Submission #6771: Acmlm's NES Destiny of an Emperor in 1:16:02.50

Nintendo Entertainment System
baseline
FCEUX 2.2.3
274201
60.0988138974405
11925
Unknown
Destiny of an Emperor (U).nes
Submitted by Acmlm on 6/5/2020 3:50:03 AM
Submission Comments
  • Recorded with FCEUX 2.2.3, old PPU
  • Fastest time
  • Luck manipulation
A 05:07.11 improvement over the current published TAS, almost exactly 10 years later.

New changes or improvements

"First character cancel" glitch, or how Zheng Mao became the new leader of the small group of Yellow Scarves

A relatively recent discovery, and the reason I started this TAS.
The glitch was previously used to remove Guo Ji (id $12) from one battle, but if this is done while Liu Bei is the only one alive, it will also exterminate all lower id's ($00-12), including:
  • Yellow Scarves: Zhang Jao, Zhang Bao, Zhang Liang
  • Cheng Yuan Zhi (only at Zhang Jao 1, then he's back)
  • Li Su and Guo Ji
  • Brigand NPC in the back cave to Zhang Jao 2
  • Worthless recruitable generals like Song Yong
So I now take an intentional death at the start, and no longer need Han Zhong since the front battle got much easier (it's even required now). As a nice side effect, I also gain less EXP, making it finally possible to reach Xu Zhe at level 11 and gain max TP for Fu Bing! (and yes the extra TP was used later)
This saved about a minute and half overall.

Savewarps

When you save the game and reload, you restart in front of Liu Bei where he was last invited. If you invite him at a wrong location (says he's going somewhere else), it still moves him where he needs to be.
This saved about 10 seconds at Ji Zhou (after Yuan Shao), but has very limited uses otherwise.

Luck manipulation

I went back to the bruteforcer I made last time for All-Out and random encounters, and expanded on it a bit (and made it faster).
Previously I would start each battle with a few different RNG values and bruteforce All-Out on each to find the best results, but now I generated savestates from thousands of step delay combinations (ideally leading to unique RNG/timing) to bruteforce all of them. The results eventually tend to converge to near duplicates since the RNG sequence is only determined by 15 bits, but there was a definite improvement from before.
This also made the An Sha sprees more efficient, by finding fast ways to clear all rebel forces first and even manipulate tactics like Yi Xin against myself to skip unneeded moves/menus.
This was probably most of the improvement after the glitch and savewarp, as I just seemed to gain time everywhere.
However it's still mostly limited to All-Out (and the forced "caught off guard" in Wei), so battles starting with manual actions (especially Ma Chao) or not directly preceded by overworld RNG (Zhao Fan) didn't gain much.

Other

  • Only 9 inn trips (previously 14), largely thanks to the improved luck manipulation
  • Minor route changes (skipped Zhou Cang, delayed Huang Zhong) and party order/items
  • No glitched text skip freezing the game
  • Gullwings before/after Xu Zhe saved close to 1 second, the other gullwings must still be bought separately (needed gold)
  • The overworld menu opens on a framerule, so I do it on screen transitions (always fast) where possible to save some frames
  • Ma Chao missed the attack bonus from 5 digit HP at the end, but healing takes longer than it's worth
Time comparison:
OldNewGain
Yellow Scarves22434168955539
Dong Zhuo53913477706143
Yuan Shu75684685117173
Jing Zhou1084619818410277
Ji Zhou12032210987910443
Shu16687515403812837
Wu23109621573715359
Wei (end)29265827420118457

Truncated: I'll judge this.
Truncated: Everything looks good, nice improvements. A worthy successor to the previous 10 year old movie.
feos: Pub.
Last Edited by Memory on 1/7/2022 9:16 PM
Page History Latest diff List referrers