Become a member to get access to these awesome emotes along with extra features, or if you want to see videos before they're uploaded become a Patron where you also get updates on what I'm working on: www.patreon.com/abyssoft
@@Mai.Calico I stream on this channel several times a month, and premiere every video, so there's lots of opportunity to use them in the community aside from comments on videos.
I hate this. What is never explicitly stated is that in order to observe something, you have to EFFECT it. To detect a particle you have to interact with it which of course changes it state. I hate that in physics things are made more confusing than they need to be constantly.
@@raafmaat I've already gone though this garbage when I took QM when I was studying to become an electrical engineer. I'm trained as a VLSI designer actually. The fact you're disputing what I said shows you don't understand QM. So tired of the stupid mysteries around all this crap. "Oh if you observe it, it changes is". This concept is CONSTANTLY promoted by the stupid fraud Deepak Chopra. I'm not going to argue something I understand, and I'm not waste time talking with somebody that does, but is fully convinced they do.
Playing Mario Kart 64 a lot growing up, I absolutely noticed the rival system even if I couldn't ever put it into exact terms. I could tell that 2 of the CPUs always were better than the others, and 1 of the 2 was also just a little better than the other. Cool to see the exact mechanics of it laid out and finally justify what I've been saying for years.
I found it in Episode 1 Pod Racer as well.... and then realised I could use the debug cheat to slow the rival down. Always funny to watch a pack of races crawling along behind a rival because their AI won't let them overtake at all in that game XD
The rival system is a staple of the series, but its mechanics vary depending on the game. My four years old self also noticed its existence in Super Mario Kart. Unlike in MK64, rivals aren't random and it's not just two but all seven of them. The game is rigged so that CPU drivers usually finish the race in the same order. One is programmed to be in first place, one goes for second place, one for third place, one for fourth place, and three for last place. If CPU drivers are not in the correct order, the game adjusts their rubber band speed in an attempt to fix it. The rival order depends on who you're playing as. For Mario, the order is Donkey Kong Jr., Peach, Yoshi, Luigi, Toad, Bowser, and Koopa Troopa. The last three have the same speed and are programmed to stay in fifth place or worse. They don't compete against one another, so how they place in relation to each other depends entirely on how much abuse they receive from the player. They can be really close together or really far apart. Since no points are awarded to drivers who finish in fifth to eighth place, it makes no difference. A good strategy is to prioritize attacking the fastest rival, especially during the final lap, in order to skew their placement in the rankings. The second fastest rival will never aim for first place. It will yield to the fastest rival in all races. Once you successfully make the fastest rival finish after you in third place or worse, it makes winning the grand prix much easier.
I also noticed it. Not the specifics of course, but it wasn't too hard to figure out that there were always 1-2 standout racers. Still, this video explains a lot of the mysterious cheating I've encountered in the game such as a 1st place CPU finishing when I barely got started on the final lap in 2nd place. I've also had it happen that a CPU just straight up kept driving at full speed for an entire race without being able to catch up on it (probably a glitch or something).
Me and my sisters always use to hate the baby characters and Mario on the Wii versions because they always seemed to outperform the others Guess it’s just cause we all picked characters who had them as rivals
My favorite was Wario Stadium, making all the turns as close to the inside as I could while constantly drift boosting and watching the CPUs try to keep up since they don't ever try to not make a wide turn. I end up getting further and further away and it's really funny
@@LunarBloodRose27 its also hilarious when they cheat hard when you do the wall hop >_< somehow it stirs their coding when you are suddenly 3/4ths across the track repetitively.
@@Abyssoft Would be curious to see how the AI cheats in Sonic Drift 2 on the Game Gear, since I feel like its rubber banding is even worse than Mario Kart 64. (especially if Metal Sonic is in 2nd, he'll just dash into first whenever he feels like it.)
@@Abyssoft In the Advance Wars games for the GBA, the CPU would cheat in Fog of War, knowing exactly where your units were & could even fire upon them when they normally wouldn't be visible to another player. Makes me wonder if the Advance Wars 1+2 for the switch does the same thing or did they finally make it fair like in Days of Ruin.
Wario coming in and clotheslining Yoshi through the wall gave me such a good laugh, I've been terribly sick the past few days so I really needed it. I appreciate it a lot, thanks for another great upload!
Fun fact: You can trap a CPU player into the train tracks on Kalimari in GP mode and it will remain "jailed" as long as at least one human player can visibly see the player on screen. The moment you drive away and place the CPU out of view, it can glitch itself outside the "jail." When I finally acquired a GameShark in the early 2010s, I combined approximately a dozen codes together to form what some in the community call the "Wonn challenge." This allows human player(s) to race against up to 3 CPUs in 4P Vs. mode where they can maintain draft mode at will and even turn on a dime in mid-air off of the big jumps, like in Royal Raceway. In this mode you can see all kinds of other interesting effects as well: Bolts will happen out of nowhere, sometimes zapping all 4 players. You might even get bolted thrice in a row! The first item box they pick up never gets used, so it's ripe for stealing with a ghost. But they'll drop bananas, fake item blocks, and activate ghosts or stars seemingly at will. Also when a CPU runs into a banana, they slide forward (since they're now always visible) and they'll additionally crash upward at the end of the slide as though they ran into a fake item. This is much like the effect you see with the last place human player at the end of an all-human Vs. race if that player didn't have a Star activated. If you're too aggressive with bananas near the end of Choco Mountain, you might get 2 or more CPUs stuck in such a way that they never get back on track, effectively soft locking the game, so this could result in failing the "Wonn challenge" depending on your rule set. Had the CPUs not been visible, they'd find a way back on track...
@DavidWonn: The player can only see his own item box, so we can't see the item box the CPU have on the road, but the same thing since we don't see the cpu item box, we wont know if we accidently broken one of his , one time the cpu behind me used no item because I must have run into one of the cpu item box that the player can't see, the same think with the cpu they don't see your item box on the ground but can still break them so they would not do on purpose to take your box because they actually do not see it. Notice in toad turnpike the player item box are on the side, and the cpu never goes their unless push into that area, how they do get item if the never go to that area, their own item box is place where they follow the path this is how I know that I accidently pass an A.I item box because the one dirrectly behind me did not toss anything.
I recently played MK64 for the first time, and while I knew CPUs cheated, witnessing it first hand was astounding. It's insane. I also did notice a lot of the BS the CPUs do such as them recovering fast from items when I couldn't see them, but didn't know all the intricacies.
@RaniaIsAwesomeAlmost makes me wonder how CPU behavior and handicapping changed in later games. I KNOW Mario Kart Wii has some shenanigans with those infamous "item chains" they get on you at the end of the last lap (lighting and then hitting you with every item launched with PERFECT timing to avoid the shock or 4 blue shells in a row). Though one neat thing i noticed, in Mario Kart Tour, the Computers do not get 3 items, but rather just one item regardless of character...and position. Seems they randomly get items anywhere but only one at a time. If they do get a frenzy it's like getting the Star item in some missions. Just something I observed when trying to see if I could manipulate getting item boxes multiple times for points and deliberately played poorly.
Since child I've always noticed the rivals and different behavior upon being hit.Toad's Turnpike is impossible to win if you let a rival get offscreen on first, you simply can't catch up no matter how flawless you play. They have a much higher max speed when ahead and offscreen. Your only chance to win is pray them to spin into bananas a couple of times.
I've always loved the Star Cup in this game, because every track has areas where you can mess with the CPU's behavior. Sometimes, on Royal Raceway, they'll get stuck in an endless loop of falling into the water over and over. Lakitu places them back before the curve where they fell in, and they fall right in again.
I would always try trapping the main rival on one of the forced tumble slopes as soon as I could so I had every other racer trying to get ahead of me all at once lol. One mistake and they all vanish into the wind
Except for the rival system, I don't think most of these are intentional. The behaviour of the non rival computers players when off screen is mainly just because the game doesn't fully simulate those characters to save CPU power. What you see when you use play with off screen camera is just an accident of what happens when the code that runs the simplified off screen physics interact with animation code that's not been designed to be played against the simplified physics. The off screen physics seems to all just be a variation of adding a time penalty to the character when they got hit by those events. The devs likely calculated/estimated that when a computer player got hit by a certain item, this is the amount of time it would need to recover compared to if they didn't get hit. So rather than simulating the item effect, they just applied a time/speed penalty instead.
I freakin' knew it! I always suspected the computer characters cheated somehow! I noticed the rival system, to an extent. The computer's inexplicable ability to just overtake me when I was in 1st place seemed so artificial! My brothers said I was crazy. They said I was just being a sore loser! The vindication is PALPABLE!
I thought it was pretty obvious they cheat. Even in MK Wii, they have a higher top speed than the player on the hardest difficulty, and certain character just always seemed to compete better based off what you play. I'm pretty sure ai also cheats in Mario party as well, beyond their inhuman button mashing abilities (the 1v3 claw game in Mario party 2)
@@GlitchLamb lol yeah, if that game came up, my bro and I didn't even bother playing if the cpu was on hard or very hard, we'd rather spare our controllers. Its funny tho watching a person try the absolute hardest trying to escape DKs grasp from the claw machine, getting winded and failing, especially when they go for the plus 30 timer when ur the last one standing XD I love that game but my family sold our Nintendo 64, we don't have that anymore.
