This Typing of the Dead: Overkill submission was recorded as a series of individual levels after getting feedback from SDA verifiers and staff following my original single segment rejection.

I implemented the verifier suggestions and managed to put together a MUCH nicer looking set of runs. All 9 levels are faster than their counterpart from the original single segment submission, most of them quite significantly so.

Adding it up, this total time of this submission is 3 minutes and 35 seconds faster than its predecessor - far and away more than I ever originally thought could be saved even by grinding out the individual levels.

Please see https://forum.speeddemosarchive.com/post/the_typing_of_the_dead_overkill__may_15_2014.html for the history of the original submission.

--------------------------------
General comments and information about speedrunning ToTD

About speed vs. accuracy:

The difference between my very best high score runs and my very fastest speedruns (PBs), is quite significant. While the comparisons below put my single segment speedrun against the much improved IL times, comparing them to my old high score runs shows an even greater difference. The play style ends up being entirely different.

High scores concern themselves with getting a full-combo on the stage without any typos. Obviously doing this requires slowing down WPM significantly. In a speedrun, the primary concern is clearing out the mutants from the screen *as quickly as humanly possible* above all things. This means combo will break quite often... but the gain is that the average WPM can be MUCH higher.

I found on average my speedruns of any given level ended up being 45 to 60 seconds faster than the standard playthroughs which had gotten me to the top of the leaderboard. There are a few reasons for this, and here's what you're going to see in the run:

Damage:

You're going to see me take a lot of damage in this run. Getting hit drops your combo... but in a speed run this is irrelevant, and in fact hitpoints are a resource. You will see a number of situations where a mutant throws a projectile (a single character). Waiting for the throw, or backspacing out to hit the single letter and starting again both cost precious time. Instead, my strat was to dispatch the mutants as quickly as possible even if it cost me lifebar.

Similarly, there are a number of sections where a far-away mutant is available to attack first... and then a closer mutant emerges and deals damage if you don't backspace out to kill it. The fastest thing to do in a speedrun is to soak this damage. There are even a couple of places in the latter sections of the game where taking damage intentionally from exploders is the fastest way to clear the screen.

"Slow Time" Powerup:

There are a number of sections where the game throws at you a dense horde of mutants and expects you to use the red "slow time" powerup to kill them without taking damage. This is simply not an option in a speedrun, and is never used.

About RNG:

While a small number of mutants have static words / phrases assigned to them, the difficulty of the phrases can be highly variable. While I found that this tends to average out in a single segment run of the whole game, redoing this as ILs allowed me to play the odds. My experience was that performing very well with "bad" (long) RNG phrases is unfortunately still slower than performing just okay with "very good" (short) RNG phrases. For these ILs I didn't stop playing a level until I had managed to get a strong performance AND better than average RNG. (Developing multiple midpoint splits for each individual level really helped me button this down.)

Another place where you REALLY want concentratedly good RNG (and of course still good play) is on the bosses where you get a finite amount of *real time* to bang out as much damage (as many phrases) as you can before they go through another cycle, usually in the form of "grey phrases" that keep appearing while the boss is in hitstun. Doing this as ILs allowed me to only turn in levels where I got the optimum performance on the boss runs, and only after I was already satisfied with the time of the main stage.

-------------------
Individual Level Information:

(note, all given times calculated include the interum loading time between stage and boss - this should be removed for comparisons between mine and other runs on disparate PCs, but is consistent across all of my runs as they were captured on the same machine)

Papa's Palace of Pain:

8:57 - Great stage overall, not a lot of variance on the boss unless you flub something. My single segment, which I thought was already very fast, was a 9:07 here.

Naked Terror:

9:18 - My single segment was a 9:22 and had a ludicrously good first split that I was always behind on by a few seconds. Managed to shave all my time on the latter section. Again this boss fight is pretty uniform if you don't stumble on phrases.

Ballistic Trauma:

11:53 - Just minor improvements here, the single segment submission had an 11:59. Boss is uninteresting. Don't choke. (Seven Days.)

Carny:

14:28 - Huge 30 second improvement here! The old single segment was 14:58. I was never happy with that boss fight because I didn't get enough damage in for the minimum amount of cycles. Getting the "two dives" finish on Nigel & Sebastian by itself saved 10 seconds from the old run. This means the main stage was separately 20 seconds faster than the old run. Part of that came from smarter survivor strats, and the rest was just solid typesmanship.

