Submission #6083: Challenger & Morrison's NES Castlevania in 11:15.10

Nintendo Entertainment System
baseline
BizHawk 2.3.0
40573
60.0988138974405
91174
Unknown
Castlevania (U) (PRG1)
Submitted by Morrison on 8/30/2018 2:40:11 PM
Submission Comments
"GOOD EEEEVENING!" -- Castlevania instruction manual
Here's a new TAS for NES Castlevania
  • Aims for Fastest Time
  • Manipulates Luck
  • Exploits Programming Errors
  • No death
  • Genre: Platform
  • Emulator: BizHawk 2.2 / BizHawk 2.3
  • Syncs perfectly on both USA PRG1 and KC version.
This movie is 236 frames faster than the previous movie, thanks to several new skips, different strategy for 2 bosses, and a better turnaround trick. This version also features a higher score: 253,110. There were a few places in the run where it was possible to get a few more points, but sometimes it causes lag or some other issue.
TL;DR table:
LevelStageFrames
110
120
132
240
25-1
2641
373
388
3910
4100
4110
4120
51391
514-1
515-6
6160
617112
618-24
Challenger: A month ago, Morrison (a.k.a Joshua) contacted me about some new testing stuff for this game. I wrote on the submission of the published 2017 run, 3 possible improvements to this game, and he showed me his first progress: Stage 13 with a stair skip that I've failed to test well before! Yes, that stair skip which scrimpeh tested years ago. On the next day, I started with the skip, improving to perfection, and also finding a new way to pick meat. So, I decided to work on this game again because I should have worked/figured more better before (this game is one of best runs of this site too - and really deserves an improvement). We TASed using my published run, and I switched to TAStudio since I got better TASing experience and for more improvements - mainly because of my recent published run - Cadash! 4-5 months of arduous work, coordination, management, stuck on several parts (especially the last boss) and revisions for a long time, pushed me to the limits - so, I'm an improved Challenger! The rerecord count is also resulted from the experience :D
We used PMs (Private Message) from the forum to keep contact while finding/showing new improvements (a bunch of PMs :p).
Joshua: After a long break from tool assisted speedrunning, I am happy to return with a new collaboration to update the NES Castlevania TAS. I was hoping to find some way to improve the Castlevania TAS from 2017, but I was not sure of any new strategies. I was watching Castlevania speedrunners on Twitch and they amazed me. I saw frame perfect strategies from the CV TASes being attempted with a good amount of success. I saw jay_cee's 11:27 world record (from control of Simon to orb grab) which includes a meatless two cycle on Dracula (uses 1 crit). The speed runners inspired me to try some different strategies. One of the strategies was the "faster bone" boost on Stage 13. This was introduced to me by speedrunners Ouijawii and Mechanical Koopa (little did I know it's also in Grincevent's pacifist run which I also studied later). I was able to use another bone boost over the highest wall in that room to save 32 frames (two 16 frame rules at the door)--and a new Castlevania project was born.
Stage by stage notes:
--1--
Stage 1: Should be the same as the previous movie. Checked to make sure that the castle is entered as fast as possible and that the best frame rule is gotten at the stage transition.
Stage 2: Does not take big heart, because...
Stage 3: ...it was not necessary to manipulate the zombies with the stopwatch and axe for a holy water drop. Instead we keep the clock for optimizing stage 6.
Two frames faster at the timer-to-score countdown (only used watch once, so less time to countdown at the end).
--2--
Stage 4: No holy water for multi-shot drop manipulation in this section (it will be missed), but the watch is not on the enemy drop table in level 2.
Challenger: One frame gained on the first room by jumping a frame earlier, but soon lost due to extra lag during the stair transiton.
Stage 5: Same, but the item counter is cycled one time (medusa head by the door drops the money bag).
Challenger: another extra lag this time.
Stage 6: Challenger knew that this crusher section could be done faster with the stopwatch, but we needed to get the boomerang immediately afterwards along with a multi-shot to not lose the time gains at the boss. The stopwatch had to be activated early to avoid bone pillar & fireball lag. A few frames are needed to avoid the last crusher with this tactic, but it's much faster than sitting through a bunch of lag frames. The bat was manipulated to drop the boomerang. Fortunately there are 8 candles and two medusa heads to hit with a boomerang to manipulate a multi-shot drop just before the boss.
Challenger: This stopwatch strategy for these crushers is also present on the very first run of this game (14 years), but there's no boomerang.
These crushers, seems for me that works through the global timer - every 16 frames - like the doors. Pausing on right frames can delay the function for +16 frames. A watch + pause would be enough to bypass the third crusher without the need to wait (if used on a certain frame and timing), but not enough to avoid lags from bone pillar. So, +1 pause, fix this but gets the same result as the other strategy.
With II multi-shot instead of III, only 2 frames were lost - same strategy as the previous run.
Since we used watch twice, more seconds = some frames lost.
--3--
Stage 7: Optimizations on movement when manipulating the hunchbacks saved a few frames. The down button can be used to turn around without losing X position. Grabbed a III multi-shot.
