DCFFVII Research Thread

Investigating Extra Mission #12, "Vincent the Destroyer", has been great fun.

Your goal here is to destroy 100 Aerial Mines (henceforth sometimes referred to as 'AM' for singular and 'AMs' for plural). If Vincent gets below a certain elevation, a 5-second countdown starts. Unless Vincent gets back to a higher elevation the countdown will eventually reach 0 and the player will get an instant Game Over. Not even an auto-life status (only possible in this mission via cheats, as Phoenix Downs are not available) could save you from this Mission Failed status if the death-countdown completes. Thankfully the time limit is always 5 seconds when the countdown is initiated, so the game has no memory of how long you've been in this death-timer state before.

The AMs don't do a lot of damage when they explode on Vincent. The real threat is that they pursue the player and are likely to knock them down from the platforms, often making it impossible to regain the high ground before a Game Over. *something something Star Wars meme* Normally the mines are slow-moving but at times they can achieve dangerous speeds.

The most common method to cheese this mission is by standing on this hill that connects to a platform. The path of the AMs is restricted to the platforms so they can't follow you down. The hill is high enough that you can easily reset the death-countdown by jumping. There is even a Goldilocks zone on the hill where you can stand just out of the enemy's reach but still not trigger the timer.

Some platform sections are damaged and the Aerial Mines can actually fall down these. But when standing on this hill that's not even a problem. The AI will prioritize helplessly trying to get back up on the platform rather than choosing to pursue Vincent!

aerial-mine-stuck-on-ground.png


Another method to cheese the mission is by standing close to this damaged platform section:

cheese-method-b.png

The floating mines will always fall down to the ground if Vincent stands close to this section. The sight is quite humorous as the mines fall down like Lemmings and stay huddled up against the wall. From here the grounded AMs can easily be destroyed.

The Goldilocks zone for this method is pretty wide and easy to find. Don't stand so close that Vincent touch the AM hitboxes but don't stand so far away that they are able to dodge the trap in the topography.

The spot Vincent is standing on, coordinate H-9, is near a dead-end. Only ONE Aerial Mine will ever spawn here. Specifically it is either the 8th, 9th or 10th Aerial Mine that spawns at H-9. Once the player is past this point in the mission, the cheese method is free to use without getting ambushed.


The first 13 mines spawn in predetermined locations and all of them are in the southeast part of the map. The mines are divided into groups so that the next group will only spawn when the previous one has been cleared.
Group 1: 1st mine
Group 2: 2nd, 3rd and 4th
Group 3: 5th, 6th and 7th
Group 4: 8th, 9th and 10th
Group 5: 11th, 12th and 13th

After this point it is random where the next series of Aerial Mines will spawn. There is a limited set of spawn locations, placing some limit on the randomness, and many of the initial 13 spawn locations are excluded (including the aforementioned H-9 spawn point). You should expect to see mines appearing in all parts of the map though and not just the southeast section.

A maximum of 10 AMs can roam the field at any given point. When one target is defeated, a new one will instantly pop up in one of the randomly selected spawn coordinates. Therefore it is valuable in a speedrun to quickly spot a target and destroy it so that the next one may be created.

More analysis to come in my next post.
 
Last edited:
No matter how weak your attack, indeed even if a bullet cause 0 damage, the Aerial Mine will explode and count as defeated. If we look at the actual HP of each Aerial Mine that shows up on the field, they all have 20000 HP respectively. Weird, huh? In normal gameplay you will only be able to make a small dent in this HP pool.

Manipulating the target's remaining HP such that our next shot will definitely deplete it to 0 reveals two things:
1) Destroying the target suddenly adds to our killchain, whereas normally this does not happen for these targets.​
2) The Aerial Mines no longer explode but fall down and slowly fade away.​

EM12_aerial-mines-alternate-state.png

The only downside is that the next targets will not spawn until the models for these defeated Aerial Mines have faded away, so that adds a few seconds of waiting. You still gain points for destroying the mines even in this unintended fashion.

Certain other targets in the game also use a large HP pool to become exempt from the mechanic where destroying them adds to your killchain. The Shrikes in the Train Graveyard (Ch8-1-3) notably do not increment your chain when they are defeated. This is because their actual HP is 20550, but they count as defeated so long as you deplete 550- or more HP. The Shrikes, much like the Aerial Mines, are programmed to go through an explosive death animation under normal circumstances. Bypassing the large HP pool and *actually* lowering the Shrikes' HP to 0 means that they will add to your killchain and the Shrikes will no longer explode. Sometimes the Shrikes will just disappear, other times they'll instantly fall to the ground and fade away.

The major downside from bypassing the intended HP depletion mechanic is that the 'Boss HP' bar at the top right will not deplete and so you end up making the chapter unbeatable. You'll clear the area of Shrikes and afterwards the game won't progress. No matter how powerful you make yourself by normal means I do not think this glitch can be triggered without cheats.

Shrikes adding to killchain
ch8-1-3_shrike-killchain.png


Fallen Shrike that skips the explosive death animation
ch8-1-3_fallen-shrike.png

Clearly then the game usually intends that so long as the enemy HP is lowered to 0 then their defeat will boost your chain. There are exceptions of course. The final boss, Omega Weiss, has its HP lowered to 0 by your final attack but your chain is not incremented. The Pegasus Riders (Ch2-5) are weird because they have a huge HP pool just like the Shrikes but they still add to your chain. This may or may not be because of how the actual soldiers on top of the hover bikes are programmed. Someday I'll look into it.

Going back to the Aerial Mines in Extra Mission #12, their huge HP pool also helps partially explain an occasional bug in this mission: There's roughly a 1% chance that your successfully landed shot, be it a bullet or a magic shot, will not cause the target to explode. What happens here is clearly that the game fails to call the explosion-death animation. Since the Aerial Mine definitely has a TON of HP left, that also explains why even a super powerful attack can be subject to this no-explosion glitch. The strength of your attack just doesn't matter here.

Funny thing though is that these AMs are the first targets I've confirmed to receive DOUBLE magic damage. Not that I've searched high and low for enemies with extra low magic defense but none of the bosses take double magic damage. If you attack one of the Aerial Mines with a Thunder Lv3+Manasoul+0-chain attack you'll deal 4892 damage (+50 for the mine's self-explosion damage). Most enemies would only receive 2446 damage from the same attack.
 
Last edited:
If you destroy an EM12 Aerial Mine while Vincent is in the death-countdown state, then your "Targets" counter will not have its value incremented. The consequence is that, even without using cheats, you could use this "Invalid Target" mechanic to learn that only a maximum of 254 Aerial Mines can spawn in this mission!

By going out of your way to destroy (at least) 155 AMs while in the countdown state you will render the mission unbeatable. Since there will only be 99 targets left you don't have enough to gain the 100th target point and finish the mission.

We can gain some understanding of what's happening here by looking at the "Actor Combatants" region in memory.