Creeping Flesh:

9:21 - This level is a nightmare to get a perfect score on, but luckily we don't care about taking damage in this run. The single segment time was 9:39 and the 18 second difference comes primarily for an optimized boss fight with Meat Katie. The old run had 5 chops (3 chops, charge, 2 chops) - this run does it in 4. It's very difficult to get three "grey" phrases in after eat chop, but necessary to reduce the number of boss cycles.

Scream Train:

12:01 - The old time was 12:31, and all of the time save came from a faster main stage. By this point in the single segment run I was suffering from fatigue and soreness in my tendons... and obviously doing it as an IL avoids this. The other advantage to IL is that I was able to play the stage a fair number of times until I had reasonable RNG. Barring major mistakes (dropped phrases) the boss fight takes more or less the same amount of time every time.

The Fetid Waters:

14:18 - You're looking at damn near a full minute time save from the old submission! First off, I knocked 16 seconds off of the main stage time. The optimum survivor strats here are really nebulous, it really seems to depend on the difficulty of the phrases whether or not it's best to let them eat it or not. I did, however, find an erroneous trigger / intentional damage strat where you get the camera to move on even though mutants are still alive. Mostly though, the old Single Segment time of 15:16 was due to a 3-lobs boss fight with Lobber. It is absolutely imperative that you not let lobber go through three lob-puke cycles. It can range from somewhat difficult to functionally impossible to get two grey phrases in during his hitstun, depending on the RNG. You don't have to get two phrases in *every* time, but if you fall short too often he'll have just enough life left to go through the lengthy lob-puke cycle a third time. This lobber was done in two lobs and the boss fight alone saved an additional 42 seconds.

Jailhouse Judgment:

13:55 - The old single segment time of 14:27 had a "perfect" Brutus (no dropped damage opportunities), but after a lot of trial and error I discovered that it seems faster to take one hit on this boss. This gets to the "There he goes" pattern which allows more damage to be put on him prior to him summoning mutants and this means he only summons once. Although this boss fight looks sloppier due to the dropped phrase, it is 4 seconds faster. The stage by itself is still 28 seconds faster, partly due to faster typing - and party due to letting all of the survivors die and avoiding the particularly lengthy cutscenes you earn from saving them.

Overkill:

12:47 - Old single segment was 13:34. The time save of 47 seconds was almost entirely due to the stage, only about 4 seconds came from improving the boss. Letting the survivors die was helpful, but I found an intentional damage strat and a few other specific mutant order sections that boosted this run further.

----------------------
Administrative:

I intend to record audio commentary for this if it is accepted, but I want to hold off until after verification.

This Typing of the Dead: Overkill submission was recorded as a series of individual levels after getting feedback from SDA verifiers and staff following my original single segment rejection.

I implemented the verifier suggestions and managed to put together a MUCH nicer looking set of runs. All 9 levels are faster than their counterpart from the original single segment submission, most of them quite significantly so.

Adding it up, this total time of this submission is 3 minutes and 55 seconds faster than its predecessor - far and away more than I ever originally thought could be saved even by grinding out the individual levels.

Please see https://forum.speeddemosarchive.com/post/the_typing_of_the_dead_overkill__may_15_2014.html for the history of the original submission.

--------------------------------
General comments and information about speedrunning TotD:O

About speed vs. accuracy:

The difference between my very best high score runs and my very fastest speedruns (PBs), is quite significant. While the comparisons below put my single segment speedrun against the much improved IL times, comparing them to my old high score runs shows an even greater difference. The play style ends up being entirely different.

High scores concern themselves with getting a full-combo on the stage without any typos. Obviously doing this requires slowing down WPM significantly. In a speedrun, the primary concern is clearing out the mutants from the screen *as quickly as humanly possible* above all things. This means combo will break quite often... but the gain is that the average WPM can be MUCH higher.

I found on average my speedruns of any given level ended up being 45 to 60 seconds faster than the standard playthroughs which had gotten me to the top of the leaderboard. There are a few reasons for this, and here's what you're going to see in the run:

Damage:

You're going to see me take a lot of damage in this run. Getting hit drops your combo... but in a speed run this is irrelevant, and in fact hitpoints are a resource. You will see a number of situations where a mutant throws a projectile (a single character). Waiting for the throw, or backspacing out to hit the single letter and starting again both cost precious time. Instead, my strat was to dispatch the mutants as quickly as possible even if it cost me lifebar.

Similarly, there are a number of sections where a far-away mutant is available to attack first... and then a closer mutant emerges and deals damage if you don't backspace out to kill it. The fastest thing to do in a speedrun is to soak this damage. There are even a couple of places in the latter sections of the game where taking damage intentionally from exploders is the fastest way to clear the screen.

