I need to address something (Script/notes)
Added 2021-05-17 01:43:27 +0000 UTCWHY I made the video
Floor clip analysis
left + right
These things seem to have been lost on some people so I would like to clear things up.
~FLOORCLIP ANALYSIS~
I've seen multiple people say that my argument for the floor clips being suspicious was not compelling enough, so lets dive a little deeper into the floor clip mechanics and how exactly they work and why it is very clear that no human could achieve a run like this without cheating.
In my original video I didn't want to take up a ton of time explaining the floor clip mechanics and then apply statistical analysis to all the clips in the run to try and prove that the run was cheated as I felt the rest of the arguments in the video would more than suffice. However I underestimated how hung-up people could get over how I structured my arguments instead of the fact that if you combine all of the suspicious activity together, the conclusion is clear that the run is cheated regardless, which was the main goal of the video. However, That is what we will be doing now as I OVER-Explain that this type of run is not feasible for a human to complete without cheating.
Floor clips in sotn are dependent on a few factors. the speed of the wolf, the timing of the jump and untransform, the collision/angle of the slope, as well as the positioning of the wolf (both pixel and subpixel positioning.) The simplest way I can explain subpixel positioning is that, you can think of it as a hidden value stored in the game that will determine where you end up after certain movement. With different subpixel values, your positioning can be different even with the same exact movement and button presses. The simplest example of this is alucard's walking speed. Alucard's walk speed is 1.5 pixels per frame, so on the first frame (assuming his subpixel value is 0) he will move forward by 1 pixel, followed by 2 pixels, then 1 pixel ect. But with a stored subpixel value, this can cause Alucard to start walking by 2 pixels forward instead of one.
Pretty much every action in the game used for speedrunnning completely scrambles This subpixel value, causing it to constantly be shifting around. Because of this, there is no way to know for sure your subpixel value without doing some form of alignment setup, such as rubbing up against walls and doing consistent movement setups, like Crouching full jumps, full backdashes and normal walking for extremely precise amounts of frames.
And to tie this back to floor clips, These floor clips are insanely precise as I stated in the previous video, and have very strict windows to perform the jump and untransform in order to actually clip into the geometry successfully. However these tricks become even tighter without a subpixel alignment and sometimes become flat-out impossible to perform. So, if someone were to attempt these floorclips with irregular subpixel values, like the cheated run did, the success of these tricks no longer becomes about executing them with the perfect timing of button presses, because even that is not enough. So if the floor clip requires pixel perfect positioning (which most of the ones in the cheated run do) The success of these tricks becomes effectively random. Making the highest possible chance of clipping into the floor, even with perfect execution, only a 50% chance assuming the floor clip subpixel requirement is binary and doesn't require smaller ranges of subpixel values to work.
For the sake of time and simplicity, Lets assume every floorclip is on average the same difficulty to perform. A frame perfect Jump input, followed by a frame perfect untransform. For reference the game runs at 60 frames per second meaning a frame perfect button press must be performed within a 16.7 millisecond window. The average blink of an eye is 100 milliseconds for reference. So one 6th the blink of an eye is one frame. hitting two frame perfect inputs back to back in a short period of time is extremely difficult.
For example, Even after putting in Tons of hours of practice into this floor clip (library floor clip), getting this unbuffered floor clips is still extremely inconsistent. This floor clip if performed unbuffered will save upwards of 2 seconds over how its performed in my world record run that uses map buffering. So I have been practicing doing this clip unbuffered every day for about 30 mins a day in order to try and save that time when I try and get a new WR in any%. I even made a video with audio cues to sync with the button presses to internalize the timing of this trick. This floor clip also has a subpixel positioning setup where you walk against the ledge and perform a mist into the wolf transformation within a 7~8 frame window which causes the wolf to dash automatically by just holding left. This is known as an autodash and puts you in the same positioning every time. And With all this preface, I think its safe to assume that this is a near perfect case scenario for someone attempting a floor clip like this in the game without map buffering.
Both myself and Talic tried to see how many floor clips we could get in 5 mins and I will be using this as my metric for the percentage chance of how likely it is to get a floor clip like this. https://youtu.be/lEbULCK9Q2w < Video Link
in 63 attempts where I dashed and untransformed into the stairs, I got 6 floor clips. So in this sample I had a 9.5% success rate. Lets even give this person the benefit of being built so differently as a human being that they are twice as good as myself (the current world record holder in every Alucard Category on the leaderboard). But because they use no positioning setup this lowers their consistency for these tricks by at least half. So Simplifying this down we can assume that the runner would have about a 10% success rate at these floor clips if I am being as generous as possible with how the subpixels will effect their success rate and If i assume they are twice as consistent as me at floor clips. Even with all of those leniencies, the odds of getting 7 different 10% floorclips in a row is 1 in 10 million. To put that into perspective, a run of the category shown takes about an hour and a half to complete. That means a run like this is statistically expected to happen after 15 million hours of gameplay. That's equivalent to 1,712 YEARS of gameplay before a run like this is expected to happen. For more perspective, the odds of getting struck by lightning in your lifetime on average is about 1 in 15 thousand. The odds of getting bit by a shark are about 1 in 4 million. So clearly this dude has been practicing his floor clips since 300AD. And that's just the odds for the floor clips being successful that doesn't include the odds of getting any of the other frame perfect tricks first try either without any mistakes. The odds become so ridiculously farfetched that its not even worth calculating the actual odds at this point and just say it is not possible without finding a way to make those clips more consistent in some way. So that is why it is suspicious that they chose not to set up those clips using some form of subpixel setup and map buffering to help make them more consistent, otherwise they would have to be resetting for LITERALLY THOUSANDS OF YEARS to get a run this good.
~left plus Right~
Real Quick
I claimed that it is not possible to perform left + right at the same time in sotn without on original hardware and without using modified controllers or forcibly activating the left and right sensors on the controller and bypassing the dpad altogether. This is why the rule is put in place on the speedrun.com page as it gives an unfair advantage to individuals who are playin on emulators that allow for left + right to be pressed by using a keyboard or any other various methods. Some individuals have claimed that it is possible to press both left and right at the same time by using a dual shock controller and pressing one direction on the d-pad and the other on the analog stick. One individual went as far as to post about this on reddit and as well as harass myself and others in the sotn community about this genius revelation of theirs. However, they forgot one simple fact that SotN was programmed and released before the release of sony's first ever analog stick controller, and the game has absolutely no analog support. The game does not respond to any inputs when the analog button is pressed. In fact, the reason there is an analog button in the first place, is because in order to be backwards compatible with games that were made before the invention of sony's analog controller, they had them switch between analog and digital modes as to not cause problems with those games that were made before the analog controller. So when in analog mode the game accepts no inputs and when in digital mode the game ignores the analog stick as it is designed to do.
~LINKS AND REFERENCES~
Mouse's video on ps1
https://www.twitch.tv/videos/1018583642?sr=a&t=1s
https://ultimatepopculture.fandom.com/wiki/Analog_stick#:~:text=On%20April%2025%2C%201997%20Sony,controller%20for%20a%20games%20console.