Snapshot from memory when Extra Mission #12 is active, before the first target has spawned:

EM12_combatant-memory-region_a.png

Each actor that counts as a potential combatant have their own 208-byte region that shows things like max HP, max MP, DEX/DEF/STR/INT stats, remaining HP, remaining MP, status effect, max walking speed, max running speed etc etc. Accordingly, objects like cargo crates, drum cans, item bags etc are not listed here since they are not "combatants".

This Actor Combatants section changes location in memory as you go through chapters and modes. This is where I go when I want to increase Vincent's maximum running speed and when I want to make him invincible.

The string that identifies Vincent is p000 and is typically the first actor item in the Actor Combatants region. If you transform into Galian Beast then the p000 string is replaced with "p001" and a few other values change. In other words the game doesn't reserve a new 208-byte region for Galian Beast: It reuses the same region in memory and adapts to the player's current state.


Finding unused combatants/targets/enemies is not uncommon and the game indeed never does anything with the first Aerial Mine (e045) in the region. Here is the same location in memory when the first floating mine enemy has spawned:

EM12_combatant-memory-region_b.png

As you destroy each target, new 208-byte instances of the e045 actor will continue to occupy memory downwards. You may find this actor data overwriting tons of stuff that looks important. That was actually my initial reason for investigating this Actor Combatants region: I wanted to see if a newly spawned Aerial Mine would overwrite important data and cause the game to glitch. It did not. The only data overwritten is old junk data.

What saves the game from overwriting important memory sections is that no more than 256 actor objects (256 * 208 = 53248 bytes) are allowed to exist here. I do not know if the limit of 256 actor objects stem from the programmer specifically declaring an array of that size or if it stems from an 8-bit variable that limits the sizes to 0-255.

This is why only 254 Aerial Mines can spawn in Extra Mission #12. Actor Object #0 is Vincent, Actor Object #1 is the unused Aerial Mine, and objects #2-#255 are the available remaining spaces for spawning targets.


A secondary reason for me looking into this was the curiosity that the game wasn't reusing Aerial Mine addresses. Normally when the game has re-spawning targets, an older instance of that actor object will be reused. So long as the defeated target has had their 3D model fade away, that 208-byte spot in memory is free for the taking. The HP value previously lowered to 0 will be restored back to max HP, a few other values get set and the new enemy will be good to go.

"The HP value previously lowered to 0" is the key condition here. The reason Actor Object spots are not reused here is because the defeated Aerial Mine's HP was never reduced to 0. It stays at the ranges between 10000-20000, depending on how much damage you caused when destroying the mine. Amusing how the oddities of this mission's main enemy keeps coming back to the fact that they start out with 20000 HP.

Whenever I used cheats to actually lower the target's HP to 0, that spot in memory would actually be taken up by the next Aerial Mine to spawn! No new 208-byte section was occupied. Through that method I had increased the total number of targets that could ever spawn in the mission.

The programmer(s) definitely went with "good enough" when they programmed EM12. Good enough it very much is. No player will ever encounter the spawn-limit of 254 unless they specifically try and find out about it. It makes perfect sense then to call it a day and spend precious time on programming other parts of the game that may have required more urgent attention.
 
Speedrun of Extra Mission #12: Vincent the Destroyer

RTA: 01:57.884
IGT: 01:58



The mission had FAR LESS randomness than I anticipated based on my initial research on emulator. So much so that each session plays out more or less the same, so long as you destroy clusters of enemies in the same order. This drove me to repeat the stage until my record got decently optimized.

I camp out in the same spot that I've done since years back, on a thin platform at the approximate center of the stage. Back then it was a safe, casual playthrough with a machine gun, Auto Reloader and Recoil Limiter to render it into a monotonous "Just push the trigger" affair until the challenge is over. A speedrun requires more awareness of your equipment though.

Sight Support: Automatic​
W1: Sonic Griffon + Nova L Barrel + Sniper Scope + Manasoul + M Adjuster Gamma​
W2: Ultima Weapon + Nova L Barrel + Materia Floater Gamma + Manasoul + Thunder Materia (Thunder Lv3)​
W3: Feather Hydra + Nova L Barrel + Sniper Scope + Auto Reloader + M Adjuster Gamma​

Thunder magic creates a protective wall from the Aerial Mine explosion. Therefore even if you blast Thunder on an Aerial Mine that is hovering right in your face, you still won't take any damage or knockback from the exploded target. When you're using normal ammo you need to be much more aware of having good distance between yourself and the target.

My initial equipment setup had Thunder Lv3 on all three weapon slots but this had to change. My machine gun was missing too many shots, so Materia Floater Gamma was replaced with Sniper Scope and the Thunder Materia was replaced with M Adjuster Gamma. On the rifle side of things, I initially favored the Velvet Hydra because of its large magazine but I was losing too much time on its slow firing rate. Feather Hydra is the fastest rifle but with a small magazine, so the only way to gain time with it was by swapping Manasoul with the Auto Reloader. Without any automatic MP restoration active, I figured it just as well to remove the materia equipment and focus on accuracy with the rifle as well.

The mission gives you the same amount of ammo as in EM07, here too without any ammo drops in the actual stage.
180 Handgun Bullets (60 shots)
300 Machine Gun Bullets
60 Rifle Bullets


Automatic aim saves you the most time but you still need to manually aim a lot in order for the game to lock onto far-away targets. You have to accept the limitations of Automatic aim since it's not good at hitting fast-moving target. When an enemy moves fast and does so perpendicular to your aim, auto-lock will always fire where the enemy used to be rather than where they ended up in the next few frames. Semi-Automatic is better at compensating for this stuff since the player is allowed to adjust the angle/trajectory of the shot themselves. Over time though, fully automatic wins out in terms of efficiency.

The glitch where a successful shot does not cause the Aerial Mine to explode appears to happen more frequently on console, much to my frustration. Feels like it happens roughly 3% of the time on console, compared to the 1-2% chance on emulator. Could just be the fact that I now have a larger sample size on console.
 
Last edited:
Speedrun of Extra Mission #14: Vincent the Grappler

RTA: 05:13.280
IGT: 05:13



Defeat 100 targets by using only melee. Magic can still knock back foes but will always cause 0 damage. Even if bullets were available, they will only cause damage if you strike the target's weak point but this is useless since the damage-per-bullet will always be 1.

Camera Speed is set to just one step below maximum in order to allow fast rotation and quick detection of enemies. Slight motion sickness may be experienced by the viewer because of the camera spinning around so much.

Sight Support: Automatic​
W1: Bayonet Rifle + Gale S Barrel + [Empty] + Manasoul + Power Cross Gamma​

Bayonet Rifle and Power Cross Gamma increase the power of your melee attacks. Without Power Cross, Guard Hounds will always take at least two melee strikes to be defeated when you have a 0-chain damage booster. The equipped Manasoul has no purpose. The equipped barrel also doesn't matter. Guard N Barrel would have provided defense but if the enemy lands enough hits for this defense boost to be relevant you've already lost the speedrun.

