Tool-assisted game movies
When human skills are just not enough

Submission #1541: Swordless Link & AKA's N64 Super Mario 64 "16 stars" in 15:08.0

Console: Nintendo 64
Game name: Super Mario 64
Game version: USA
ROM filename: Super Mario 64 (U) [!].z64
Branch: 16 stars
Emulator: (unknown)
Movie length: 15:08.0
FrameCount: 54480
Re-record count: 29754
Author's real name:
Author's nickname: Swordless Link & AKA
Submitter: Swordless Link
Submitted at: 2007-05-03 02:03:15
Text last edited at: 2007-05-10 16:15:31
Text last edited by: DeHackEd
Download: Download (7651 bytes)
Status: published
Click to view the actual publication
Submission instructions
Discuss this submission (also rating / voting)
List all submissions by this submitter
List pages on this site that refer to this submission
View submission text history
Back to the submission list
Author's comments and explanations:

Video Plugin: Jabo's Direct3D8 1.6 Sound Plugin: Jabo's DirectSound 1.6 Input Plugin: TAS Input Plugin 0.6 RSP: RSP emulation Plugin

Together, AKA and Swordless Link worked in an attempt to beat Super Mario 64 in the lowest possible time, through several new opimisations as well as an entirely new route for one part of the game. They managed to finish the game in a mere time of 15 minutes and 8 seconds.

The new route saved about 6 seconds. It consists of avoiding Bob-omb Battlefield completely, and instead getting a third star from Hazy Maze Cave and Lethal Lava Land. The stars we chose were "Swimming Beast in the Cavern" for HMC (AKA's idea) and "Hot Foot it into the Volcano" for LLL (Swordless Link's idea). These stars seemed to be the fastest alternatives to BoB.

The never-before seen optimisations in this run contributed greatly to reaching the improved time. The optimisations may seem to be little things, like a dive in place of a long jump, or a cleverly executed wall kick, but overall they saved the rest of the time that wasn't saved through the new route.

AKA's comments

When I first saw the currently published run, I thought the movie was near perfect and could only be improved by 1 or 2 seconds, I had no interest in doing another run, mainly becuase it would visually look very simalar. However, when I saw a unique glitch in HMC being performed, I suddenly thought it might possible to grab the star and sure enough, with brief testing, it was, and it also worked out to be faster than the current slowest star. I was making minor improvements until by chance I stumbled across Swordlesss Link's WIP, which showed quite a lot of neat stategies I wasn't aware of. It had many improvements I didn't think were possible, although there were some bits where I was faster or I thought could be improved a little further, I proposed to him that we should work on it together and we also had a complete rethink on the all the strategies and lines of thinking at pratically every part in the game, some of the new strategies saved a very surprising amount of time. It was also shrewdly calculated that another star over the Chain Chomp star was faster to obtain when all the time diversions were calculated. In terms of work put in, it was near enough 50/50, a lot of time the person whose playing changes 6-7 times in a very short period of time, so no one was assigned specific sections to do. It should also be strongly noted that the real re-record rate is significantly higher than whats stated my estimate is probally over double what's stated.

Swordless Link's comments

I had been working on a TAS of this game in secret, and one day contacted AKA for help with editing a m64 for another run. It was at this time that he informed me that he had also been working in secret on a new SM64 TAS. We decided to show each other what we had so far, and discovered that there were parts where he had done better than I had, and parts where I had done better than he had. Granted, neither of us were that far into the game, but we decided at that point that the best solution would be to co-author the run, so that's exactly what we did. AKA suggested going to HMC for a third star, instead of the Island star in BoB. I then suggested getting a third star from LLL to eliminate any need for BoB, and AKA timed this to find that it was considerably quicker than the old route. Most of the time saved over the previous run, however, comes from our optimisations, and little things we discovered along the way. For example, I found that you could grab the third Bowser while he was in the air, which saved a reasonable amount of frames. Although this altered his final position, which placed us not quite under the star, the amount of frames gained by it outweighed the few frames lost by it. Another new strat we used was when backwards long jumping on the 50 star door stairs, I came up with the idea of long jumping onto them first. AKA then came up with the idea of getting Mario inside the next set of stairs to save time. Both of these combined saved a lot of time in that part of the game. We achieved a time of 15:08, after hoping initially to beat the current run by around 3 seconds.

People we want to thank

Mukki - For discovering the glitch we used in HMC.

Rikku - For your previous runs which served as a fantastic guide for most of this run.

DeHackEd: AVI under way

Similar submissions (by title and categories where applicable):