@@halinaqi2194 I have always thought that Peach is the CPU who presses the A button the fastest, at least in Mario Party 2, in the Claw mini game it is impossible to escape from her once she grabs you, another thing, when you pay Boo 5 coins to steal Peach's coins, he can only steal 5 coins from her xD
@GlitchLamb do *NOT* Get me started about losing coin in the claw game in Mario party. That son of bit h Luigi cost me the game on 2 turns before the end I lost 110 coins! I'll never forget it
Here's a slight technical explanation for the CPU skip on Royal Raceway: This course and a few others have special code that overrides the default Lakitu placement behavior. There is a second set of path point values (memory address: 0x80165320) , that only updates under certain conditions and that the game will choose to place back on the road when Lakitu rescues the racer. That second set also only updates when the CPUs are considered on-screen. Here are the special cases on which the "Lakitu placement" will update for each course: -Yoshi Valley: Updates always unless the racer is over 600 units above ground or it detects an out of bound area / tumbling wall / void plane straight below. -Frappe Snowlad: Updates always unless the racer drives on the off-road snow. -Royal Raceway: Updates always unless the racer is in the air through any means or touching the large boost pads. -Rainbow Road: Updates always unless the racer is over 600 units above ground or it detects the void plane straight below. They never update when the CPU racers are considered off-screen, which means if you switch that property (via having the player camera come close enough) and time it correctly, you can observe the invisible Lakitu pick them up and place them where they've last updated their "Lakitu path point" That's also why the Frappe Snowland skip works. You can basically enter the off-road snow and drive half the lap, then touch an out of bound surface and Lakitu will place you all the way back, where you've last been on the road.
I also always hated that their speed seemed to stay up even when they are shrunk from Lightning. Watching those icons catch up to you swiftly IMMEDIATELY after using lightning really is frustrating.
I know it maybe helps with challenge or competition… but I hate racing games with such rubber band AI where it literally doesn’t matter how fast or good you drive, because in the last 3 seconds you’ll randomly get hit by a blue shell or etc and passed by a random computer out of nowhere. MK is definitely the biggest offender of this
F Zero was awful too. No matter how far you think you're ahead, if you bump into something you WILL get passed every time, also the CPUs never blow up, and if you knock them off the road they'll just get back on
Indeed. This artificial difficulty forces the player to lose usually by not their skill level being not good enough and rng instead. Crash Team Racing for example on PS1 had inflated speed given to the racers, but if you boost enough and use shortcuts, you can in fact even with the slowest characters, have a decent gap between CPUS. Showcasing a level of skill and difference.
Take a look at Fatal Racing/Whiplash by Gremlin Interactive for DOS/Win 95. Despite how difficult the AI could be there (up to 16 racers), it is possible to lap 2nd place twice in a race, while simultaneously being very difficult to catch up to a slower car ahead just on the course's geometry. Off the top of my head, I want to say this was a 1996 release.
It's just bad coding. Make the game too easy, and it's boring, but what if you can't make the game challenging enough, i.e. you can't make game mechanics that makes it competitive? Well, in this case, they just cheated. That also ruins the game.
i laughed so hard at the part with Wario literally smashing into Yoshi, i can just image Wario going "WAAAAAAAAAAAAAAAAAAAHHHHHHHHHHH" in the distance as he literally flies in lightspeed towards Yoshi.
>I have always wondered how that Royal Raceway shortcut worked. It always made me curious. I've learned a ton about the weird off screen mechanics from this video. >I've also noticed on certain tracks about Marty's getting hugely in front and winning races when they otherwise shouldn't, and wondered how that worked. It especially happens in Kalamari Desert, if the rivals get hit by the train, while a random Marty gets through. As long as they can avoid hazards through the rest of the race, a Marty can become unstoppable on that track. >I've seen the reverse to Royal Raceway happen to the loop in Rainbow Road - where instead of a shortcut, it screws the off screen character over. Usually a rival will fall through the track in the loop around the star, and get stuck on another part of the track, doomed to come last. >Something I try to do is to take out a rival in Wario Stadium or Royal Raceway's jumps, and watch them fly up the rankings - or do it on the last lap and doom the rival to one of the last place positions, giving me a huge advantage for the rest of the cup.
On any track, if a rival or marty enters an off road area and go off screen while under the speed penalty, they'll keep the speed penalty until they're on screen again. That's likely what's happening on KD.
8:36 wario going through a wall sounds like something wario would actually do. as he would probably yell, “WARIO! IM GONA WIN!” As he rams yoshi of the track.
The only thing I didn't really talk about was what items the cpus can or can't use, as they will never use red or green shells. Aside from that there hasn't been much research into AI behavior in MK64 so this video was the first concise attempt at observing what they do and how they work. There is a long video about the AI by midiguyDP linked in the description if you want a super in depth dive
There's no reason this system can't be toned down significantly and make the game more balanced, but still challenging. Also commenting here because if someday this becomes a thing, and someone sees this comment, please let me know! :)
I'd like to see how the system performs in this scenario. I get a feeling a lot of this is done for perfomance reasons, and if you make it completely simulate all 7 racers, it probably lags on original hardware or an unboosted emulator.
If you've seen the work going on with the Mario 64 optimization, a little work like that on Mario Kart and it would be a masterpiece, able to handle all kinds of improvements in standard hardware. Granted... Mario Kart is not nearly as interesting.. but nonetheless, it would be awesome.
A few Rom Hacks (Amped Up and Hooting Time atleast) allow such options including what weapons they can use, so they can finally use shells as they don't in vanilla Mario Kart 64. If you can't find a video of the bots without rubberbanding I'll record footage of it later today.
This is surprisingly clever, honestly. Having a couple of “skilled” CPUs means you’ll consistently be somewhere in the top 3, but vying for first place with a couple of opponents who are slightly better than you. It’s not just imitating competing with other players, it’s constructing a narrative that’s going to be satisfying for the player. It’s almost got a parasocial quality to it. I didn’t know this kind of thinking was even being applied to game design as far back as the N64.
Yeah it's actually pretty brilliant, since it would've been easy to just have a bunch of CPUs behaving under the same rules as the player, but the rival system makes it more exciting while being subtle enough that the player doesn't feel cheated.
@@coltonwilkie241 they rubber band way too hard and if they go off screen ahead of you, they will pull away. I personally do feel cheated on stuff like Banshee Boardwalk, where if you fall off anytime after the first lap, it's almost impossible to catch up.
Sometimes, in 150cc mode, a CPU racer in the lead will start pulling away, and you can't catch up to them without shortcuts. One can assume that said CPU racer has been put in "catch-up" mode when it shouldn't have been. I'm not sure what causes this. As for CPUs not obeying the usual rules and physics of the game, I noticed this even as a kid in the 90s. I always knew that CPUs had a "catch-up mode" where they would be faster than the player character, but only when trailing (unless the above glitch happens). I also caught CPUs using items when they hadn't driven through an item box (most noticeable on Rainbow Road) and yes, I also observed that CPUs recover from crashes faster when they are behind your field of view. While I never thought about it at the time, it couldn't be more obvious today that the reason for the CPUs "cheating" was ultimately due to hardware limitations. Later Mario Kart games have CPU racers that behave much more like the player-controlled ones. Though even in these more modern Karts, I bet there's still some things going on behind the curtains in order to keep the game running smoothly, especially in Mario Kart Wii where Grand Prix now has TWELVE racers.
I might look at the CPUs in other Mario Karts, this one was easy to do since I'm so familiar and didn't need much help, but I'd be happy to learn about others
In the new Mario kart8 deluxe, it’s pretty balanced overall, but I did notice a little bit of cheating with the CPU/ AI. I remember seeing a CPU miss an item block completely and still managed to block my red shell with a green shell, and especially when it comes to turning in 200cc! I swear no matter what vehicle they have they all turn at god like level barely hitting the curb when you know that even some of the best kart players can drift off stage with ease!
Them not needing item boxes isn't even really hidden is the wild part. Like yeah they grab them, but the CPUs never actually enter the little turnoffs in Toad's Turnpike which are the only places the item boxes spawn. Though the CPU can't use every item so it somewhat balances out.
Another stage that had something off with the AI was kalimari desert. It was the only track that I was able to get a huge lead in without a shortcut. Wasnt 100% of the time but it happened often where I'd be able to get far ahead enough to lap the cpus
I didn't mention this, but if an AI goes into an offroad area and gets a sleed penalty then goes off screen, the penalty persists until they're on screen again, which happens on KD a lot
You're likely getting the main rival CPU off screen when they're on the train tracks. They're forced to move slower than normal until they get back on screen and will drag everyone else down with them
In high school I would purposely mess with the rivals all the time and could get them to glitch out on certain tracks. Love this video! Brought back a lot of memories.
I think this video explains why mirror Toad's Turnpike was incredibly difficult to first place even when driving perfectly. Something about the AI going out of sight made it extremely difficult to keep up. With the whole thing about them ignoring collision when out of sight, it makes so much sense.