The following targets will appear and have the following HP pools.
*Enemy.....................HP*
SOLDIER...................200
Guard Hound...............60
Gargoyle..................400
Heavy Armored Soldier S...300
Cactuar...................100


The Gargoyles in this mission can actually use melee attacks. Pretty sure this is the earliest instance in the game when the Gargoyles can be seen using melee.

The initial 17 targets are predetermined to a huge extent, meaning these can be memorized. From this point there is heavy randomness and the only way you'll get a decent time is by having great luck with enemies spawning near you. You'll also lose time to bad RNG if an enemy decides to not instantly track you down. It is much preferable that an enemy runs straight to you, rather than them staying idle, running in zig-zags or even running past you. A maximum of 5 targets can roam the field at any given time, plus the one Cactuar that can show up. Defeating enemies quickly is key so that the next target may spawn, though the Cactuar is exempt from prompting another enemy spawn when defeated.

The Cactuar will spawn in front of the church doors either when you hit 40, 50, 60 or 70 defeated targets. I have not checked the game code to confirm the probability algorithm but the rarest instance was for the Cactuar to show up at the 70-score mark. It is possible for the Cactuar to never show up but this is extremely rare, to the point that I have only achieved this on emulator.

The defeated Cactuar drops a cardkey that is intended for use either on the barriers to the southeast (near the church stairs) or far to the west. The former holds a capsule for the Extra Features and the latter holds two item cases, each one holding a Limit Breaker. Picking up the Limit Breakers would hurt our speedrun and wasting time on beast transformations is out of the question.


My casual EM14 record was 06:47 and it's been nice seeing it shaved down to 05:13. My previous speedrun time of 05:17 was achieved pretty quickly and it took about two hours of attempts after the 05:17 before I got my 05:13. Not much was improved in my gameplay performance, I just had to play until RNG was enough in my favor. The sheer mission length and the heavy RNG elements makes it far less satisfying to speedrun than EM12, Vincent the Destroyer. EM14 is decently enjoyable for a "Defeat 100 Targets" mission but I wish there was more room for optimizations and strategies. A record below 5 minutes is possible but I highly doubt we'll ever see a record below something like 04:30.


The main thing I take away from my recording is that I need to mash the melee button more fervently. There are so many combos that could have been executed faster. My instinct is to carefully avoid button mashing because you can easily lock Vincent in an additional melee animation that then mess up your positioning and recovery. No other mission has motivated me to do the dash-melee combo this much so that's a refreshing element. The dash-melee uses an attack animation that was part of the main melee combo in the original Japanese version and usually it's not the melee technique I invoke.
 
Last edited:
Speedrun of Extra Mission #15: Red and Blue

RTA: 00:08.609
IGT: 00:08



Sight Support: Manual​
W1: Ultima Weapon + Gale S Barrel + [Empty] + Power Booster Gamma + S Adjuster Gamma​

Camera Speed and Sight Speed is set to maximum sensitivity so to diminish the amount of time spent moving the camera around.

Doing this mission with Automatic or Semi-Automatic aim may grant you an IGT of 10 seconds. It is certainly possible that this auto-aim method may even give you a record of 00:09, both in IGT and RTA, but I do not believe this strategy can get you below nine seconds.

To reach the 8-second range you need manual aim and frequent critical hits. With the above equipment you need a minimum of 6 hits on Azul and a minimum of 2 hits on Rosso. Their hitbox defense data is the same as their story campaign counterparts (Ch8-2-3 Rosso and Ch9-5 Azul) so I could actually use my Damage Calculator to confirm potential damage outputs. The big difference is the low HP of the Tsviets in the Extra Missions, here with Azul at 2500 HP and Rosso at 2000 HP. Of course, their "actual" or "internal" HP is 12500 and 12000 respectively, as part of the game's quirky method of making these targets exempt from the killchain-boosting mechanic (see the Aerial Mines example in earlier post).

The enemy starts out at just below 12 meters away, and any distance shorter than 12 meters just so happens to be short-range. Thanks to S Adjuster Gamma the final trajectory-deviation of bullets is exactly zero. All three bullets fired from an Ultima Weapon shot will converge on the exact same point and none will diverge from the path you aimed at.


Potential damage outputs on Azul per shot from W1 setup:

EM15_azul-hitbox-damages.png

In the video I start out with a shot on Azul's arms, causing 306 damage. I miss the next shot and then I perform five critical hits in a row. You need to land at least 6 shots in order to defeat Azul. If I had equipped the Nova L Barrel then only 5 shots would have been required but the fast firing rate of the Gale S Barrel compensates for its relatively weak output.

Starting out by attacking Azul is important for two reasons. Firstly, he is the most stationary target and will not move during the first 3-4 seconds. Rosso's AI is such that when you move the aiming reticle over her, she will usually move aside. Azul's AI does not have this complication. Secondly, it is after those first 3-4 seconds that Azul's giant rapid-fire gun will go off and quickly stun- and knock back Vincent. We can afford Rosso's weaker bullets in our direction, so preventing Azul from initiating his attack is a high priority.

By moving closer to Azul while firing the shots I am able to compensate for the recoil of Vincent's gun: My reticle is being moved up and up, but getting closer to Azul adjusts the angle so that I still land the critical hits.


Potential damage outputs on Rosso per shot from W1 setup:

EM15_rosso-hitbox-damages.png

My record session misses the first shot, then lands two critical shots. The game displays a final damage of 2032 on Rosso, which is equal to 1524 (three critical bullets) + 508 (one critical bullet) damage. It is a near-certainty that all three bullets in the second shot did land, but because Rosso's defeat condition had already been fulfilled (2000 HP depleted) the game does not display the superfluous damage from the last two bullets.


(See next post for deeper analysis of IGT and RTA in EM15)
 
Last edited:
*Continued from previous post for ease of reading and browsing*

The video may not look pretty but it is the result of over two hundred attempts. I don't think I ever got a perfect streak of firing the minimum number of bullets and still getting a good completion time. When the mission starts there is a huge time variance in regards to when the game starts registering inputs from your controller's analog sticks. This variance of ±0.2 seconds (roughly estimated) destroys many attempts, leading to either exaggerated camera movements at the start or even no movement at all.

Thankfully, variances in the game's loading times were almost nonexistent for the mission's ending. From the moment that the final target is defeated (regardless if that target was Azul or Rosso) and up until the save icon appears in the upper right corner, the loading time on console was almost always 5.472 seconds. That's 164 frames on a 29.97 fps recording. No matter how the mission is completed, no matter where you are aiming or what you are doing, this time of 5.472 seconds was highly consistent and usually only changed by adding one frame or two. EM15 is thus way simpler than EM02 (Widow Missileer), a mission where we had to move Vincent- and the camera around in order to save 0.1 seconds. There are no such optimizations to be found in EM15, at least according to my cursory research. What probably helps here is that EM15 takes place in a small area and with no explosion effects to accompany the targets' death animations.

