the p00X string also include a "bzd" (think of it as "basic zone data") string which changes depending on which zone is being loaded. Example: String 'bzd048:2006/ 8/17:19:38:56' for Kalm in DCFFVII International.

The bzd data actually appears to be useless, as when I tested changing every value in this approximately 128-byte sized region to value 0 nothing at all changed when I loaded up the corresponding checkpoint in-game. But the data that follows the p00X string not only impacts if the checkpoint is able to load up the player UI and player behavior correctly, it will also DISPLACE addresses later in the scene0XX files.

Among other things each scene0XX file region contains your current stage results and your current "cumulative results" as of the last chapter/stage finished. But the location of these addresses will shift depending on if the scene0XX region also contains the p00X string and its associated data region of 100 bytes. Regions with the p00X data region will displace these addresses with an offset of +100 bytes (0x64) compared to normal. Documenting the addresses in my Excel document becomes something of a hazzle because of this.

Before my beast mode playthrough I was only familiar with the p000 string showing up but now I know that most regions (that include the bzd and p00X regions) will change this to the string "p001" if you ended the previous segment/checkpoint while in Galian Beast form. The scene0XX regions also remain consistent between playthroughs and game versions as to which ones include the bzd/p00X data regions and which ones don't.

Seeing as I haven't tested erasing every single bzd region, it is still possible that some areas might not properly load up if these values have been changed to 0 in your memory card save. The p00X data region has already confirmed how important it is for checkpoint load-up, since most player behavior won't load at all if this region is deleted. Curious then that so many scene0XX file regions make do perfectly fine without the p00X data regions. 30 out of the 56 used regions (counting the unused scene0XX file regions the total is 66 regions) don't include the bzd and p00X data portions and the game is able to load all these checkpoints no problem.
Last edited:
Revisited the reference data from the DC Complete Guide that I posted earlier in this thread page.
Here is what the Dirge of Cerberus Complete Guide says about range-based accuracy on page 40.

※ 距離の目安は、カーゴの一辺が1.2m、ドラム缶の高さが1.5M​

Google Translate
HIGH: Hit accuracy at a distance of 38m or more.​
MED: Hit accuracy at medium distances from 12m to 37.99m.​
LOW: Accuracy at close range (0-11.99m).​
* As a reference for the distance, each side of the cargo crate is 1.2 m and the height of the drum can is 1.5 m.​

Looking at the in-game values, I confirmed that each cargo crate side is 12 units long and the drum cans are 15 units tall. Meaning that to translate into meters, we just have to divide the floating point values by ten.


Vincent's canonical height is 1.84 meters. Comparing Vincent's in-game height with the drum can shows consistency with canon.

The area map in the menu scales differently from area to area, meaning there is no universal reference that may help spot which lengths correspond to the three ranges: Short/Low, Medium and Long/Far. Of course even if they *were* universal distance references on the 2D-plane, they'd still be a flawed reference because of the height dimension. Still, I'd have appreciated a constant scaling scheme with each area map as viewed in the menu.

Chapter 9 elevator: Azul battle

Chapter 3: Edge plaza
New WR holder for Dirge of Cerberus NTSC-U/PAL Any%: NT1_Evolution with a time of 1:36:42, beating GocuComeBack's old record by seven seconds.

NT1_Evolution is employing the new dash-jump-gun-cancel technique to diminish air time and thus shorten the time between dashes. The technique is way more involved than the simple dash-jump but it looks effective when mastered.

Another new submission to the leaderboard from two weeks ago is by CityHall who is now on 5th place in the NTSC-U/PAL Any% category. I think this is the oldest video on the board to include the dash-jump-gun-cancel strat.

These are the first submissions that the category has seen in a year. ZZ_Yoshi did submit a new Ex Hard Any% record 7 months ago but other than that the board had been silent for almost a year.
"Erroneous Target Defeat" bug in Ch4-2

NT1_Evolution highlighted an interesting bug. It happens in Ch4-2, WRO HQ, as the player progresses from the 2nd floor up to the 3rd floor.

Observe the bottom left of the screen as Vincent steps over the lowered EM barrier and moves up to the third floor. Time stamps should take you where you need.
- NT1_Evolution practice video
- Same thing happening in my Ch4 All S Rank video

Your current killchain is drained/reset and then a new 5-second chain counter begins, as though you had just defeated an enemy. In fact that is exactly what just happened: The game thinks that you've just defeated an enemy!


When you cross this line at the precipice of the stairs leading up to 3F, an event occurs up on 4F and 5F (top floor).


The event is an explosion on the opposite side of the building that will kill two WRO members who are standing on the top floor and one common DG soldier who is standing on 4F. The explosion happens without the use of an actual drum can, so there is no way for you to activate this explosion early (yes, even if you destroy the actual drum can that stands on that very spot). While the fourth floor DG soldier is physically present from the very start of the level, the two WRO soldiers only spawn at the moment that you trigger the explosion event!


The WRO soldiers will dramatically fall down to the 3F stairs just below. It can be difficult to realize at all that a DG soldier was just defeated, as from the 2F you can never get the soldier into view (other than seeing its gun when the enemy reloads). The DG foe is only quickly thrown up in the air by the explosion and then disappears from view again as the defeated enemy fades away.


Normally, events that happen outside of the player's control will not affect your kill count, your chains etc. But the defeated DG soldier in this obligatory event counts as a valid kill, meaning not only is your "Targets Destroyed" count incremented by 1, but the killchain mechanics are active as well. You even gain 5 EXP from this automated kill!