I didn't know about them skipping or being invulnerable to items sometimes but I knew they would speed up crazy. My nephew and me played the whole GP's in 2 player mode and we noticed how these AI drivers were glued to us. Not even Rubber banding but literally glued. Also I miss the old item boxes which were truly random and not distance and position based
Yeah, that Royal Raceway shortcut happened to me one time years ago. I always used the placement UI so I noticed it immediately. Bowser just skipped half a lap and was so far away that I couldn't catch him anymore. It's nice to finally have confirmation that I didn't imagined it that day.
Long time ago, I was playing with my nephew and tried to do the ramp jump on Royal Raceway to land on the left side of the track, I slammed into the slope and fell underwater, but somehow, Lakitu picked me up and placed me on the top of the left hill. We were so surprised by it that we had no reaction, and then later, Lakitu placed me on the finish line. I never did that again, Lakitu always placed me on the track near the finish line curve. Seeing your video made me remember this.
I have added this video to my "Exemplary Video Essays/Review Videos" playlist because based on my experience with TH-cam for the past Decade, this one checks many if the boxes for me to consider a video of top quality. Easy to follow script, excellent visuals, the visuals matching the source topic (literal Mario Kart 64 graphics), Clear audio, no audio layer conflicts with hiw listenable any other layer is, and there are bonus points for putting such an amazing effort to cover a small/niche bit of a greater whole, in this case, just a piece of whqt Mario Kart 64 is. Awesome work.
Thanks for this comprehensive breakdown! My personal 1 player GP challenge for MK64 is to come in 1st while leaving the largest possible gap between 1st and 2nd total points at the end, which means identifying your rivals during the first race and ensuring neither places 2nd more than once. This is incredibly difficult but very rewarding.
This is why I went crazy with a GameShark as a kid with unlimited lightning, constantly spamming it and also spamming spiny shells, sometimes I would hold the spiny shell and they would fall onto it eternally screaming as they would fall on it over and over again by positioning myself correctly. :^) this game made me get a sadistic glee seeing cheaters suffer. 😂🤣
Even as a kid in the 90s, I always knew that the game didn't play fair when it came to the CPU characters. I just wasn't aware of what they were capable of doing or what the triggers were. This was legitimately interesting and I'll definitely be more perceptive, regarding the CPUs, the next time I play MK64. Thanks for a great watch!
It's actually wild the amount of thought that went into giving the player a *fun* and *engaging* experience instead of a *balanced* one This really demonstrates how different old Mario kart was to the modern titles
There was something I have observed on Kalimari Desert, which I'm pretty sure is a glitch. It also only happened on 150ccm, I think. Once the leading rival gets too far from you, it becomes impossible to catch up, as he always seems faster and no matter how often you hit him with shells, he wouldn't slow down. But if, by some miracle, you do catch up, then he would behave normal again.
Oh yeah, I also experienced that on that level. Well, without the miracle of ever catching up. I think the detection messes up or something and then puts the CPU in full speed. And that's way faster than a player can go.
Nintendo fanboys: "Mario Kart 64 and F-Zero X on high difficulties aren't hard, you just suck" The rubber-banding mfs in these games: (awesome video btw, thanks as always!)
I'm also curious about the programming of a red shell. I notice that if you're somewhat in the lead, a red shell will be 'stupid' and hit walls all the time, but if you were closer to last, it seemed to be smarter and hit targets more accurately.
Red and Blue Shells have an approximate path when further from CPU's. You'll see this usually if you throw the shells at a slight angle, in which case after a second they either direct at a target or re-align itself in the pathway of a track. Since it does take a second for the shells to do this, and since they aren't guard protected like green shells, red and blue shells thrown by humans near walls often can't align properly in time and collide with the walls which nullifies and destroys them once made contact. There are exceptions to this though, such as if their is a hairpin turn and a CPU is on the other side of the wall, the shell's target priority will take place and try to go through the wall to hit the CPU to no avail. Also I think Nintendo knew how bad this programming was because every game onward always gives red and blue shells a proper and more efficient tracking when targeting CPU's even from a mile away.
I played this game today after maybe 20 years or something, and it struck me really hard how much the rivals cheat even on 50cc. They are on your tail at all times, it's crazy. Makes mk8 seem fair and balanced lol
MK is always rigged, the question is just how much If it’s so bad that you get punished for being a good racer and keeping first, then it’s not a good racing game
Thanks for the video! This is a topic that always intrigued me since I was a kid, so having a more detailed explanation on how/why this happens is great.
Geeze, I knew the cheating was bad, but I didn't realize HOW bad it was! No wonder I remember this game as being frustrating rather than fun... Got a question for you: Say I want to make a speedrun of a game that no one has ever run before - has even tried to find glitches, breaks, RNG manips, or the like before. What sort of advice do you have for such a scenario?
For the speedrunning thing, I'd say play the game. A lot. Experiment with all sorts of different builds in the game, and try to come up with unusual combinations or maneuvers that the game isn't strongly encouraging to do. Making multiple things happen at the same time is common in breaking games, so try to go for stuff like that. Button mash occasionally, try to slam into walls, you know, stuff that game testers do. For RNG manips, learn to juggle a lot of Save States (assuming you're playing on emulator) before a given RNG event might occur, and then load it and go about it very slightly differently, also applying some of the above methods like button mashing (to see if useless inputs adjust the RNG, because they often do, especially in older games). I'm not huge into speedrunning but I do like to try and break games for fun. Hope you have some fun with it, too.
It depends on their weight class and your speed class. If you race at 50cc, any CPU can become a rival. At 100cc, medium and heavyweight CPUs can be rivals. And on 150cc/Mirror, only heavyweights are rivals.
This is very smart programming thats makes the gane feel more organic and fun. Games which lack this and onky use speed rubber banding just feeling dull and/or frustrating. Especially kart racers where the player can abuse boosting and items. Games where items have full effect offscreen become extremely easy to cheese and you can easily lap NPCs multiple times in a race.
This is so cool... I'm 37 and a DevOps engineer. However, at the age I was when playing this game I didn't really know anything about programming or anything tech really. I've never thought much about this stuff, and watching this video it seems so obvious! But it's really cool to learn all this and actually think about all the experiences I had as a kid. Even now... since I'm not a "programmer" or "developer" per se ( i do write code... just not in their sense) I don't really think about games in this sense, but watching this kind of makes me want to think a little bit more about the programming challenges that go into this or decisions for why things work the way they do.
I grew up playing mario kart for snes and this was extreme in that game, they even had AI exclusive abilities/item effects they could just toggle whenever, Peach's cupcake, bowser's fireball, Mario and Luigi has the star effect ability, etc.
Another interesting fact about CPUs and their behavior with bananas: when throwing a banana forward, it will most likely land in their own path (even throwing them at an angle to do so) resulting in them driving right into them. It's rather goofy. And their general item pool is also restricted. They will never get any shells no matter what. they won't get any multi items either, like the banana bunch or golden mushrooms. If I remember correctly, they also don't get any lightning bolts. pretty sure that's because of hardware limitations And on another topic, let's talk red shells: Did you know/notice that red shells will only follow the road to get to the next racer when used on 5th place or lower? When on 4th or higher it will go straight for the next racer regardless of any obstructions, just hit the wall and be destroyed.
I encountered CPU cheats like this on the original Snes version, it meant you had to stay in second place the entire race and only pull ahead on the final lap if you wanted to win, the same technique also worked with the PS1 version of Grand Turismo.
Hmm, I think it'd be fun to test that theory about performance by limiting physics calculations/interactions. Surely there's some sort of flag that CPU racers have that tells them they're being observed, is it possible to trip this flag for all of them and see if it affects performance at all?
You can trip it while they're on camera with a game shark code which is how I captured the off screen effects. But enabling them to always behave as if on screen is a bit different, and I'm not sure the code exists.
@@Abyssoft I don't remember the game shark code, but I do remember playing with a code with the real console that was cpu cheating turn off, and it is a really anyoing mess of screaming in Toad turnpike. They where off the player screen and still get hit by the trucks and car. with that game shark code on.
This Royal Raceway skip actually happened to a CPU when I was playing, young me wasn't very happy about it lol It was always a mystery to me how and why did this happen, so thank you for clarifying this ^^
4:05 Mario kart, the only racing game that you press left and it moves you right. Also Mario kart, the only racing game that cheats with EVERY way available.
You might as well play Time Trials then, since you won't really see much of the CPU racers once you're front running. Yes, there is a gameshark code for this! CPU Drivers Process Offscreen 81028BAA B49A
Ignored this one for a few days assuming it was a video about rubber banding (as most videos about CPU cheating in racing games are) and left insanely informed on new jank that i had no idea about. Great video as usual man!
This is why MK64 is easily my least favorite entry in the series. That damn clipping and rubber banding was so frustrating. It just made the game feel like no matter how well I did, they would always cheat me out of my hard-earned victory.
Mario Kart really didn't get good until Super Circuit. They cheat in all games with rubber bands, but in Super MK and 64, it's barely even a racing game, it's just "you, 2 super cracked hackers and 5 pointless non-entities that make your life hell through random chaos".
This video was amazing, I always noticed and hated how the cpu would be immune to red shells and would only slow them down for a bit instead of doing the whole fling animation
Im a big Halo Wars fan, and this video reminds me of the legendary difficulty Computer A.I. and how they cheat as well. They make supplies quicker, and their units are slightly stronger than the player's units. As in: A marine squad from a Legendary AI will always beat a marine squad from a player in a 1v1. Same unit, but theirs is slightly stronger in a one-on-one scenario. That goes for EVERY unit in the game. This damage disparity is very noticeable when your Scarab is fighting their Scarab. It can be annoying, but it makes it more rewarding in the end when you beat them.