My RTA mission time can be expressed thusly:
8.609 = 3.103 + 5.506

Bad luck granted me one extra frame at the end, not giving me the usual 5.472, but the initial time spent on actual gameplay was decent. The 5 1/2 seconds that the game spends waiting for the final target to start fading away puts quite the hard limit on this mission's completion time. Is an RTA below 8 seconds possible by a human? Maybe. It's hard to say without using TAS methods on this mission. But I will say that I do not think a record below 7 seconds is possible, neither in RTA nor IGT. Not without glitches of some kind.


EM15 rewarded me with a frustration that was bound to happen at some point. In my initial attempts, I did not use an external stopwatch to optimize my IGT. In those sessions I managed to get an RTA of 8.642 seconds, but my IGT remained at 00:09 instead of moving down to 00:08. Then once I started the stopwatch technique, I got an RTA of a paltry 8.843...and the game chose that I had earned a 00:08!

MY FEATHERS WERE RUFFLED! It was my desire that the final record video would mark the transition to 00:08 but now that had been ruined. I considered uploading both the 8.642 and 8.843 into one video, to act as a good example of the difficulties with IGT, but that felt too cumbersome. So I decided to make due with whatever my next new RTA record would be and upload that. Silly me still used an external stopwatch when I got my 1-frame improvement of 8.609, even though that was completely extraneous at that point, ergo why you see me waiting by the mission-description card at the start of the video for so long.

I will never know if the game internally decided that I had an IGT of 00:09 or 00:08 in my final record but I know at least that I technically *qualified* for the better IGT of 00:08. Fortunately your best IGT is never overwritten by a worse completion time (unless your previous record was 00:00).

EM15 "Red and Blue" is a great example of the challenges with both IGT and RTA. Without a timer on the screen, and without image-analyzing software to initiate and stop the timer automatically, competition in short missions like these is a particularly blind affair. I found multiple times that the actual RTA of a recording was better- or worse than what my guts told me. One day I will become professional enough and actually stream my attempts online, timer included, but the ultimate dream is to have a program in conjunction with your recording software that can analyze frames in the game and set the timer for you to at least give a reliable impression down to the milliseconds of what your actual real-time achievements are.
 
Last edited:
Very brief history of Dirge of Cerberus speedrunning: The "Main Game"

Starting with the main game, what I like to call the "story campaign" (and "Single Player" if I'm talking about the original JP release), the oldest run available online is by French runner Daitsukii. They achieved an RTA of 2:20:08 on October 29, 2013. They were racing against a time of 2:29:37.5, which must have been the PB of either Daitsukii or somebody else. Daitsukii remains the only person to have speedrun the PAL/European version of Dirge, which on the speedrun.com leaderboard is grouped together with the North American version.

Worthy of note is that Daitsukii is not employing dash-jumping to traverse the stages faster. This was also true of their contemporary, sk84uhlivin, up until and through the first quarter of 2014. The oldest available recording by sk84uhlivin is their third speedrun of the game, from 2014-03-30, and here dash-jumping is not used. But by the time of 2014-04-12, as sk84uhlivin is on a roll lowering the WR, we finally see some dash-jumping for faster movement albeit sparingly so. Four days later he broke the 2-hour barrier and there was indeed more frequent use of Vincent's famous HEGH-HUH HEGH-HUH HEGH-HUH movement.

The WR would be lowered a few more times by sk84uhlivin during the second quarter of 2014. In June of 2014 he achieved a respectable 1:49:57 which was then uploaded to YouTube. This marked the ending of sk84's time with the North American version of Dirge. They would briefly run the JP International version in April 2015 but this became the last time they did speedruns of DoC.

The runs done by sk84 inspired NT1_Evolution and DynamiteDanTNT to run the game. The trio conversed via stream chats and Skype DMs back in those early days as there was no formal place where Dirge runners gathered. By July 2014, runners had begun playing the game WITHOUT a memory card as this was shown to result in shorter loading times.

Normal Mode Any% across game versions was the only category that had been run thus far. This changed in 2018-03-28, when ZZ_Yoshi submitted their first run of Ex Hard Any%. Ex Hard has continued to be a niche with primarily ZZ_Yoshi at the helm and I hope it eventually gets more attention.

NT1_Evolution in particular would carry the torch in an important way by introducing more killchain-optimization routes and by forming the Dirge of Cerberus Speedrunning Discord in 2018-07-26.

GocuComeBack was the first to achieve a time below 1h40m, back in 2018-08-03. Both NT and Gocu have continued to be top runners in the community.

Keep in mind that during all this time there was yet one more paradigm-shift in the movement technique waiting to happen: The introduction of the dash-jump-aim-cancel.

City Hall joined the Dirge speedrunning community in July 2020. By the end of November that same year he had introduced and implemented the dash-jump-aim-cancel movement technique, which he dubbed the "City Dash" after himself. This technique requires more dexterity with the PS2 controller but it saves good time by lowering the amount of time the player spends in the air and it increases the frequency of dashes. Measurements show the average velocities to be...
Running........6 m/s
Dash-jumping...6.51 m/s
City-dashing...6.81 m/s

On straight- and downhill paths the city-dash is especially useful. Measurements show that uphill movements benefit more from the simple dash-jump, due to airtime already being limited when jumping forwards on an uphill.

The next big paradigm-shift in the game's routing then came with Gocu's WR of 1h33m23s back in 2021-01-23, as they demonstrated just how powerful it was to employ high-level Thunder- and Fire spells in a speedrun. It was with a combination of these magic strats and the city-dashing technique that City Hall would then take the WR in 2021-07-30 with a 1h33m21s.


Dirge of Cerberus remains largely a speedrunning game without glitches. There is a complete lack of out-of-bounds tricks, sequence breaks, item duplications, warp methods etc. A good speedrun performance so far boils down to refined movement technique, killchain setups, magic spamming and good RNG.
 
Last edited:
Very brief history of Dirge of Cerberus speedrunning: The Extra Missions

When you think about the story campaign of Dirge, speedruns and record times is usually not what comes to mind. The Extra Missions lend themselves easier to the "Time Attack" mindset thanks to the game saving- and showing your last record times in a menu that's visible at any time. I will confidently assume that many casual players give some missions a few extra tries just to see how far down their IGT gets lowered.

The Extra Missions have two big downsides. Firstly, you need to complete at least two playthroughs of the main game, acquiring all the necessary equipment therein, if you want to truly grind down your record times. Secondly, your starting equipment in a mission depends on what you had when you last defeated Omega Weiss, which makes it awkward and tedious to change what setup you start with.

Getting started with a speedrun of the main game is thusly far less cumbersome. You don't even need a memory card if you're aiming for Normal Mode Any%. Just boot up the game for the first time and start running. You barely even need to start out with a casual playthrough.