Stage 8: Same drop manipulation from the last run to set up a potion drop. Uses a brief pause to manipulate the skeleton to throw an extra bone for a boost. Keeps the bone pillar pattern for the 7ooo scores on the last two fireballs.
Stage 9: Same special drop and score manipulations up to the mummies. Uses a better mummy pattern with Simon on the left side of the room to set up the boomerang kill.
Challenger: We used whip for no reason a few times to manipulation luck. Like the previous 2 runs, we waited a very few frames (this run waits 1 more frame) before loading the boss area just for manipulation.
On stage 8, that frame lost on the stair transition, was caused by whipping for some reason (I used for manipulation before). Fixed this.
--4--
Stage 10,11,12: Same patterns as last movie
--5--
Stage 13: This stage was a gold mine. Uses the hunchback manipulation technique from Grincevent's pacifist run to save ~43 frames. Had to be very careful to avoid lag on this section. Grabs a money bag on the way up the stairs. On the second part of the stage, uses the faster bone boost from Grincevent's pacifist run in conjunction with an additional bone boost at the top of the screen. This is enough to save two-three 16 frame rules at the door. Also got the meat faster with a new jump and whip technique which is a lot better (saves 6 frames) than waiting for a full 22 frame standing whip.
Challenger: I have watched Grincevent's pacifist run sometimes, but forgot to pay more attention for some parts, including the new hunchback strategy. It surprises me how nobody figured/pointed out since the publication on 2012.
Stage 14: Changed some of the drop manipulation in favor of grabbing a boomerang later in the run. Also leaves the second red bone due to lag issues at the axe-man.
Challenger: On the first room, I didn't kill that axeman in order to avoid an extra lag during transition this time.
Stage 15: Has to spend a few frames and throw a holy water at the bone pillar to conserve life for a critical strike at Death. Another extra lag transition added. Grabs a few throwaway hearts at Death.
Challenger: Death could be improved by using axes instead of holy water, but the enemies on the way between 14-15...holy water is still better.
--6--
Stage 16: Presses left and jump.
Stage 17: Wanted to switch up the Dracula fight. It's the most complex encounter and deserved a serious rework. Skipped the meat to save time. Also grabs a boomerang instead of the clock at the third skeleton. The boomerang was grabbed earlier in order to rack up a multi-shot before the Dracula encounter. Had to take a crunch at the big leap down towards the end of the stage.
Stage 18: Uses a 36 frame first cycle (solid, closed cape timing varies at 16,24,36,50 frames) on Dracula for 9 pips of damage. 5 from boomerang hits (one extra from the candle by his head resetting the boomerang), 1 from whip, and 3 extra boomerang hits using the fireball projectiles to reset the cross to damage Dracula in rapid succession. Second cycle is 2 boomerang hits, 2 whip hits, and 3 extra damage from boomerang & fireballs. No crits needed. (!) The second form uses extra holy water throws and a III multi-shot to skip one of Simon's jump cycles. The holy water does additional damage when inside the top section of Dracula's demon. It helps make up some of the time lost from not critting Dracula on the first frame of his second teleportation cycle. Retains the crit at the end to end input early like some of the previous versions.
It's very difficult to throw holy waters for extra damage with the second form fully standing in real time play (requires frame perfect whip and holy water or else you lose the extra damage). You'll tend to see speed runners let the "cookie monster" jump once so that he's in a crouching position (this allows you to get the extra holy water damage for each of Simon's jump, whip, holy water cycles).
Challenger: First of all, I'd like thanks to my partner Morrison. Even after nearly 10 years since this last run, your great knownledge with this game and TASing still works very well. This was the key how we managed to improve a "perfect" run to even more perfect. Also thanks to the previous authors of this game.
A few frames and this run could be sub-11:15, but for other side, this is still great (and this game was maxed even more).
Since I learned more of this game (as well as some new discoveries), I'll improve the pacifist TAS, someday.
Joshua: Thanks to Challenger who is a great TASing partner. His keen eye for new tricks and his knack for optimization were invaluable. I never would have imagined this kind of improvement was still possible for this game. Best of luck on your future projects. Thanks to previous Castlevania TAS authors Phil, Genisto and Bisqwit. I sent Phil a PM last year, but I don't think he visits the site regularly anymore or didn't see my message. I know he was proud of the 2008 TAS and might be interested in a new run. Also big thanks to Bisqwit for his disassembly and notes--for example, using pause for manipulation and knowing the drop system helped immensely. Thanks to the Castlevania speed run community for inspiration, ideas and rekindling my interest in NES CV1 (I have started speed running this game and have a time of 11:45 by speed run rules). I guess I can go back into hiding now until someone finds fresh strategies or something. Cheers.

Memory: Judging
Memory: Optimization is extremely on point, improving over previous TASes.
The audience loved this TAS. I'm not as super into it as some others but I definitely recognize the appeal and the work put into this.
Accepting to Stars as an improvement over [3492] NES Castlevania by Challenger in 11:19.03.
fsvgm777: Processing.
Note to self: Inherit tier.
Last Edited by adelikat on 10/27/2023 3:29 AM
Page History Latest diff List referrers