If more than 5 seconds have passed since the last increment to your chain, triggering the explosion will USE UP your most recent killchain and the 5-second counter starts over again due to the game thinking that you just defeated an enemy.

But if less than 5 seconds have passed since your last kill, the explosion-event will actually add in continuing the build-up of your chain! That is what happens in the sniper scope image I posted above: As I casually walk and trigger the event, my 11-chain is incremented to a 12-chain.

For speedrunners this poses a timing challenge. If they don't time the lowering of the EM barrier and the destroyed targets up until that point, the 5-second countdown will expire and the event will cancel their old killchain rather than help in building it up.

With a well-placed Fire spell shot you could kill this fourth-floor DG soldier early and not have to worry about the explosion-event but that's a lot of precision and MP for just one kill.

From the start of the level this DG soldier will be under fire by the WRO and receive damage by increments of 6 and 8 per shot. By the time you reach the gate on 2F, their HP may have been lowered from 280 down to 220 or even 140 depending on the RNG. It's pointless to wait around for several minutes just so this soldier can self-die under WRO fire. If you let this DG soldier be killed by the WRO then their defeat won't affect your "Targets Destroyed" counter or your killchain, so at least here the game's behavior is consistent.

As another pointless aside, when Ch4-2 begins there is a hidden DG soldier on the corner of the top floor. Unless you use cheats, like my super-jump below, you will never see this soldier.


Much like the fourth-floor DG soldier, this one exists solely to fight it out with a WRO member on the opposite end of the same floor. But unlike the 4F soldier, the player has no chance of ever seeing the 5F soldier. By the time you reach the top floor, this soldier will have spawned out of the scene even if they were not defeated by WRO fire.

With a well-placed Fire Lv2 shot (Fire Lv1 also works but it is needlessly difficult) you can defeat this soldier while you are standing on 3F or 4F and thus gain a rare kill, maximizing your Targets Destroyed count.

This reminds me of the obscure Omega Bud kills I achieved in Ch12-1-3 with help of Blizzard magic. :monster:
Last edited:
Found another instance of the glitch where your actions performed in the point-conversion screen and in the shop gets deleted. The first one I found was in the transition from Ch11-6 to Ch12-1-1. If you are playing the original Japanese version or the North American release, accessing Ch12-1-1 from the main menu or turning off your console during this section will result in your aforementioned use of EXP and gil being deleted. All the EXP you gained in the stage before is essentially deleted. The bug was fixed for the PAL and International versions so that the game now included an extra save prompt just as you leave the shop.

I found now that this bug is also true for the transition from Ch8-2-3 (Rosso fight) and Ch9-1 (Shinra HQ, only cutscenes).


Leave the game during Ch9-1 or enter here from the menu and your actions after defeating Rosso will have been deleted...if you are playing the original JP version and the NA/NTSC-U release. The bug was fixed in PAL and International by adding a memory card save right as you leave the shop.

In other words...I was subjected to this glitch in my All S Rank playthrough without knowing it! I end Ch8-2 with 81120 gil while also having sold all my Potions and my Hi-Potion. But when I load Ch9-2 I'm back down to 63860 gil, same I had before EXP conversion and shopping, and my Potions and Hi-Potion are back. All because I was meticulous and recorded myself clicking on the cutscene-only checkpoint that is Ch9-1. Should edit my video commentary to reflect this but I'm not in the mood to revisit the subtitles just yet.

The reason I got around to find this now, over four years later, is because the question was asked in the DoC speedrunning Discord channel just how many save-cancels that a speedrunner has to perform. The overall loading time of the game is faster if you don't have a memory card in your console but the game will nag at you with pop-ups asking "You REALLY don't want to use a memory card and save?".

Looking at playthroughs of the North American version of Dirge, the most commonly speedrun version, these were the number of save-cancels per chapter.

5 : Ch1
5 : Ch2
4 : Ch3
4 : Ch4
4 : Ch5
4 : Ch6
2 : Ch7
3 : Ch8-1
3 : Ch8-2
4 : Ch9
5 : Ch10
6 : Ch11
3 : Ch12-1
1 : Ch12-2

5+5+4+4+4+4+2+3+3+4+5+6+3+1= 53

A total of 53 save pop-ups when you don't have a memory card inserted.

As we now know however, PAL and International have +1 save instance when Ch9 starts and +1 when Ch12-1 begins, thus fixing the aforementioned glitch of point-conversion and shopping being deleted. Presumably then those versions have 55 save prompts in total.

*EDIT, 2021-09-23: There are in fact 55 save prompts in all versions. I missed the two save prompts that happen during Stage Results screens if you are playing the two first releases (J1.0 and NA). For PAL and INT, these save prompts were moved to after the Shop section when a chapter ends. This makes sure that both the Stage Results screen and your shop actions are saved, instead of only the former being saved.
Last edited:


astray ay-ay-ay
I'm so frustrated that most games I've been trying to play recently are such a pain in the backside in terms of controlling the characters. Either I have an only brain cell or I need to buy a new gamepad. DC has been a positive experience so far, but I can't move my camera freely for some reason, and anyway, moving it manually in combat is just uughhhh so inconvenient. I made it to the first boss, but the controls keep killing me. Literally. So guys, I appreciate your affinity for this game, I probably won't be able to advance much in here.

On a positive note, Kalm looks pretty big for a small town. Makes you think of the scale of the other FF7 settlements.
Top Bottom