My go-to reference for the earliest days of elite play of the Extra Missions are fiel2c's uploads from September- and October of 2011. Many of their completion times are formidable and display the kind of forethought and planning that is so rare to see in DoC sessions. What you can tell however is that these are not "speedruns": They are displays of skill, strategy and can be a helpful tool for others, but fiel2c is not actively trying to trim down records. There is lots of time spent being idle and there are no dash-jumps in sight.


One lone runner broke the mold however. During 2012 and 2013, Brazilian player Carlos Cauê played the Extra Missions with the express intent of speedrunning. He uploaded 24 mission completion videos and they display not just great efficiency with routing and equipment...Carlos Cauê is dash-jumping and CITY-DASHING near constantly. The earliest recorded evidence of Cauê using city-dashing (the dash-jump-aim-cancel technique) for fast movement is in one of his tutorials uploaded on January 2012. This is almost nine years before the technique was re-invented and implemented by City Hall in the speedrunning community of the main game.

Carlos Cauê proclaimed himself a speedrunner and the intent shows. Many of his record times are close to that of my own. Sadly, not all of his recorded accomplishments were uploaded. A harddrive failure appears to have deleted many of his videos, leaving some record times to only be hinted of when seeing the missions menu. His last Dirge of Cerberus upload is from February 2013 and none of his knowledge ever spread to the "main" speedrunning community that effectively began one year later. I only found out about Cauê roughly two weeks ago and his videos were a shock to me as well.


***TANGENT***
Interestingly, Carlos completes each mission without a memory card which results in a save prompt at the end of each challenge. Does playing this way result in better IGT and/or RTA? According to my research in EM15, this method does not yield better records.

When you clear a mission there will be a save icon in the upper right part of the screen. For my RTA measurements, I use this first frame with the save icon as the moment when the timer ends. I always play with a memory card in the PS2 console.

Clearing EM15 without memory cards reveal that the save icon actually shows up BEFORE any save prompt window pops up. The usual time it takes for the mission to end when the final target is defeated, 5.472 seconds, remains. If you were to use the save prompt window as your end-time measurement, your RTA would always be worse by 1-4 frames. There is no benefit, based on my EM15 research, to playing the Extra Missions without a memory card.

I am happy though that Cauê's crazy move made me test this all out. Otherwise I would have never learned that this technique is the only "normal" way to hear the full "mission completed" jingle at the end.

*Please excuse the loud bullets in the first second of this video*

In the past I unlocked this full track with cheats and then uploaded it to my Audio Library section of the Multiplayer Archives. Even the official Multiplayer OST lacks a few seconds of the track compared to unlocking it in-game. At least now we know that by removing the memory card you can hear it all (unfortunately accompanied by the background noise of the mission map, of course).
***END OF TANGENT***


Apart from fiel2c and Caue/Cauê, the only other person to upload impressive completion times for a great number of Extra Missions is DevilHunterXYZ. This was in June-July of 2017 and the videos were primarily meant as an efficient demonstration- and explanation of each mission. Speedrunning was not intended and there are no dash-jumps or dash-jump-aim-cancels to be found here.

Because DevilHunterXYZ is the only person to upload videos of ALL missions, he is the only one I can compare "Total Mission Completion Time" with. If you add together the IGT of every mission DevilHunterXYZ completed, except for the three missions that don't save IGT, his total completion time ends up at 3h45m25s.

In my current savefile, my completion time just reached below four hours: 3h59m28s. Keeping track of this total time, thanks to my spreadsheet, is one of the joys with my speedrunning adventure. I look forward to seeing how far down this total time can go.


This post has all revolved around the completion of individual missions but what about a marathon run of all 46 missions in one go? Far as I know that is still unprecedented. Once I have completed individual EM speedruns, I might look into doing a marathon.
 
Last edited:
An out-of-bounds video just for fun.


When the moving cylinder has completed one cycle, a bug will become available that causes the cylinder's position to be temporarily reset while the game is paused. This glitch can easily be spotted by anyone playing the game normally, whether it be in Ch9 or in one of the missions that employ this map. You'll also notice in my EM15 video that the cylinder position (and its velocity) is reset when the mission itself starts properly.

The designers really should have let the top elevator entrance drift further away. Casual players can easily spot the entrance even when it has reached its maximum height coordinate. This destroys the illusion of being on a moving elevator.

All the out-of-bounds observations here are the same whether you visit the elevator in the main game or in one of the Extra Missions.
 
Last edited:
Speedrun of Extra Mission #16: Stygian Sewers

RTA: 00:49.616
IGT: 00:49



Sight Support: Manual​
W1: Ultima Weapon + Guard N Barrel + [Empty] + Manasoul + Fire Materia (Fire Lv1)​
W2: P Hydra Gamma + Guard N Barrel + [Empty] + Manasoul + Blizzard Materia (Blizzard Lv1)​
W3: Sonic Griffon + Guard N Barrel + Materia Floater Gamma + Manasoul + Thunder Materia (Thunder Lv3)​

The mission gives you 15 minutes to find the Gigas EX Medal held by the Heavy Armored Soldier B.

The Heavy Armored Soldier B (henceforth "HASB") has 20000 HP internally. When their HP has been depleted to below 17200, the target disappears in an orb of darkness and teleports to the storage room. To get there you need a cardkey. Once in the storage room, you must deplete the HASB's HP to below 15200. They'll drop the Gigas EX Medal and the surrounding barriers will drop, BUT WAIT! The game teases you. A new set of barriers trap the medal when you reach the platform and the HASB is "resurrected" into a Heavy Armored Soldier S (HASS). Defeating the HASS grants you the cardkey needed to finally pick up the Gigas EX Medal and end the mission.

Because of how the HASB target is programmed they effectively have 2801 + 2000 = 4801 HP from the player's perspective. You can also cause the HASB to spawn in the storage room in an already defeated state. For example: Manasoul + Thunder Lv3 (no killchain) deal 2446 damage. So with two Thunder Lv3 shots you'll deal 2446*2 = 4892 damage. The target will count as completely defeated and the Gigas EX Medal will spawn in the storage room. You'll still have to go through the trouble of triggering the HASS to appear and get the last cardkey. For a speedrun, double-defeating the HASB right at the start is not beneficial enough.

If there was a way to instantly deplete the HASB's HP to actual zero then they would drop the Gigas EX Medal on the immediate spot where they died. No storage room detour. This would end the mission very quickly but you'd have to break the 9999 damage cap in order to conceivably accomplish this. This "method" of clearing EM16 is thus not in the cards.

A debug zone called "Test 1" contains a barebones version of EM16. In that map the HASB will drop a Mini EX Medal and picking up the medal ends the test mission.


The first tunnel in EM16 will randomly spawn one of the following:
- Handgun Bullets x9​
- Machine Gun Bullets x30​
- Rifle Bullets x4​
- Potion x1​
- Ether x1​
- Phoenix Down x1​

I have not confirmed the probability rates of each item drop but my impression is that they are evenly distributed. Potion and Ether hold the potential of being useful in a speedrun, the Potion if you take heavy damage from bad RNG and the Ether if you want to increase magic usage.