Do you have the bell clicked as well? Sometimes the algo will wait a bit to notify subscribers after a video has been uploaded, but if you have the bell activated it should be instant.
If Im not wrong you can change your rival in mid GP since the rival always need to be the 1st, 2nd or 3rd. So if you somehow manage to buly your Rival while making the 4th place steal the 1st place, he will become the new rival. while the new 4th place will act like a normal npc (all in point system)
Mario Kart 64 is not really pushing the hardware. It's a case of an early N64 game frankly not being very well made. Similarly, there's lots of bad, slow code in Super Mario 64 (now famously described by Kaze) or bad rendering decisions in Turok.
This system is actually genius for those who want a challenge, I just wish they implemented this exact system or made it it's own difficulty in later mario kart games. Having the CPUs cause more trouble and be chaotic brings about all the fun in Mario.
i didn’t know this game had a hidden rival system, I thought i was going crazy when I saw the exact same two racers targeting me through the entire cup
I actually knew most of this from just playing hours of this game as a kid. I first realized when the CPU actually stopped for banana peel hits but seemed to only BRIEFLY stop if hit with a collision item. On Toad's turnpike they were ONLY hit by cars if I literally SAW it or if they where VERY close behind with the collision occured. I can go on all day... Glad someone finally documented this though 😌
I noticed this problem on the SNES version of Mario Kart -- whichever order the CPUs spawned in determined what rank they thought they should be in. On high difficulties like 150cc this was way more apparent. If a CPU got significantly behind their expected position, they would rocket forward to catch up. When I discovered this I took it to the extreme for fun: After my friend slowed down the expected-to-be-in-first-place character (I think it was Princess Peach) with a couple attacks (I think red shell + ran them over when they were small), I stopped at the powerup area and hit her I think 3-4 more times (multiple red shells and a green one). At this point Princess Peach was almost an entire lap behind on one of the beach levels. I just sat there watching the mini-map and laughing my ass off as Peach went around the map like an airplane. It was ridiculous how fast she was going, around 10x the max player speed! Now, as far as racing games where the computers don't cheat, Rock N' Roll Racing for the SNES is my all-time favourite! On Warrior difficulty those computers are so tough to beat, and they can hit every obstacle, fall off every cliff, and can be blown up the same as a player. The only cheats they employ are an unlimited supply of lap-powerups (weapons, boosters, etc.). But I always figured that was just because they were richer and could afford better vehicle upgrades, lol. I never really minded the CPUs on RnR having endless nitros and missiles, they didn't usually spam them.
I always wondered why I seem to get harassed by two specific CPUs during races, and knew the CPUs cheated in some way shape or form (namely the rubber band effect) but this thorough breakdown about their behavior is fantastic!
Honorable mention to the original F-Zero for also having ridiculous rubber-banding There's TASes of the SNES version(s) that're split into real time-focused and IGT-focused, the sole difference in practice being that the real time TASes use some tricks to slow down just enough at the finish line to let one of the rival cars pass em by like one frame - getting 1st place proper burns time as the finish animation is much longer than if you get 2nd or 3rd The fact though that the competition catches back up in jUST A FEW SECONDS despite having a lower top speed stat than the TAS's typical choice, the Fire Stingray (or I think the Luna Bomber in the case of the SatellaView ROMs), and no manual boosts like the player does is insane
@@Abyssoft CatadorDeLatas on SMWCentral has a good thread explaining F-Zero's AI along with a patch that makes the AI control your vehicle if you're looking for somewhere to start.
I knew it! I've just started playing this game and like I can hit these guys all day and they will catch right back up. I can barely stay in the lead and if I try 150cc I've have no chance to win 😢
Thats even more terrifying when you realize theyve only gotten deadlier with newer installments i swear ive seen them calculate the trajectory of a green shell bounce off a wall and hit mee LIKE HOW !!!!
We all knew it. We just didn't know what or how. The same applies to many other racing games. This prevents the player from winning too easily or losing too hard. SOME games have a setting to disable certain things like this. For example. I believe in some NFS games, there is a setting to disable AI Slowdown when they are ahead. This is meant to make the game more challenging.
Content at the tail end of the year is a tough gap for my sub feed. Much appreciation for taking the time out during the end of the year! Have a good one.
I was playing Mario kart 64 on the Switch VC two weeks ago and the entire time no matter the lead, no matter how well I did and amount of drift boosts I did I always had NPCs passing me non-stop on all cups, 50cc - mirror. I was like, wtf is going on here. It's impossible to keep any lead! And now I know rivals exist and they are out for blood. Thanks for sharing, that was really cool to learn about. lol.
When I was like 10 I noticed a lot of these things because you can turn on that track outline on the screen border and they stick out a lot. They dont bounce to indicate they were hit by red shells. They will, however, always spin out when hitting bananas. I never was bothered by this since the player has so many advantages over them. They can't ever use shells or take advantage of any shortcut. Bananas also force them to a complete stop and spin unlike you being able to recover super quick if you use triple trap acceleration. You don't even need a hacked camera to see a lot of this. If you just stay far, but can still see them, you can see them not do the tumble. The game also relocates shells to specific places on the track once one goes off screen. Easiest one to see it is Wario Stadium because its in the dip underneath the big jump
![[MV] A leap of faith...ความเชื่อใจครั้งสุดท้าย (From "Petrichor the series")](http://i.ytimg.com/vi/U1piZH2CNXA/mqdefault.jpg)
Become a member to get access to these awesome emotes along with extra features, or if you want to see videos before they're uploaded become a Patron where you also get updates on what I'm working on: www.patreon.com/abyssoft
It's just badly coded.
no offense but in what context would i use those in
@@Mai.Calico I stream on this channel several times a month, and premiere every video, so there's lots of opportunity to use them in the community aside from comments on videos.
Make a Subscribestar instead, at least Subscribestar doesnt shut down their artists for drawing charsacters. 💀
Did you say Super Saiyan 3 lol makes sense they seem to have no limits in this mode lol
If you observe a CPU it acts like a wave, but if you don't, it acts like a particle
Imagine running the Double Slit Experiment on a bunch of CPU riders ^^
i can't believe this.
it's the other way around
I hate this. What is never explicitly stated is that in order to observe something, you have to EFFECT it. To detect a particle you have to interact with it which of course changes it state. I hate that in physics things are made more confusing than they need to be constantly.
@@raafmaat I've already gone though this garbage when I took QM when I was studying to become an electrical engineer. I'm trained as a VLSI designer actually.
The fact you're disputing what I said shows you don't understand QM. So tired of the stupid mysteries around all this crap. "Oh if you observe it, it changes is". This concept is CONSTANTLY promoted by the stupid fraud Deepak Chopra. I'm not going to argue something I understand, and I'm not waste time talking with somebody that does, but is fully convinced they do.
Playing Mario Kart 64 a lot growing up, I absolutely noticed the rival system even if I couldn't ever put it into exact terms. I could tell that 2 of the CPUs always were better than the others, and 1 of the 2 was also just a little better than the other.
Cool to see the exact mechanics of it laid out and finally justify what I've been saying for years.
I found it in Episode 1 Pod Racer as well.... and then realised I could use the debug cheat to slow the rival down. Always funny to watch a pack of races crawling along behind a rival because their AI won't let them overtake at all in that game XD
@@jasam01
Was that on the n64 version?
I've played both it and the Xbox port, and the Xbox one felt a LOT easier.
The rival system is a staple of the series, but its mechanics vary depending on the game. My four years old self also noticed its existence in Super Mario Kart. Unlike in MK64, rivals aren't random and it's not just two but all seven of them. The game is rigged so that CPU drivers usually finish the race in the same order. One is programmed to be in first place, one goes for second place, one for third place, one for fourth place, and three for last place. If CPU drivers are not in the correct order, the game adjusts their rubber band speed in an attempt to fix it.
The rival order depends on who you're playing as. For Mario, the order is Donkey Kong Jr., Peach, Yoshi, Luigi, Toad, Bowser, and Koopa Troopa. The last three have the same speed and are programmed to stay in fifth place or worse. They don't compete against one another, so how they place in relation to each other depends entirely on how much abuse they receive from the player. They can be really close together or really far apart. Since no points are awarded to drivers who finish in fifth to eighth place, it makes no difference.
A good strategy is to prioritize attacking the fastest rival, especially during the final lap, in order to skew their placement in the rankings. The second fastest rival will never aim for first place. It will yield to the fastest rival in all races. Once you successfully make the fastest rival finish after you in third place or worse, it makes winning the grand prix much easier.
I also noticed it. Not the specifics of course, but it wasn't too hard to figure out that there were always 1-2 standout racers. Still, this video explains a lot of the mysterious cheating I've encountered in the game such as a 1st place CPU finishing when I barely got started on the final lap in 2nd place. I've also had it happen that a CPU just straight up kept driving at full speed for an entire race without being able to catch up on it (probably a glitch or something).
