For run submissions, we allow the Dolphin emulator, but not Project64. Dolphin supports .wad files for Virtual Console games, so hopefully that will work for you instead of Project64.

Since we've started to receive leaderboard submissions for the NSO/SwitchVC of Yoshi's Story, I took the time to do some coarse time comparisons between SwitchVC and WiiVC. This will help inform us on how we might want to handle the SwitchVC platform in the future.

SwitchVC time saves:

• 4 seconds saved per reset; 20 seconds total
• 15 frames saved per level complete animation; 1.25 seconds total
• 15 frames saved per level load; 1.5 seconds total
• 40 frames saved per screen transition startup, 4.5 seconds total
• 2.75 seconds saved over all the any% screen transition animations
• Total: 30 seconds saved

These are mostly approximations, but SwitchVC comes out almost exactly 30 seconds faster than WiiVC for any%. Most of this time comes from faster resets: SwitchVC can properly "soft-reset" the game, whereas a reset on Wii is actually a hard reset that takes a few seconds to load the ROM again.

Aside from resets, SwitchVC improves loading times and eliminates even more lag than WiiVC does.

See the other thread (link) for more information on how we've decided to handle NSO/SwitchVC submissions. Right now, we're keeping them on the same leaderboard because it's the most inclusive way of accepting runs from new runners, and giving them the satisfaction of having a run on the leaderboard. If we end up getting a whole bunch of SwitchVC runs with competitive leaderboard times (eg. sub-12) then we'll consider moving them to their own leaderboard, such as to not discredit the fantastic performances of our top N64 and WiiVC runs.

NSO/SwitchVC is now added, and runs for this platform can now be submitted.

The NSO version of the game is faster than the other platforms. However, NSO is a subscription service that may disappear someday, so we're discouraging it for players who are aiming for the top of the leaderboard.

So, our decision is to allow NSO version runs as long as the leaderboard position is 6th place or lower. In doing this, we can make the game as accessible as possible to more casual runners, but still maintain the other platforms as the definitive competitive versions of the game. We already have this rule in place for emulator runs, for the same reason.

In the future, when Switch is a last-gen console and the NSO N64 library is retired and no longer accessible, we will look at the homebrew options for accessing it. If the game is easily accessible via Switch homebrew like the WiiVC version, then we may be comfortable allowing it for top-placing runs, in which case it will effectively replace WiiVC as the #1 platform.

Hey everyone, I was working on getting SM64 Randomizer to work on Wii VC, and along the way I decided to look into that annoying darkness issue that SM64 and some other VC games have. The fix is very simple to do, so I figured I would share it here.

Whether or not this fix would be allowed for leaderboard qualification is up to the mods, but since this is beneficial to game enjoyment regardless of leaderboards, here's how you do it.

1. Unpack your SM64 wad (I use a program called ShowMiiWads)
2. Open 00000001.app in a hex editor
3. Search for the hex "00 00 00 BE 00 00 00 BE 00 00 00 BE". There should only be one search result. In my file, the location is at hex offset 0x16B0B0.
4. Replace each BE with FF and save the file.
5. Repack the wad and you're done.

Enjoy!

Comparison image:

https://gfycat.com/TalkativeSatisfiedInvisiblerail

As a quick preface, this glitch currently does not have any use in speedrunning and likely never will, but it would be nice to keep documented in some way.

General info

We've known for a long time that Yoshi's swallow animation has a strange behavior where state changes don't always work properly during swallow. This is most notable in the JP version, where you can jump midair while swallowing if you slide off an edge while eating a fruit.

It turns out this also applies while swimming, where Yoshi cannot transition from swimming to walking while he is swallowing a fruit. This means you can use a slope to nudge Yoshi above the water line during his swallow animation, and the game will put him back into the swimming state, because none of the conditions to leave the swimming state have been triggered. (The "swimming swallow" state does not have any conditions to leave the swimming state.)

Entering the loading zone while Yoshi is swimming simply puts Yoshi on the ground again.

How to control air swimming

The swimming state itself checks regularly to see if Yoshi is transitioning from water to ground. Here is how the checks work.

Swimming upwards: Ground detection never happens. You can swim upward infinitely and soar into the skies. Swimming horizontal left/right: Ground detection happens every stroke when the animation loops. Yoshi will exit swim and start to fall at the end of the stroke. By interrupting the stroke animation with tongue, you can avoid this. Example: https://gfycat.com/ConcernedElasticElephantseal Swimming downwards: Ground detection happens every frame. Yoshi will exit swim and start to fall immediately.

Where is this possible?

The glitch is only possible in levels that have a slope entering the water, and something to eat next to the slope. I believe this is only 4-2.

Special thanks to LoveBot for accidentally discovering this glitch.

After being curious for a long time exactly how much time VC saves over N64 in any% runs, I finally did a side-by-side comparison. It turns out that VC doesn't save any time at all!

I used my PB on VC and compared it to michaelgoldfish's PB on N64. I removed all gameplay so that only the screen transitions and resets remain. In the end, N64 is about 1 second faster, but VC might save a tiny bit of time on Bowser due to no lag.

VC also doesn't require 2-2 lag reduction, which saves about 2 seconds if you don't use that strat. If your 4-1 uses the method where you enter two huts, then VC will save you an extra second or two because of one additional screen transition. Between Bowser, 2-2, and 4-1, VC might save 3-4 seconds for most players, but the time becomes more negligible as the run becomes more optimized. This means neither console will be a stronger contender than the other for WR, and you're free to use either one and trust that the leaderboards are accurate.

VC benefits much more in other categories because there are more screen transitions.