My initial strategy against the HASB in the tunnel was to shoot their head with bullets and hope for a good streak that would end up causing at least 2801 damage. This resulted in too much hesitation and waiting. Thus I switched to Fire Lv1 as a means of activating the drum can in the HASB's vicinity. The drum cans in the Stygian Sewers are powerful, with a base damage of 3000, and have a wide range. What's mainly challenging about my strategy is landing the correct angle with the magic shot and to have the good positioning and RNG so to not get hit by the HASB's bazooka projectile.


EM16 marks the first time we get to use Blizzard magic in a speedrun. The Black Widow in this mission is invulnerable to Thunder magic when faced from the front. Why? I do not know. Normally the Black Widow has to be upside-down in order to be invulnerable to Thunder. Maybe the weird angle of the Black Widow on the sewer grounds causes this behavior to act on the front.

Anyway, what this means is that our usual choice of overpowered magic, Thunder, will not do. The Black Widow has 1000 HP. Manasoul + Fire Lv3 is too weak as it only deals 979 damage. But Manasoul + Blizzard Lv1 deals 1102 damage! The sheer power and the low MP consumption suits us perfectly. It makes me happy that Blizzard became useful in Stygian Sewers.


In the storage room I launch a Thunder Lv3 strike in midair so that I avoid triggering the drum can and instead only deal damage on the HASB and the lone DG soldier on the boxes. The "magic trick" against the HASS ended up way more precise and dangerous. The sword-wielding Heavy Armored Soldier has 4000 HP, meaning none of my magic is powerful enough to one-shot the enemy and get a cardkey. That's why I kept the drum can intact earlier. By attacking downwards with Thunder magic at just the right angle and position I can simultaneously trigger the drum can and deal magic damage on the HASS. The enemy is thusly defeated in "one shot" and this benefits the speedrun. It is easy to get the angle wrong and to end up caught in the explosion.


Stygian Sewers is something of a cacophony to watch. The action happens fast and the initial rain of bullets from the enemy is an auditory assault. The moment is quickly over but it's still unpleasant. Visually there is also a lot of lag due to all the action happening and when playing the mission you have to learn where the game is likely to eat up your controller inputs. The input-eating in particular made me curse and yell. That is unfortunately a big part of Dirge of Cerberus speedrunning though: Learning where you need to hold down the buttons for a fraction of a second longer in order to guarantee that Vincent will dash, use magic etc.

My casual time record for EM16 was 06:49. My speedrun, after much suffering, became 00:49. Oh what a beautiful, round shave of six minutes. I will be immensely impressed by anyone who is able to get a 00:48 in this mission.
 
I have a habit of going into a mission and thinking "Okay, surely THIS ONE will have nothing interesting at all about it" even though I'm almost always proven wrong. Extra Mission #17, Messenger from Hell, proved me massively wrong by teaching me something new about the game.

Drum cans and cargo crates, when taking damage, can ignore the damage cap of 9999.

The game is strict about applying the normal damage cap when you are using melee and Thunder magic. But if you are using bullets (handgun/rifle/machine-gun) or Blizzard- or Fire magic, the game will allow damage above 9999 to be dealt on drum cans and cargo crates.

To the player this will be invisible. No damage digits are displayed when you attack these destructible objects and on top of that none of these objects typically have HP values even close to 10000.

The way that the crates are usually programmed is so that melee strikes deal a set amount of damage. The crate's HP might be 60 and a melee strike from Vincent will always deal 20 HP of damage no matter his level, equipment and killchain. This ensures that if you're gonna use melee then you'll always need to land three consecutive strikes on the box before it is destroyed (assuming no other damage was dealt on the box). The easy way to go around this is to use bullets and/or magic since damage from these are still calculated based on the player's actual power.

EM17 in-game mission description:
"The dead have returned from hell and won't stop until they take you back with them. Weapons have no effect against these undead foes, leaving you with only one option--escape."​

Ackshually, weapons do have an effect. Bullets are weak here, yes, but the indigo-colored enemies each have only 4444 HP (Coincidence? I think not!) and can be defeated somewhat easily with powerful magic. The soldiers and beasts spawn indefinitely though and they are not the targets to focus on. Your goal is to escape the linear corridor within three minutes, clearing the cargo crates in your path before finally reaching the Goal Point beyond the exit.

Most of the cargo crates on your path are programmed with 800 HP and some of them have 1000 HP. Melee is extra useless here: Ten strikes are needed to destroy a single box. EM17 also sets itself apart by nerfing magic on these boxes in a similar way that melee is nerfed. No matter the inherent power of your magic attack, it will always deal a predetermined set amount of damage on the boxes. To destroy a box you need to land three magical attacks.


Once we approach the exit and freedom appears within our grasp, eight crates spawn in front of the gateway and these must be destroyed before we can reach the Goal Point. This is where the underhood of EM17 becomes truly interesting. Neither bullets, melee nor magic is nerfed on these boxes. They take damage same as any common, moving target would...except that now we can skip the normal damage cap of 9999.

EM17_cargo-crate-boss.png

The eight boxes have 20000 HP respectively. After they have spawned, the "Boss HP" bar will show up in the upper right corner and the game will now internally recognize that this "Cargo Crate Boss" has 5000 HP.

Once the boss HP has been depleted, the crates will explode one by one and open the path to the Goal Point. But here's the kicker: You don't need to wait for the HP bar to show up or even for the game to retrieve the "intended" HP value of 5000. If you can deal 20000 HP worth of damage to even a single one of the crates, the boss will count as defeated and initiate the exploding death animation. Internally the boss HP of 5000 will have generated and remain untouched but you'll still have destroyed the boxes and opened the path to the outside.

A potential speedrun strategy here is to accumulate a 14-chain on your way up through the corridor, then launch a single Blizzard Lv3 attack on the first box to spawn. This will instantly deal more than 20000 HP worth of damage in a single shot, thanks to the normal damage cap being ignored, thus activating the destruction of all eight boxes. The boxes all have normal magic defense.

magic-damage-chain-14.png

Measurements still have to be taken to check whether a rapid-fire approach with Ultima Weapon or the single-shot Blizzard Lv3 strategy is the fastest. Whichever method ends up used I love the fact that EM17 provides the opportunity to legitimately deal more than 9999 damage with a single projectile. Going above the damage cap just means so much more when your target actually has a total HP above said cap.
 
Last edited:
Speedrun of Extra Mission #17: Messenger from Hell

RTA: 00:17.684
IGT: 00:17



Sight Support: Manual​
W1: Ultima Weapon + Gale S Barrel + [Empty] + Power Booster Gamma + S Adjuster Gamma​

The Blizzard Lv3 strategy had to be scrapped and so I go in guns blazing with only the W1 setup. If you arrive at the top with a 14-chain it *is* actually 0.3 seconds faster to use a Blizzard Lv3 on the cargo crate boss than to use rapid-fire with normal ammo. Numerous obstalces prevent this from being the prime strategy however.