Me and my sisters always use to hate the baby characters and Mario on the Wii versions because they always seemed to outperform the others
Guess it’s just cause we all picked characters who had them as rivals
It's always fun to turn on the character location border, take a shortcut, and watch your 2 rivals move at extreme speed to catch up to you.
My favorite was Wario Stadium, making all the turns as close to the inside as I could while constantly drift boosting and watching the CPUs try to keep up since they don't ever try to not make a wide turn. I end up getting further and further away and it's really funny
@@LunarBloodRose27 its also hilarious when they cheat hard when you do the wall hop >_< somehow it stirs their coding when you are suddenly 3/4ths across the track repetitively.
It knew it wasn't just because I sucked! Now do a video on every other game I suck at. I'll wait.
You're good at every game so I can't.
@@Abyssoft Would be curious to see how the AI cheats in Sonic Drift 2 on the Game Gear, since I feel like its rubber banding is even worse than Mario Kart 64. (especially if Metal Sonic is in 2nd, he'll just dash into first whenever he feels like it.)
@@Abyssoftunexpectedly wholesome
@@Abyssoft In the Advance Wars games for the GBA, the CPU would cheat in Fog of War, knowing exactly where your units were & could even fire upon them when they normally wouldn't be visible to another player. Makes me wonder if the Advance Wars 1+2 for the switch does the same thing or did they finally make it fair like in Days of Ruin.
I sure felt better about myself when I saw how street fighter 2 cheats 😂
Wario coming in and clotheslining Yoshi through the wall gave me such a good laugh, I've been terribly sick the past few days so I really needed it. I appreciate it a lot, thanks for another great upload!
Same, that genuinely made me cackle, I wasn't expecting it LMAO
Wario jumpscare
WAAAAAAAAAAAAAAAAAA
Lol! 😂😂😂😂😂
Add in Wario's maniacal laugh unique to MK64 and it's perfect!
Fun fact: You can trap a CPU player into the train tracks on Kalimari in GP mode and it will remain "jailed" as long as at least one human player can visibly see the player on screen. The moment you drive away and place the CPU out of view, it can glitch itself outside the "jail."
When I finally acquired a GameShark in the early 2010s, I combined approximately a dozen codes together to form what some in the community call the "Wonn challenge." This allows human player(s) to race against up to 3 CPUs in 4P Vs. mode where they can maintain draft mode at will and even turn on a dime in mid-air off of the big jumps, like in Royal Raceway. In this mode you can see all kinds of other interesting effects as well:
Bolts will happen out of nowhere, sometimes zapping all 4 players. You might even get bolted thrice in a row! The first item box they pick up never gets used, so it's ripe for stealing with a ghost. But they'll drop bananas, fake item blocks, and activate ghosts or stars seemingly at will. Also when a CPU runs into a banana, they slide forward (since they're now always visible) and they'll additionally crash upward at the end of the slide as though they ran into a fake item. This is much like the effect you see with the last place human player at the end of an all-human Vs. race if that player didn't have a Star activated.
If you're too aggressive with bananas near the end of Choco Mountain, you might get 2 or more CPUs stuck in such a way that they never get back on track, effectively soft locking the game, so this could result in failing the "Wonn challenge" depending on your rule set. Had the CPUs not been visible, they'd find a way back on track...
Wow, a name I haven't seen in forever. I remember having a lot of fun visiting your old website back when I was a kid!
I can only imagine the CPU players chatting like:
-C'mon dude, lets get out of here and cheat our way into 1st place.
-Shhh! The player is looking!
It's pretty difficult for me to avoid a SimpleFlips joke here. It's almost like a jail that I can't escape. It's quite infantile of me.
@DavidWonn: The player can only see his own item box, so we can't see the item box the CPU have on the road, but the same thing since we don't see the cpu item box, we wont know if we accidently broken one of his , one time the cpu behind me used no item because I must have run into one of the cpu item box that the player can't see, the same think with the cpu they don't see your item box on the ground but can still break them so they would not do on purpose to take your box because they actually do not see it. Notice in toad turnpike the player item box are on the side, and the cpu never goes their unless push into that area, how they do get item if the never go to that area, their own item box is place where they follow the path this is how I know that I accidently pass an A.I item box because the one dirrectly behind me did not toss anything.
Fun fact: tRump lost the election
I thought they had just, increased speed.
Didn't expect them to have a multitude of techniques at their disposal.
I recently played MK64 for the first time, and while I knew CPUs cheated, witnessing it first hand was astounding. It's insane. I also did notice a lot of the BS the CPUs do such as them recovering fast from items when I couldn't see them, but didn't know all the intricacies.
The Japanese game Devs even back them were doing a brilliant job.
@RaniaIsAwesomeAlmost makes me wonder how CPU behavior and handicapping changed in later games. I KNOW Mario Kart Wii has some shenanigans with those infamous "item chains" they get on you at the end of the last lap (lighting and then hitting you with every item launched with PERFECT timing to avoid the shock or 4 blue shells in a row). Though one neat thing i noticed, in Mario Kart Tour, the Computers do not get 3 items, but rather just one item regardless of character...and position. Seems they randomly get items anywhere but only one at a time. If they do get a frenzy it's like getting the Star item in some missions. Just something I observed when trying to see if I could manipulate getting item boxes multiple times for points and deliberately played poorly.
Since child I've always noticed the rivals and different behavior upon being hit.Toad's Turnpike is impossible to win if you let a rival get offscreen on first, you simply can't catch up no matter how flawless you play. They have a much higher max speed when ahead and offscreen. Your only chance to win is pray them to spin into bananas a couple of times.
It's worse if one of non rival CPUs are in first somehow. They slow down for absolutely nothing
Yeah. Real
Yeah when they are off-screen ,they can through the vehicles unscatted while you are forced to lose time dodging them .
Blue shells don't do anything either. I mean... the effect of a redshell? And a nerfed redshell? That's just stupid lol
@@icycloud6823 at least they aren't running straight into walls lol
I've always loved the Star Cup in this game, because every track has areas where you can mess with the CPU's behavior. Sometimes, on Royal Raceway, they'll get stuck in an endless loop of falling into the water over and over. Lakitu places them back before the curve where they fell in, and they fall right in again.
That's probably the most broken track when it comes to Lakitu placements
I would always try trapping the main rival on one of the forced tumble slopes as soon as I could so I had every other racer trying to get ahead of me all at once lol. One mistake and they all vanish into the wind
Homosexual Infant Incarceration Facility
What @@arciks11
@@arciks11Shoutouts to Simpleflips
Always knew the computer cheated, never knew it went this deep though. Great video!
Except for the rival system, I don't think most of these are intentional.
The behaviour of the non rival computers players when off screen is mainly just because the game doesn't fully simulate those characters to save CPU power. What you see when you use play with off screen camera is just an accident of what happens when the code that runs the simplified off screen physics interact with animation code that's not been designed to be played against the simplified physics.
The off screen physics seems to all just be a variation of adding a time penalty to the character when they got hit by those events. The devs likely calculated/estimated that when a computer player got hit by a certain item, this is the amount of time it would need to recover compared to if they didn't get hit. So rather than simulating the item effect, they just applied a time/speed penalty instead.
my ex cuda learned a few things and uped her game
i bet people took shortcuts so that's not cheating?
I freakin' knew it! I always suspected the computer characters cheated somehow! I noticed the rival system, to an extent. The computer's inexplicable ability to just overtake me when I was in 1st place seemed so artificial! My brothers said I was crazy. They said I was just being a sore loser! The vindication is PALPABLE!
I thought it was pretty obvious they cheat. Even in MK Wii, they have a higher top speed than the player on the hardest difficulty, and certain character just always seemed to compete better based off what you play.
I'm pretty sure ai also cheats in Mario party as well, beyond their inhuman button mashing abilities (the 1v3 claw game in Mario party 2)
@@halinaqi2194 Bro, you remind me of that claw minigame in MP2 💀
@@GlitchLamb lol yeah, if that game came up, my bro and I didn't even bother playing if the cpu was on hard or very hard, we'd rather spare our controllers.
Its funny tho watching a person try the absolute hardest trying to escape DKs grasp from the claw machine, getting winded and failing, especially when they go for the plus 30 timer when ur the last one standing XD
I love that game but my family sold our Nintendo 64, we don't have that anymore.
@@halinaqi2194 I have always thought that Peach is the CPU who presses the A button the fastest, at least in Mario Party 2, in the Claw mini game it is impossible to escape from her once she grabs you, another thing, when you pay Boo 5 coins to steal Peach's coins, he can only steal 5 coins from her xD
@GlitchLamb do *NOT* Get me started about losing coin in the claw game in Mario party. That son of bit h Luigi cost me the game on 2 turns before the end I lost 110 coins! I'll never forget it
Here's a slight technical explanation for the CPU skip on Royal Raceway:
This course and a few others have special code that overrides the default Lakitu placement behavior. There is a second set of path point values (memory address: 0x80165320) , that only updates under certain conditions and that the game will choose to place back on the road when Lakitu rescues the racer.
That second set also only updates when the CPUs are considered on-screen.
Here are the special cases on which the "Lakitu placement" will update for each course:
-Yoshi Valley:
Updates always unless the racer is over 600 units above ground or it detects an out of bound area / tumbling wall / void plane straight below.
-Frappe Snowlad:
Updates always unless the racer drives on the off-road snow.