Edit: Not included in this video are level load times. VC loads each level about 20 frames faster, so that's 2 seconds saved for VC throughout the run. It's still close enough that there isn't a strong difference between the two consoles.

JP leaderboards are now timed according to the Japanese niconico community: Timer starts on reset, and ends when the final score is finished displaying on the high score screen. All the existing JP runs have had their times adjusted to match this. When running on JP, please time and submit your run according to this.

Also, Japanese players rule N64 and WiiVC into separate categories, because they decided WiiVC carries too much of an advantage over N64. So there are now three regions: US/EU, JP (N64), and JP (WiiVC).

I picked up the N64 version of Yoshi's Story to compete with Japan, and it turns out that 2-2 is extremely laggy (VC has no lag). The main culprit is the dynamic lava and the lighting effects created by the fireballs.

Yoshi has a "sniff" mechanic that zooms the camera in while holding the R button. Using sniff for the entire duration of the room saves about 2 seconds over the normal method. Unfortunately, this has some ridiculous downsides:

¤ You have to hold R the entire time, and you already need to mash B while holding A during every jump, so this is a genuine hand killer ¤ Yoshi walks really slowly while sniffing, so you can't spend any time on the ground whatsoever ¤ You can't see shit

The clip on the left took me about 2 hours of attempts, but any% is such a short category that I'm probably going to incorporate it anyway.

Still working on some routing stuff, but I confirmed that if you successfully manip four melons 5 times in a row (with deaths in between, and a different RNG pattern each time), then you save about 30 seconds on the level compared to the normal all melons route. I'm thinking about tackling a doable route for this. It would make sub 20 possible for melon% 1-lap.

The above method saves only ~2 seconds compared to getting good luck on 5-3. So while it greatly improves consistency, it'll be worth finding an RNG pattern that doesn't require Yoshi to flutter back to the left. That would also mean less time is wasted if manipulation fails.

Here's my progress on a faster route, but I'm having some trouble getting the fifth melon to reliably spawn. Everything in the gif is reliable. This would save ~8 seconds over a lucky Lakitu. https://gfycat.com/RealScalyAlbertosaurus

Hello everyone, I want to talk about a ground-breaking 5-3 strat that we knew was around for a while, but I finally got around to routing it.

What you're seeing is RNG manipulation of the Lakitu so that he drops all 5 melons in a row, without any spiked balls. Normally, the chance of dropping a melon is around 10-15%, so getting unlucky can be a run killer, especially in All Melons 1-Lap.

This is possible because we reset the console before each level in the US version, which also resets the RNG seed. The seed only increments every time a random event is triggered, meaning we can manipulate it for a short period of time.

There are a few Yoshi actions that trigger the RNG, the most important being his footsteps. Every time he takes a step, the RNG determines which footstep sfx to play. This means we can roll the RNG a specific number of times by taking a certain amount of steps. So, if Yoshi's footstep sequence is the exact same every reset, then the Lakitu will always drop the same pattern.

I start by taking a specific number of steps at the beginning of the level, jumping at the right side of the green bush. I then trigger the Lakitu and flutter back to the left side of the screen. Once there, I ground pound, which prevents the RNG from rolling as Yoshi hits the ground.

I continue holding down on the control stick and do Yoshi's crouch turnaround. By staying in crouch, Yoshi doesn't hum along with the music, which would again throw off the RNG.

With this specific seed, the Lakitu will drop two melons in a row. After that, we roll the RNG a specific number of times in between each Lakitu drop to make sure it's on a seed that will drop a melon.

Once you see the second melon come out of the cloud, take a small step forward and eat the melon in front of you. While Yoshi is in his swallow animation, he isn't using the RNG. That one step we took prepares the RNG for the Lakitu's third melon drop, so we let Yoshi's swallow animation finish completely while the Lakitu drops the third melon.

You now run toward melon #2. Your footsteps are again preparing the RNG so that Lakitu's fourth drop will also be a melon. To take the exact number of steps, run directly on top of melon #2 and tongue to stop your footsteps and eat the melon. While you're swallowing, you'll see Lakitu drop melon #4.

Yoshi's swallow animation takes too long this time, so we need to cancel it by tapping the B button. Then, run past melon #3 and as soon as you're barely past it, press B again. You'll eat melon #4, and Lakitu will drop the final melon while you're in your swallow lag. From there, you're done and can eat the rest of the melons.

This might all sound crazy, but trust me that you can get it consistently with practice. It's based entirely on visual cues, and it's not frame-perfect - you just need to put in the correct number of steps in between each melon drop. However, it's almost impossible to practice without save states, because you need to reset the console in between attempts. So take this one to the emulator.

I hope to publish a video tutorial in the near future, which will help make things a little more clear.

Hello friends,

I am proposing that we create sub-categories to differentiate between US and JP. In doing so, there will be a separate leaderboard for US and JP regions.

I believe this is necessary because the JP version is substantially different enough that it is meaningless to compare it to the US time. For example, cutscenes cannot be skipped in the JP version, adding minutes to the any% time.

Furthermore, Japan has an active Yoshi's Story speedrun community, and it would be neat to add their leaderboard info to this site. I will also try to convince them to keep the leaderboard up-to-date with their times, as they currently rely on a shabby HTML page for their leaderboard. It will allow us to compete against JP players' times.

Unless there are any objections, I will soon add the sub-categories and re-organize the current leaderboard to take regions into account. Thanks for your consideration and input.

No need to ask us if it's legitimate - route it if you're interested in it. The most interesting thing about the route would probably be figuring out which power hearts to get. It's not something I'm personally interested in investing time in, because I run Yoshi's Story strictly for melon%.