Following the shortest path, it is entirely random if one of the boxes you destroy ends up dropping an Ether for you to gain the MP needed for the spell. Sometimes you'll pick up a Potion or a Phoenix Down before picking up an Ether, meaning you'll have to press left on the d-pad (item scrolling) before then finding the Ether and consuming it with the right arrow on the d-pad. That's a lot of extra mental power, and thus timeloss, for using just one item. Add to that I never know when the Ether will be dropped.

Often I'll reach the top with a chain lower than 14. This renders Blizzard Lv3 too weak to one-shot the cargo crate boss. Just like spending brain power on looking for- and using an Ether will result in timeloss, so will switching to a different weapon likely result in loss of time. I deem then that the magic strategy is more reasonable for a TAS.

The boxes on your way up all have super-tall collision. This means you can neither jump on top of them nor jump over them. You can stand on top of the drum cans but there's no benefit from this. Thankfully the crates at the exit have normal collision so you can jump on the last one before it disintegrates, saving a little bit of time in moving towards the Goal Point.

If you had superspeed then it would be possible to exit the gateway before the cargo crate boss fully forms and blocks your path, but we're talking a required speed of around 30 m/s which is far greater than what can be achieved without cheats. I believe a human may be able to achieve a time below 17 seconds but for a record below 16 seconds it'd take some game-breaking strategy.
 
Last edited:
If you complete Extra Mission #18 normally and devote yourself to getting a respectable speedrun record you'll get somewhere between 01:10 and 01:20.

With this shortcut however you can cut about 20 seconds from the mission, making a record below one minute possible.


The most important part of this trick is clipping in between the platform and the ladder. This is what activates the platform collision for Vincent and what makes it possible for Vincent to grab the cardkey. If the platform-player collision is not activated you won't be able to grab the cardkey even if it looks like Vincent is physically touching the key when the player does some impressive double-jump-melee-combo maneuver.

It's helpful to have a decent amount of forward momentum when jumping off from the box and the fast-firing machine gun is necessary to make you clip in between the platform and the ladder.

By far this is one of the most exhilarating gameplay discoveries I've ever made in Dirge of Cerberus. I had to show it off immediately. Could not wait until my actual completed speedrun.

*Update: I have now confirmed that this does work on console too but holy moly is it difficult! Took me five minutes to grab the cardkey the first time. With practice it does get easier. Slightly wiggling the right analog stick as you are firing the machine gun seems to help you clip in.

**Update #2, 2021-09-14: Further tests reveal that you actually can get the cardkey by doing a more straight jump towards it. You don't need to jump in between the platform and the ladder. My initial theory about collision-activation seems to be erroneous. The trick is primarily to optimize your height and angle. In order to get the three-strike melee combo in midair you just need to raise the reticle after the double-jump. Otherwise you'll get the six-strike melee combo that launches Vincent downwards by a lot.
 
Last edited:
Speedrun of Extra Mission #18: Badlands

RTA: 00:57.524
IGT: 00:57



Sight Support: Automatic​
W1: P Hydra Gamma + Nova L Barrel + Sniper Scope + Manasoul + Cerberus Relief Gamma​
W2: Sonic Griffon + Gale S Barrel + Materia Floater Gamma + Manasoul + Thunder Materia (Thunder Lv3)​

While playing this mission I gradually ended up lowering the number of snipers I destroyed on the way, until I shaved it down to only one sniper defeated in the mission's first half. Once the first cardkey is used up then most of the enemies in the first section will instantly die, including the snipers. The odds of going mostly unscathed while running through this no-man's-desert turned out to be pretty good. Enemies don't fire very often and your dash can deflect bullets if the timing is right.

How grateful I am to the cargo crate that makes the cardkey shortcut possible. The box contains a Potion but its blessing turned out so much greater. The stars really had to align for this awesome trick to be available. To see the long way round you're supposed to take, see 00:54 - 02:16 in Caue's run.

While practicing this trick-jump on console I moved away from the analog-stick-wiggle strategy and found the following to be more important:
1) Good angle
2) Great momentum into the wall, not letting go of the left analog stick
3) Fast-paced melee combo

The slightest delay in the melee combo means you'll fall and lose the height needed to pick up the cardkey. My number of successful cardkey-grabs increased when I made sure to mash the melee button with extra deliberation. The scenario I got in this run is the most optimal: Grabbing the cardkey but not actually landing on the platform. Often when landing on the platform it will take me time to react and resume Vincent's movement in the correct direction. In addition, it looks and sounds cool to activate the mine but fall fast enough to avoid its explosion. You have enough HP to survive one explosion but it's slower and riskier to get caught in the blast.

I have not done extensive testing but I believe the cardkey-grab trick could be done with the rifle and/or handgun as well, it's just that the extended firing of the machine gun makes it easier to clip in between the platform and the ladder.


The speedrun of EM18 also marks the moment when I've finally surpassed DevilHunterXYZ in total mission completion time.
DevilHunterXYZ's total IGT: 3 hours 45 minutes 25 seconds
My current total IGT: 3 hours 39 minutes 36 seconds

If I knew Caue's total completion time I'm sure it would be somewhere below 3 hours. We'll get there eventually.
 
Last edited:
Galian Beast has a flame bullet technique that I've been unaware of for all this time: The grounded triple-fireball technique.

galian-beast_triple-fireball.png

The ground-type fireball is launched by pressing R1, whether Galian Beast is on the ground or in midair. It's a single, homing flame bullet from the beast's right arm. Not very strong, especially so in the original Japanese version, and made extra useless by the midair slashing+fireball mayhem that the player can use in the post-original releases.

The beast has one more trick up their sleeve IF you are playing any of the post-original editions though. Perform two melee strikes and then follow up with a press of R1: Three flame bullets will be launched simultaneously! The animation is unique and shows the fireballs being launched from Galian Beast's LEFT ARM. After years of only seeing ground-type flame bullets launched from the right arm it feels uncanny to see the left arm performing the action in this attack.

triple-fireball-01.png


triple-fireball-02.png


triple-fireball-03.png

The three homing projectiles are lined up pretty close so often they'll hit the same target. If the target dies from the first or second fireball, the remaining ones will go on to seek out new targets.

You'll only get this attack by following up with R1 after the second strike in a melee combo. It won't happen from the first strike or even the third one. Indeed there is no way to launch flame bullets in conjunction with the ground-pound finisher either. The special window for the triple-bullet move definitely adds to why I did not observe it before now. Heck, when discovering it today my mind was only on the unique left-arm animation at first. It was only after using the attack half a dozen times that I noticed the multiple projectiles.