-Royal Raceway:
Updates always unless the racer is in the air through any means or touching the large boost pads.
-Rainbow Road:
Updates always unless the racer is over 600 units above ground or it detects the void plane straight below.
They never update when the CPU racers are considered off-screen, which means if you switch that property (via having the player camera come close enough) and time it correctly, you can observe the invisible Lakitu pick them up and place them where they've last updated their "Lakitu path point"
That's also why the Frappe Snowland skip works. You can basically enter the off-road snow and drive half the lap, then touch an out of bound surface and Lakitu will place you all the way back, where you've last been on the road.
I also always hated that their speed seemed to stay up even when they are shrunk from Lightning. Watching those icons catch up to you swiftly IMMEDIATELY after using lightning really is frustrating.
then lets talk about that shortcut in wario stadium ;-) the cpu never uses that shortcut so that makes the player also a cheater
@@theelitegamer1980no.
more like we have to cheat because the entire cpu cheats and not just by shortcuts@@theelitegamer1980
I know it maybe helps with challenge or competition… but I hate racing games with such rubber band AI where it literally doesn’t matter how fast or good you drive, because in the last 3 seconds you’ll randomly get hit by a blue shell or etc and passed by a random computer out of nowhere.
MK is definitely the biggest offender of this
To be fair they didn't have many case studies to base how the AI should be tuned on back in 1996
F Zero was awful too. No matter how far you think you're ahead, if you bump into something you WILL get passed every time, also the CPUs never blow up, and if you knock them off the road they'll just get back on
Indeed. This artificial difficulty forces the player to lose usually by not their skill level being not good enough and rng instead.
Crash Team Racing for example on PS1 had inflated speed given to the racers, but if you boost enough and use shortcuts, you can in fact even with the slowest characters, have a decent gap between CPUS.
Showcasing a level of skill and difference.
Take a look at Fatal Racing/Whiplash by Gremlin Interactive for DOS/Win 95.
Despite how difficult the AI could be there (up to 16 racers), it is possible to lap 2nd place twice in a race, while simultaneously being very difficult to catch up to a slower car ahead just on the course's geometry.
Off the top of my head, I want to say this was a 1996 release.
It's just bad coding. Make the game too easy, and it's boring, but what if you can't make the game challenging enough, i.e. you can't make game mechanics that makes it competitive? Well, in this case, they just cheated. That also ruins the game.
i laughed so hard at the part with Wario literally smashing into Yoshi, i can just image Wario going "WAAAAAAAAAAAAAAAAAAAHHHHHHHHHHH" in the distance as he literally flies in lightspeed towards Yoshi.
Wario Dies After Colliding With Yoshi
@@aprinnyonbreak1290I'm pretty sure if a planet collided with a lizard, the lizard would die
>I have always wondered how that Royal Raceway shortcut worked. It always made me curious. I've learned a ton about the weird off screen mechanics from this video.
>I've also noticed on certain tracks about Marty's getting hugely in front and winning races when they otherwise shouldn't, and wondered how that worked. It especially happens in Kalamari Desert, if the rivals get hit by the train, while a random Marty gets through. As long as they can avoid hazards through the rest of the race, a Marty can become unstoppable on that track.
>I've seen the reverse to Royal Raceway happen to the loop in Rainbow Road - where instead of a shortcut, it screws the off screen character over. Usually a rival will fall through the track in the loop around the star, and get stuck on another part of the track, doomed to come last.
>Something I try to do is to take out a rival in Wario Stadium or Royal Raceway's jumps, and watch them fly up the rankings - or do it on the last lap and doom the rival to one of the last place positions, giving me a huge advantage for the rest of the cup.
On any track, if a rival or marty enters an off road area and go off screen while under the speed penalty, they'll keep the speed penalty until they're on screen again. That's likely what's happening on KD.
@@Abyssoftare the train tracks you cross considered off road?
8:36 wario going through a wall sounds like something wario would actually do. as he would probably yell, “WARIO! IM GONA WIN!” As he rams yoshi of the track.
i was prepared for a 30 min vid, this will do it
it was a nice explanation on what we suffered as kids
The only thing I didn't really talk about was what items the cpus can or can't use, as they will never use red or green shells.
Aside from that there hasn't been much research into AI behavior in MK64 so this video was the first concise attempt at observing what they do and how they work.
There is a long video about the AI by midiguyDP linked in the description if you want a super in depth dive
"We" what you mean "we"? Nobody cares and get lost already
Can somebody patch this out so they behave? Would be fun to see how the CPU's perform in that scenario.
That would be awesome to see. I imagine you would be able to lap them as you can do to the CPU in DKR
There's no reason this system can't be toned down significantly and make the game more balanced, but still challenging.
Also commenting here because if someday this becomes a thing, and someone sees this comment, please let me know! :)
I'd like to see how the system performs in this scenario. I get a feeling a lot of this is done for perfomance reasons, and if you make it completely simulate all 7 racers, it probably lags on original hardware or an unboosted emulator.
If you've seen the work going on with the Mario 64 optimization, a little work like that on Mario Kart and it would be a masterpiece, able to handle all kinds of improvements in standard hardware.
Granted... Mario Kart is not nearly as interesting.. but nonetheless, it would be awesome.
A few Rom Hacks (Amped Up and Hooting Time atleast) allow such options including what weapons they can use, so they can finally use shells as they don't in vanilla Mario Kart 64.
If you can't find a video of the bots without rubberbanding I'll record footage of it later today.
This is surprisingly clever, honestly. Having a couple of “skilled” CPUs means you’ll consistently be somewhere in the top 3, but vying for first place with a couple of opponents who are slightly better than you. It’s not just imitating competing with other players, it’s constructing a narrative that’s going to be satisfying for the player. It’s almost got a parasocial quality to it. I didn’t know this kind of thinking was even being applied to game design as far back as the N64.
Yeah it's actually pretty brilliant, since it would've been easy to just have a bunch of CPUs behaving under the same rules as the player, but the rival system makes it more exciting while being subtle enough that the player doesn't feel cheated.
@Blasco_Snap the player doesn't feel cheated if they ignore the blatant cheating AI.
@@coltonwilkie241 they rubber band way too hard and if they go off screen ahead of you, they will pull away. I personally do feel cheated on stuff like Banshee Boardwalk, where if you fall off anytime after the first lap, it's almost impossible to catch up.
Sometimes, in 150cc mode, a CPU racer in the lead will start pulling away, and you can't catch up to them without shortcuts. One can assume that said CPU racer has been put in "catch-up" mode when it shouldn't have been. I'm not sure what causes this.
As for CPUs not obeying the usual rules and physics of the game, I noticed this even as a kid in the 90s. I always knew that CPUs had a "catch-up mode" where they would be faster than the player character, but only when trailing (unless the above glitch happens). I also caught CPUs using items when they hadn't driven through an item box (most noticeable on Rainbow Road) and yes, I also observed that CPUs recover from crashes faster when they are behind your field of view.
While I never thought about it at the time, it couldn't be more obvious today that the reason for the CPUs "cheating" was ultimately due to hardware limitations. Later Mario Kart games have CPU racers that behave much more like the player-controlled ones. Though even in these more modern Karts, I bet there's still some things going on behind the curtains in order to keep the game running smoothly, especially in Mario Kart Wii where Grand Prix now has TWELVE racers.
I might look at the CPUs in other Mario Karts, this one was easy to do since I'm so familiar and didn't need much help, but I'd be happy to learn about others
In the new Mario kart8 deluxe, it’s pretty balanced overall, but I did notice a little bit of cheating with the CPU/ AI. I remember seeing a CPU miss an item block completely and still managed to block my red shell with a green shell, and especially when it comes to turning in 200cc! I swear no matter what vehicle they have they all turn at god like level barely hitting the curb when you know that even some of the best kart players can drift off stage with ease!
Them not needing item boxes isn't even really hidden is the wild part.
Like yeah they grab them, but the CPUs never actually enter the little turnoffs in Toad's Turnpike which are the only places the item boxes spawn. Though the CPU can't use every item so it somewhat balances out.
@nicholasharvey1232 Nobody cares and get lost
Another stage that had something off with the AI was kalimari desert. It was the only track that I was able to get a huge lead in without a shortcut. Wasnt 100% of the time but it happened often where I'd be able to get far ahead enough to lap the cpus
I didn't mention this, but if an AI goes into an offroad area and gets a sleed penalty then goes off screen, the penalty persists until they're on screen again, which happens on KD a lot
You're likely getting the main rival CPU off screen when they're on the train tracks. They're forced to move slower than normal until they get back on screen and will drag everyone else down with them
My mom never believed me when I said the game was cheating.
Your mom just was worried that you might choose a life of dishonesty.
the satisfaction of proving people wrong
In high school I would purposely mess with the rivals all the time and could get them to glitch out on certain tracks. Love this video! Brought back a lot of memories.
I think this video explains why mirror Toad's Turnpike was incredibly difficult to first place even when driving perfectly. Something about the AI going out of sight made it extremely difficult to keep up. With the whole thing about them ignoring collision when out of sight, it makes so much sense.
I didn't know about them skipping or being invulnerable to items sometimes but I knew they would speed up crazy. My nephew and me played the whole GP's in 2 player mode and we noticed how these AI drivers were glued to us. Not even Rubber banding but literally glued. Also I miss the old item boxes which were truly random and not distance and position based
That video of Yoshi getting bodied by Wario is top tier comedy.
