Since May 31, 2018, I’ve been creating a 120 Shine TAS (Tool-Assisted Speedrun) of Super Mario Sunshine solo. As of January 2020, 1 year and 8 months in, I finally surpassed the 30 Shine mark — one-fourth of the total.
Rather than doing one massive retrospective after finishing the entire run (which would be exhausting and I'd forget the early parts anyway), I’ve decided to write a reflection every time I surpass 30 Shines. This time, I’ll be looking back at the portion covered in videos up to Part 4 — specifically, up to the Shadow Mario chase in Ricco Harbor. That’s 29 Shines + 86 Blue Coins.
RNG manipulation and route testing are such a pain!
That pretty much sums it up. I remember constantly being either furious at RNG or exhausted from researching which story segment is the most efficient for collecting each Blue Coin. But I think I worked really hard on optimization, considering I improved nearly every area over the Any% route. Realistically, this isn't a workload one person should take on, but Dolphin’s movie file desyncing issues are so bad (a problem even recognized by TASVideos) that I basically had no choice. There are also only a handful of people in the world capable of optimizing to this level.
That said, it was rough knowing that before reaching Corona Mountain, most of the content is still based on Any%—and yet I was already struggling this much. I often worried whether I’d make it through the later, more complex parts like 100-coin missions, dandelion seeds, and Cheep Cheep Flower (Chuuhana). But every time I watched my completed TAS at full speed, those worries would melt away.
In terms of optimization, I regret that I had to leave some room for improvement in a few places. I’ll talk more about that later, but so far there haven’t been any fatal oversights. Most of the remaining optimization is due to either RNG I couldn’t control or techniques nobody had thought of at the time. So I don’t think I missed anything obvious. It’s possible that if I redid some levels, I could still get a mysterious time save somewhere—but overall, I’m proud enough of the current version to say I did a great job. Honestly, I believe no one else could’ve taken it this far but me.
Of course, I hope that even after I eventually retire, someone else will continue exploring Super Mario Sunshine and maybe even pick up TASing. But at the same time, this game’s RNG is such hell that I sometimes doubt anyone would want to. Even RNG elements that don’t affect RTA runs — like the RNG on Hover Nozzles — can totally mess with you in TAS. And having to manipulate RNG while adjusting coordinates down to the decimal? I can’t count how many times I wanted to scream in my room. So yeah, those are my general thoughts on the project so far. Below, I’ll talk about the sections that left the strongest impression.
This route, which started out as a fun experiment, ended up becoming the most powerful strategy.
Normally, the water striders disappear if they reach land, so I was just messing around with that when I accidentally discovered that one could be stomped on from solid ground. Up until then, the best we could do was dive recovery stomps, but now we could use water sliding to reach max speed (100) and pull off huge jumps. I got chills when I found this.
The tradeoff is that you need pinpoint RNG manipulation, and guiding the strider is a huge pain. Thinking about having to do this again in the secret red coin missions makes me depressed. But hey, the water-slide jump from the strider just looks so damn cool — I’ll push through it.
The objective is the same as in Any%, but this Shine really showcased my optimization skills. I honestly didn’t think I could save 8 frames compared to Any%, and there's still one more frame that could be saved at the start.
Adding two extra Blue Coins and still defeating Petey Piranha using the fastest route with zero time loss — this was another major win for my optimization skills. I don’t think many people could’ve pulled this off. Oddly enough, aside from collecting those two extra coins, the rest of the level was relatively simple.