This is an "Any Percentage" completion of Mario Kart 64, to see the end credits players are required only to beat the Special Cup on 150cc and that's exactly what I do. This run began as the next step in proof-of-concept runs I'd been doing for my research work with ComicalFlop's 16 course run. I had done a Flower Cup run earlier this month and concentrated purely on strategy exploration with time and optimization as a secondary importance. As I began the Special cup I remembered that one need only beat this single cup to view the credits and be greeted by Mario saying "Hey, you're very good! See you next time" as the words "THE END" bounce across the screen. So, I figured I'd test out the TASing skills I'd been developing thanks to Comical Flop's guidance and make this cup as optimized as I could stand. I think the result is quite impressive and certainly entertaining. I developed new strategies, especially on RRd, with surprising results.

Mario Kart 64, Any Percentage Completion

SETTINGS

Course Times:
DK: 20"21 (10"48, 6"21, 4"22)
YV: 29"43 (9"33, 10"33, 9"77)
BB: 1'29"80 (33"36, 22"99, 32"45)
RRd: 1'02"32 (34"95, 13"92, 13"45)

Frames:
19,045 Last race frame (5 min 17.42 sec)
20,018 Second to last frame of necessary input (5 min 33.63 sec)
24,194 Last frame of input to initiate the End Credits (at 6 min 43.23 sec)

8317 rerecords

Credits end at about 9 min 34 sec.

Suggested Picture:
Frame 10,923

First, I use two extra frames at the beginning to choose Toad rather than Yoshi. For the record, I have no reason to believe that there is any difference between Toad, Yoshi or Peach in terms of performance and control (and the variety of characters used for WR’s over at mariokart64.com is further proof of that). However, there is a discernable difference in weight between Toad and Yoshi/Peach. The difference is very small but every bit helps for SC’s like Banshee Boardwalk. So, I chose Toad to ensure maximum hops off of cpu players.

DKJP:
I tried two very different strategies on this level but neither panned out. They were: 1. CPU hop back through the trees on the left and somehow drop down into the tunnel while triggering the regular SC. This completely failed.
2. More promising, I drove forward and cut around the trees by the water towards the waterfall. I managed to get behind those trees and my hope was that I would be able to jump down into the cave from here to speed up the first lap. Unfortunately there is a large, invisible polygon that forces a tumble and put me right into the drink no matter what I did.

So, I ended up using the typical SC strategy and just optimizing it. The very first turn is more aggressive and direct than normally attempted on a console. Please keep in mind that if you try to compare this run to the WR times, I am playing in 150cc when Time Trials is approximately equivalent to 100cc. However, I have only two mushrooms.

YV:
Now for a course where the SC brings me nowhere near an item box. With no mushrooms I was forced to get creative with the strategy. The first lap was fine, the SC is perfectly doable without an extra shroom. It’s the second and third laps that are tricky. The third lap strategy is obviously the fastest (and just barely possible without a mushroom) but when Lakitu puts you back on the track it is further forward and just far enough to make it impossible to trigger the SC (at least based on my testing). So, I came up with what you see in lap two as a way to allow three quick SC’s because this strategy places me in the right spot for lap 3. I would like to dedicate this particular trick to the late Martin McMullan who discovered the fast lap strategy that uses a similar fence bouncing technique.

BB:
First, I’d like to mention that the wall-hit early on in this level was intended to demonstrate how some corners can be used to make abrupt corners. This is especially true in Bowser Castle (not shown in this run) but it may not have been faster here. No matter, I had to wait a bit for Bowser later on anyway. For the record, manipulating cpu movement is next to impossible. Hopefully, with time, a greater understanding of how to do this can be obtained. I first performed the massive jump about 2 years ago using save-states but no slow down or re-recording. I think it is worth noting that I cannot guarantee this jump will be possible in Comical Flop’s run that Fried and I are supporting. Bowser happened to swing wide and fast and I was just barely able to get enough speed by shroom-sliding.

RRd:
I would like to thank Malcolm Young for making a video of a lap trick on this course back in 2001. He used toad to jump off of Bowser in Versus mode to trigger the strange SC. This was largely an inspiration to me to search for more interesting GP mode SC’s involving cpu jumps and items. In this course I took Malcolm’s discovery to another level and used items readily available to me from the conveniently close item boxes behind the finish line. My hope was that I could go back to those item boxes and grab two sets of triple green shells never to return. However, I was surprised to discover that it is impossible to get triple green shells in last place (a strange predicament). So, I decided to try my luck with red shells. Unfortunately, if you hit yourself with a red shell and fall off the side, you continue to tumble straight down (unlike the green shells). So, I grabbed a set of mushrooms as well (triple rather than golden, so I could use more after Lakitu brought me back). I knew that one way to protect yourself against a red shell was to jam a mushroom at the last possible and hope and, if you were lucky, your character would make a big hop rather than tumble. With TASing I didn’t have to hope, but I did have to experiment to see the cause. Turns out this occurs if you use a shroom the frame you would normally get hit the jump occurs on its own. In fact, if you press R to jump during this frame, you will get hit.

After this strange part of the first lap, I used the magic of AB-spin-shrooming to make some tight maneuvers and grab a set of triple green shells. Please note that hitting the wall at the very end was somehow “01 faster than not hitting the wall. If you watch the speedometer, it indicates that this does not slow me down but it does allow me to spend less time turning.

Truncated: Firstly, I don't like the idea of running only one cup. That would either mean that we publish every cup as a different movie, or have both a "full run" and an "any % run" which would be identical for the last cup. Neither of these is a good idea.

Secondly, some people have pointed out that this movie is badly played.

For these two reasons, this submission is rejected.