"Slow Time" Powerup:

There are a number of sections where the game throws at you a dense horde of mutants and expects you to use the red "slow time" powerup to kill them without taking damage. This is simply not an option in a speedrun, and is never used.

About RNG:

While a small number of mutants have static words / phrases assigned to them, the difficulty of the phrases can be highly variable. While I found that this tends to average out in a single segment run of the whole game, redoing this as ILs allowed me to play the odds. My experience was that performing very well with "bad" (long) RNG phrases is unfortunately still slower than performing just okay with "very good" (short) RNG phrases. For these ILs I didn't stop playing a level until I had managed to get a strong performance AND better than average RNG. (Developing multiple midpoint splits for each individual level really helped me button this down.)

Another place where you REALLY want concentratedly good RNG (and of course still good play) is on the bosses where you get a finite amount of *real time* to bang out as much damage (as many phrases) as you can before they go through another cycle, usually in the form of "grey phrases" that keep appearing while the boss is in hitstun. Doing this as ILs allowed me to only turn in levels where I got the optimum performance on the boss runs, and only after I was already satisfied with the time of the main stage.

-------------------
Individual Level Information:

(note, all given times calculated include the interum loading time between stage and boss - this should be removed for comparisons between mine and other runs on disparate PCs, but is consistent across all of my runs as they were captured on the same machine)

Great stage overall, not a lot of variance on the boss unless you flub something. My single segment, which I thought was already very fast, was a 9:07 here.

My single segment was a 9:22 and had a ludicrously good first split that I was always behind on by a few seconds. Managed to shave all my time on the latter section. Again this boss fight is pretty uniform if you don't stumble on phrases.

Just minor improvements here, the single segment submission had an 11:59. Boss is uninteresting. Don't choke. (Seven Days.)

Huge 30 second improvement here! The old single segment was 14:58. I was never happy with that boss fight because I didn't get enough damage in for the minimum amount of cycles. Getting the "two dives" finish on Nigel & Sebastian by itself saved 10 seconds from the old run. This means the main stage was separately 20 seconds faster than the old run. Part of that came from smarter survivor strats, and the rest was just solid typesmanship.

This level is a nightmare to get a perfect score on, but luckily we don't care about taking damage in this run. The single segment time was 9:39 and the 18 second difference comes primarily for an optimized boss fight with Meat Katie. The old run had 5 chops (3 chops, charge, 2 chops) - this run does it in 4. It's very difficult to get three "grey" phrases in after each chop, but necessary to reduce the number of boss cycles.

The old time was 12:31, and all of the time save came from a faster main stage. By this point in the single segment run I was suffering from fatigue and soreness in my tendons... and obviously doing it as an IL avoids this. The other advantage to IL is that I was able to play the stage a fair number of times until I had reasonable RNG. Barring major mistakes (dropped phrases) the boss fight takes more or less the same amount of time every time.

You're looking at damn near a full minute time save from the old submission! First off, I knocked 16 seconds off of the main stage time. The optimum survivor strats here are really nebulous, it really seems to depend on the difficulty of the phrases whether or not it's best to let them eat it or not. I did, however, find an erroneous trigger / intentional damage strat where you get the camera to move on even though mutants are still alive. Mostly though, the old Single Segment time of 15:16 was due to a 3-lobs boss fight with Lobber. It is absolutely imperative that you not let lobber go through three lob-puke cycles. It can range from somewhat difficult to functionally impossible to get two grey phrases in during his hitstun, depending on the RNG. You don't have to get two phrases in *every* time, but if you fall short too often he'll have just enough life left to go through the lengthy lob-puke cycle a third time. This lobber was done in two lobs and the boss fight alone saved an additional 42 seconds.

The old single segment time of 14:27 had a flawless Brutus (no dropped damage opportunities), but after a lot of trial and error I discovered that it seems faster to take one hit on this boss. This gets to the "There he goes" pattern which allows more damage to be put on him prior to him summoning mutants and this means he only summons once. Although this boss fight looks sloppier due to the dropped phrase, it is 4 seconds faster than my "perfect" run from the single segment. The stage by itself is also 28 seconds faster, partly due to faster typing - and party due to letting all of the survivors die and avoiding the particularly lengthy cutscenes you earn from saving them.

Old single segment was 13:34. The time save of 47 seconds was almost entirely due to the stage, only about 4 seconds came from improving the boss. Letting the survivors die was helpful, but I found an intentional damage strat and a few other specific mutant order sections that boosted this run further.