Yeah, that Royal Raceway shortcut happened to me one time years ago. I always used the placement UI so I noticed it immediately. Bowser just skipped half a lap and was so far away that I couldn't catch him anymore.
It's nice to finally have confirmation that I didn't imagined it that day.
And here i was thinking i was crazy 🥴..
Long time ago, I was playing with my nephew and tried to do the ramp jump on Royal Raceway to land on the left side of the track, I slammed into the slope and fell underwater, but somehow, Lakitu picked me up and placed me on the top of the left hill. We were so surprised by it that we had no reaction, and then later, Lakitu placed me on the finish line. I never did that again, Lakitu always placed me on the track near the finish line curve. Seeing your video made me remember this.
This is good. Love that you’re making MK64 vids!
I love that I'm making mk64 vids as well, and that you're watching them
I have added this video to my "Exemplary Video Essays/Review Videos" playlist because based on my experience with TH-cam for the past Decade, this one checks many if the boxes for me to consider a video of top quality. Easy to follow script, excellent visuals, the visuals matching the source topic (literal Mario Kart 64 graphics), Clear audio, no audio layer conflicts with hiw listenable any other layer is, and there are bonus points for putting such an amazing effort to cover a small/niche bit of a greater whole, in this case, just a piece of whqt Mario Kart 64 is. Awesome work.
Thanks for this comprehensive breakdown! My personal 1 player GP challenge for MK64 is to come in 1st while leaving the largest possible gap between 1st and 2nd total points at the end, which means identifying your rivals during the first race and ensuring neither places 2nd more than once. This is incredibly difficult but very rewarding.
This is why I went crazy with a GameShark as a kid with unlimited lightning, constantly spamming it and also spamming spiny shells, sometimes I would hold the spiny shell and they would fall onto it eternally screaming as they would fall on it over and over again by positioning myself correctly.
:^) this game made me get a sadistic glee seeing cheaters suffer. 😂🤣
Gameshark codes helped a lot in the making of this video for sure
As a kid of what. Shut up and get lost
Don't know if you mentioned it but, the primary rival is pretty much impossible to catch up to if they get ahead in 1st.
I implied it, while it will go at max speed you can catch up by doing mini turbos
Always wondered how that CPU Yoshi was able to get that huge shortcut on Royal Raceway. Thanks taking a look into it!
Even as a kid in the 90s, I always knew that the game didn't play fair when it came to the CPU characters. I just wasn't aware of what they were capable of doing or what the triggers were. This was legitimately interesting and I'll definitely be more perceptive, regarding the CPUs, the next time I play MK64. Thanks for a great watch!
The CPU skipping to first place in Royal Raceway happened to me many times.
It's actually wild the amount of thought that went into giving the player a *fun* and *engaging* experience instead of a *balanced* one
This really demonstrates how different old Mario kart was to the modern titles
yeah, i noticed back when i used to play it that the cpu were faster when behind. It was nuts how good cpu could recover
There was something I have observed on Kalimari Desert, which I'm pretty sure is a glitch. It also only happened on 150ccm, I think.
Once the leading rival gets too far from you, it becomes impossible to catch up, as he always seems faster and no matter how often you hit him with shells, he wouldn't slow down. But if, by some miracle, you do catch up, then he would behave normal again.
I've experienced that many times as well.
Oh yeah, I also experienced that on that level. Well, without the miracle of ever catching up. I think the detection messes up or something and then puts the CPU in full speed. And that's way faster than a player can go.
4:54 thwomp enjoys his job a little too much.
Nintendo fanboys: "Mario Kart 64 and F-Zero X on high difficulties aren't hard, you just suck"
The rubber-banding mfs in these games:
(awesome video btw, thanks as always!)
I'm also curious about the programming of a red shell. I notice that if you're somewhat in the lead, a red shell will be 'stupid' and hit walls all the time, but if you were closer to last, it seemed to be smarter and hit targets more accurately.
Red shells will act stupid if your target is in 4th place or above. Why? No clue lol
Red and Blue Shells have an approximate path when further from CPU's. You'll see this usually if you throw the shells at a slight angle, in which case after a second they either direct at a target or re-align itself in the pathway of a track. Since it does take a second for the shells to do this, and since they aren't guard protected like green shells, red and blue shells thrown by humans near walls often can't align properly in time and collide with the walls which nullifies and destroys them once made contact.
There are exceptions to this though, such as if their is a hairpin turn and a CPU is on the other side of the wall, the shell's target priority will take place and try to go through the wall to hit the CPU to no avail. Also I think Nintendo knew how bad this programming was because every game onward always gives red and blue shells a proper and more efficient tracking when targeting CPU's even from a mile away.
I played this game today after maybe 20 years or something, and it struck me really hard how much the rivals cheat even on 50cc. They are on your tail at all times, it's crazy. Makes mk8 seem fair and balanced lol
MK is always rigged, the question is just how much
If it’s so bad that you get punished for being a good racer and keeping first, then it’s not a good racing game
Thanks for the video! This is a topic that always intrigued me since I was a kid, so having a more detailed explanation on how/why this happens is great.
As a kid of what. Get lost
Geeze, I knew the cheating was bad, but I didn't realize HOW bad it was! No wonder I remember this game as being frustrating rather than fun...
Got a question for you: Say I want to make a speedrun of a game that no one has ever run before - has even tried to find glitches, breaks, RNG manips, or the like before. What sort of advice do you have for such a scenario?
For the speedrunning thing, I'd say play the game. A lot. Experiment with all sorts of different builds in the game, and try to come up with unusual combinations or maneuvers that the game isn't strongly encouraging to do. Making multiple things happen at the same time is common in breaking games, so try to go for stuff like that. Button mash occasionally, try to slam into walls, you know, stuff that game testers do.
For RNG manips, learn to juggle a lot of Save States (assuming you're playing on emulator) before a given RNG event might occur, and then load it and go about it very slightly differently, also applying some of the above methods like button mashing (to see if useless inputs adjust the RNG, because they often do, especially in older games).
I'm not huge into speedrunning but I do like to try and break games for fun. Hope you have some fun with it, too.
I noticed that some CPUs were better than others in double dash, but I thought it was based on their current scores.
It depends on their weight class and your speed class.
If you race at 50cc, any CPU can become a rival.
At 100cc, medium and heavyweight CPUs can be rivals.
And on 150cc/Mirror, only heavyweights are rivals.
THAT explains why it’s only Bowser, Wario and DK f*cking with me
I feel like I've been gaslight my whole life by Nintendo and now a therapist is telling me it's not my fault lol
Mario Kart 64 multiplayer is fun. The single player game, however, is the most frustrating thing, especially on higher difficulty.
150cc isn't for the faint of heart
This is very smart programming thats makes the gane feel more organic and fun. Games which lack this and onky use speed rubber banding just feeling dull and/or frustrating. Especially kart racers where the player can abuse boosting and items. Games where items have full effect offscreen become extremely easy to cheese and you can easily lap NPCs multiple times in a race.
I still haven't forgiven Nintendo for this.
This is so cool... I'm 37 and a DevOps engineer. However, at the age I was when playing this game I didn't really know anything about programming or anything tech really. I've never thought much about this stuff, and watching this video it seems so obvious! But it's really cool to learn all this and actually think about all the experiences I had as a kid. Even now... since I'm not a "programmer" or "developer" per se ( i do write code... just not in their sense) I don't really think about games in this sense, but watching this kind of makes me want to think a little bit more about the programming challenges that go into this or decisions for why things work the way they do.
So all these years later and the game really was cheating??? Crazy
Makes you wonder how many games in general were rigged.
If you're too good then they need the advantage.
I grew up playing mario kart for snes and this was extreme in that game, they even had AI exclusive abilities/item effects they could just toggle whenever, Peach's cupcake, bowser's fireball, Mario and Luigi has the star effect ability, etc.
Makes me wonder if similar things are done in newer Mario Kart games as well.
Another interesting fact about CPUs and their behavior with bananas: when throwing a banana forward, it will most likely land in their own path (even throwing them at an angle to do so) resulting in them driving right into them. It's rather goofy.
And their general item pool is also restricted. They will never get any shells no matter what. they won't get any multi items either, like the banana bunch or golden mushrooms. If I remember correctly, they also don't get any lightning bolts.
pretty sure that's because of hardware limitations
And on another topic, let's talk red shells: Did you know/notice that red shells will only follow the road to get to the next racer when used on 5th place or lower? When on 4th or higher it will go straight for the next racer regardless of any obstructions, just hit the wall and be destroyed.
“Wario cheats!!!”
- Peralta, Jake
I encountered CPU cheats like this on the original Snes version, it meant you had to stay in second place the entire race and only pull ahead on the final lap if you wanted to win, the same technique also worked with the PS1 version of Grand Turismo.
Hmm, I think it'd be fun to test that theory about performance by limiting physics calculations/interactions.
Surely there's some sort of flag that CPU racers have that tells them they're being observed, is it possible to trip this flag for all of them and see if it affects performance at all?
You can trip it while they're on camera with a game shark code which is how I captured the off screen effects. But enabling them to always behave as if on screen is a bit different, and I'm not sure the code exists.