The midair-melee flame bullet mashing move is still the most overpowered but you can see the intent here by the designers in trying to level up the ground-fireball moveset as well. Even if I accidentally triggered the triple-flame attack at some earlier point while playing Dirge of Cerberus it is not something that I ever put into my conscious memory. :monster:
 
Last edited:
Speedrun of Extra Mission #19: Chamber of Ghosts

RTA: 00:03.503
IGT: 00:04



In my opinion the funniest Dirge of Cerberus video that I have ever uploaded.

Sight Support: Manual​
W1: Ultima Weapon + Gale S Barrel + [Empty] + Power Booster Gamma + Recoil Limiter​

The undead/ghost target spawns randomly in about a dozen locations. This is entirely RNG. The target will spend about 2 seconds in their first location and afterwards it's largely random just how much- and fast they teleport around. Landing shots *seem* to extend the target's time spent in a given coordinate but I don't know how the ghost is actually programmed.

The ideal pattern is one where the targets spawns in front of you and then decides to teleport back to exactly where they just were. I managed to get that in the second-best possible spawning point.

spawn-point-close.png

But the best one could well be this:
spawn-point-near-straight-ahead.png

Being almost straight ahead, less time would be spent on positioning the reticle if you start out assuming the enemy will appear here.

The pattern I got was so insanely lucky though that I feel blessed enough to leave EM19 behind. The bad luck with the IGT barely even bothers me. With an external timer I could have prevented that extra half second from being added to the in-game time, yes. But even my obsession with IGT has its limits.

The target has 20000 HP and their head takes 4.0 extra damage.

Ultima Weapon + Gale S Barrel + Power Booster Gamma​
(4.0) *Critical* Head of Ghost/Undead​
1 bullet: 814​
2 bullets: 1628​
3 bullets: 2442​

There is some short-range scatter because I don't have the S Adjuster Gamma accessory equipped. Recoil Limiter still feels like the right decision because the recoil is otherwise very large and I would have had to constantly adjust the reticle to strike the target's head.

With this setup we need at least 9 shots to defeat the ghost...and amazingly that is how many shots it took for me! I fired one extra shot *after* the enemy had died but even so I am amazed at how well this one went. Damage-wise this is how the mission went down.

2442 + 1628 + 2442 + 2442 + 2442 + 2442 + 2442 + 2442 + 1628 = 20350​

7 perfect headshots and two headshots where two bullets landed respectively. The third bullet of the 9th shot may have landed as well, but because the target's HP was already depleted then no more damage digits would be displayed.

One of my weaknesses however is in the button-mashing department. This fact has presented itself to me on numerous occasions in the recent missions. I'm just not young anymore. XD

The equipment provides a maximum firing rate of 4,2857 shots/second if you achieve perfect trigger-efficiency (equivalent to having an auto-fire feature on). My average firing rate ended up at 3,996 shots/second. That's a lot of time that could be saved if I had better rapidfire technique.

I do believe that an RTA record below 3 seconds is possible and even an IGT of 00:02 might be achieved one day if a player can get both the best pattern and the best execution (which may or may not include optimizing the in-game timer). EM19 is definitely a contender for being the shortest Extra Mission of them all when cleared in the most ideal conditions. You could argue that EM02 is shorter since it has to be "finished" in less than a second and then you spend three mostly dead seconds waiting for the explosive death animation of the Black Widows to finish before the game acknowledges that you've won.

Whenever an Extra Mission does NOT have a needless delay like the one in EM02, I am pleased. EM19 is short and random but I prefer it over EM02. There's also no lag and input-eating in EM19 since the game is not being overwhelmed by map size, objects and animations.
 
Last edited:
I have done it. I have discovered the first barrier skip in Dirge of Cerberus.


The discovery truly is one of those "how did I not see this before" type of moments. It was only when I made note of Vincent's Z-coordinates, while standing on top of the barrier with the use of cheats, that I realized the barrier's collision wasn't overwhelmingly high. The preserved cardkey can be used for any subsequent point in Ch10. Barriers in Ch10 only consume one cardkey per barrier. Some chapters have the barriers set your cardkey count to 0 but fortunately we don't have to worry about that here. Cardkeys can't be preserved over to Ch11.

The best timesave this offers for speedrunners is to completely skip the gun turret section in Ch10-3. Runners will often spend 80-90 seconds on the gun turret area. Backtracking to climb the ladder in Ch10-1 and jump over the barrier takes about 13 seconds. In other words you are pretty much guaranteed a timesave of over one minute so long as you make the jump on the first try.

Compared to the cardkey shortcut in Extra Mission #18, Badlands, this extended jump arc in Chapter 10 is way easier to achieve! A well-placed midair shot at the end of Vincent's extended jump arc can help with airtime but it's not necessary.

What a glorious day. The dream of achieving a barrier skip and preserving a cardkey has been fulfilled.
 
I kinda-sorta-maybe did it again. This trick jump is the most difficult that I've ever come across however. After 40 minutes of intense attempts on console I am yet to replicate this on original hardware. Skipping this barrier, exclusive to the Ex Hard difficulty mode, might only be possible on emulator.


Knowing just how close Vincent was to reaching above the barrier I tried the jump for three hours on emulator without success. Only when I equipped Sniper Scope and entered sniper-mode when touching the top edge of the barrier did the barrier skip work. What is it about entering sniper zoom mode that makes the jump possible? I do not know. I only have speculations.

If a functional setup was found on console for this barrier skip then Ex Hard players would be able to preserve one cardkey. By keeping the cardkey until Ch6 they could then use it to skip the lengthy choppers segment, shaving 4 1/2 minutes off the run. There's a reason that skipping the Ch6-2 choppers is one of the holy grails of Dirge of Cerberus speedrunning.
 
Dirge of Cerberus has now been utterly broken by yours truly. September 18, 2021, is now FF7 history.


Raising the reticle gives Vincent a jolt upwards. This is the same with raising the sniper scope zoom, at least when exiting with the melee button. The normal first- and third person reticle is limited to one raising per midair flight. However, the sniper scope does not have that limit.

Set "zoom" to an easy-to-mash button. Jump, then press the zoom trigger and then use the melee button to exit. Repeat this process as fast as you can. Vincent's fall will slow down! If your timing is perfect, you'll slow Vincent down to a halt.

Vincent can still move around in midair. This means you'll now have an immensely extended hover so long as you can keep the mashing going. This is what I call the "Sniper Hover". It has multiple use cases and blows the game open. Confirmed on both console and emulator.

The cardkey can be preserved and its best use is for Ch6 where it can be used to skip the choppers.
 
You can gain height indefinitely. By doing the zoom-unzoom technique when Vincent is in his first jump, he'll gradually climb upwards. If you do the technique after his double-jump you can only maintain your current height or fall down.



*UPDATE: Sniper Scope isn't even needed! You can do the Hover Climb anyway. Thanks to NT1_Evolution for noticing. It is the "R3"-zoom that allows infinite re-raisings, not the sniper scope itself.
 
Last edited:
Top Bottom