@@Abyssoft I don't remember the game shark code, but I do remember playing with a code with the real console that was cpu cheating turn off, and it is a really anyoing mess of screaming in Toad turnpike. They where off the player screen and still get hit by the trucks and car. with that game shark code on.
This Royal Raceway skip actually happened to a CPU when I was playing, young me wasn't very happy about it lol
It was always a mystery to me how and why did this happen, so thank you for clarifying this ^^
8:36 IM CRYING WHY DOES WARIO FLY OUT THE WALL
4:05 Mario kart, the only racing game that you press left and it moves you right.
Also Mario kart, the only racing game that cheats with EVERY way available.
They should have a rubberband free mode. Let me have the enjoyment of being much, much faster. Alas, they are but a bunch of cheats.
There might be gameshark codes that enable this
You might as well play Time Trials then, since you won't really see much of the CPU racers once you're front running.
Yes, there is a gameshark code for this!
CPU Drivers Process Offscreen
81028BAA B49A
I believe NBA Showtime or some other Midway game had he option to turn off “CPU Assistance” or somesich
Ignored this one for a few days assuming it was a video about rubber banding (as most videos about CPU cheating in racing games are) and left insanely informed on new jank that i had no idea about. Great video as usual man!
This is why MK64 is easily my least favorite entry in the series. That damn clipping and rubber banding was so frustrating. It just made the game feel like no matter how well I did, they would always cheat me out of my hard-earned victory.
Mario Kart really didn't get good until Super Circuit. They cheat in all games with rubber bands, but in Super MK and 64, it's barely even a racing game, it's just "you, 2 super cracked hackers and 5 pointless non-entities that make your life hell through random chaos".
This video was amazing, I always noticed and hated how the cpu would be immune to red shells and would only slow them down for a bit instead of doing the whole fling animation
Really good and detailed video, I wish you good luck and a happy new year.
You as well, thanks for watching!
Im a big Halo Wars fan, and this video reminds me of the legendary difficulty Computer A.I. and how they cheat as well. They make supplies quicker, and their units are slightly stronger than the player's units. As in:
A marine squad from a Legendary AI will always beat a marine squad from a player in a 1v1. Same unit, but theirs is slightly stronger in a one-on-one scenario. That goes for EVERY unit in the game.
This damage disparity is very noticeable when your Scarab is fighting their Scarab.
It can be annoying, but it makes it more rewarding in the end when you beat them.
I'm subscribed and didn't even got notified lmao
Do you have the bell clicked as well? Sometimes the algo will wait a bit to notify subscribers after a video has been uploaded, but if you have the bell activated it should be instant.
If Im not wrong you can change your rival in mid GP since the rival always need to be the 1st, 2nd or 3rd. So if you somehow manage to buly your Rival while making the 4th place steal the 1st place, he will become the new rival. while the new 4th place will act like a normal npc (all in point system)
Mario Kart 64 is not really pushing the hardware. It's a case of an early N64 game frankly not being very well made. Similarly, there's lots of bad, slow code in Super Mario 64 (now famously described by Kaze) or bad rendering decisions in Turok.
Mario 64 has the excuses that it was compiled badly AND it was developed in unfinished prototype hardware.
It's really amazing how the devs think of all this stuff, great work
Always hated cheating AI. Can't take the game seriously.
I should have called this video "Nintendo devs exposed"
Nah your title is perfect as is hahaha
Thats why is by far the worst Mario Kart game. Even MK on SNES played better.
@@sebastiankulcheNah.
@@urbanoutlaw858 Just saying facts.
11:08 "The game was pushing the limits of the hardware"
By looking like this.
Now I am curious about other racing games for N64. I wonder if CPUs cheat as much in games like Diddy Kong Racing or Excitebike 64!
Diddy kongs racing was a nightmare. I took me forever to get 100% on it.
This system is actually genius for those who want a challenge, I just wish they implemented this exact system or made it it's own difficulty in later mario kart games. Having the CPUs cause more trouble and be chaotic brings about all the fun in Mario.
This is why Mario Kart Super Circuit is the best early Mario Kart game
i didn’t know this game had a hidden rival system, I thought i was going crazy when I saw the exact same two racers targeting me through the entire cup
I wonder if there is a way to mod in the proper behavior for emulators and such.
Only if a romhacker has the will to do it. Not very likely though.
I actually knew most of this from just playing hours of this game as a kid. I first realized when the CPU actually stopped for banana peel hits but seemed to only BRIEFLY stop if hit with a collision item. On Toad's turnpike they were ONLY hit by cars if I literally SAW it or if they where VERY close behind with the collision occured. I can go on all day... Glad someone finally documented this though 😌
Happy to confirm childhood suspicions
Despite how much the AI cheats this is still my favorite Mario Kart in which I can consistently get first place in.
I noticed this about the CPUs back in the 90s. I always wondered why there was only two prominent cpu characters that kept up with me.
That's why I love this game. The cpu actually hates you, and I get such satisfaction with beating the challenge.
I prefer challenge to come from actual good design.
I noticed this problem on the SNES version of Mario Kart -- whichever order the CPUs spawned in determined what rank they thought they should be in. On high difficulties like 150cc this was way more apparent. If a CPU got significantly behind their expected position, they would rocket forward to catch up.
When I discovered this I took it to the extreme for fun:
After my friend slowed down the expected-to-be-in-first-place character (I think it was Princess Peach) with a couple attacks (I think red shell + ran them over when they were small), I stopped at the powerup area and hit her I think 3-4 more times (multiple red shells and a green one). At this point Princess Peach was almost an entire lap behind on one of the beach levels.
I just sat there watching the mini-map and laughing my ass off as Peach went around the map like an airplane. It was ridiculous how fast she was going, around 10x the max player speed!
Now, as far as racing games where the computers don't cheat, Rock N' Roll Racing for the SNES is my all-time favourite! On Warrior difficulty those computers are so tough to beat, and they can hit every obstacle, fall off every cliff, and can be blown up the same as a player. The only cheats they employ are an unlimited supply of lap-powerups (weapons, boosters, etc.). But I always figured that was just because they were richer and could afford better vehicle upgrades, lol. I never really minded the CPUs on RnR having endless nitros and missiles, they didn't usually spam them.
0:24 When you don't understand the meaning of the word, "coincidentally" 😂
I always wondered why I seem to get harassed by two specific CPUs during races, and knew the CPUs cheated in some way shape or form (namely the rubber band effect) but this thorough breakdown about their behavior is fantastic!
Yes anyone who played the game in the 90's could already tell the AI cheated. Glad you all figured this out 25+ years later.
No wonder how good someone gets you can never truly escpae the sound of burning rubber.
8:36
I'm imagining the "Wario crashes and dies" SFX rapidly approaching in this clip.
Honorable mention to the original F-Zero for also having ridiculous rubber-banding
There's TASes of the SNES version(s) that're split into real time-focused and IGT-focused, the sole difference in practice being that the real time TASes use some tricks to slow down just enough at the finish line to let one of the rival cars pass em by like one frame - getting 1st place proper burns time as the finish animation is much longer than if you get 2nd or 3rd
The fact though that the competition catches back up in jUST A FEW SECONDS despite having a lower top speed stat than the TAS's typical choice, the Fire Stingray (or I think the Luna Bomber in the case of the SatellaView ROMs), and no manual boosts like the player does is insane
I'm going to check these out for sure
@@Abyssoft CatadorDeLatas on SMWCentral has a good thread explaining F-Zero's AI along with a patch that makes the AI control your vehicle if you're looking for somewhere to start.
I knew it! I've just started playing this game and like I can hit these guys all day and they will catch right back up. I can barely stay in the lead and if I try 150cc I've have no chance to win 😢
Thats even more terrifying when you realize theyve only gotten deadlier with newer installments i swear ive seen them calculate the trajectory of a green shell bounce off a wall and hit mee LIKE HOW !!!!
3:23 DK Jr. thinks he’s one of the boys, who’s going to tell him?
We all knew it. We just didn't know what or how. The same applies to many other racing games. This prevents the player from winning too easily or losing too hard. SOME games have a setting to disable certain things like this. For example. I believe in some NFS games, there is a setting to disable AI Slowdown when they are ahead. This is meant to make the game more challenging.
Content at the tail end of the year is a tough gap for my sub feed. Much appreciation for taking the time out during the end of the year! Have a good one.
I was playing Mario kart 64 on the Switch VC two weeks ago and the entire time no matter the lead, no matter how well I did and amount of drift boosts I did I always had NPCs passing me non-stop on all cups, 50cc - mirror. I was like, wtf is going on here. It's impossible to keep any lead! And now I know rivals exist and they are out for blood. Thanks for sharing, that was really cool to learn about. lol.
When I was like 10 I noticed a lot of these things because you can turn on that track outline on the screen border and they stick out a lot. They dont bounce to indicate they were hit by red shells. They will, however, always spin out when hitting bananas.
I never was bothered by this since the player has so many advantages over them. They can't ever use shells or take advantage of any shortcut. Bananas also force them to a complete stop and spin unlike you being able to recover super quick if you use triple trap acceleration. You don't even need a hacked camera to see a lot of this. If you just stay far, but can still see them, you can see them not do the tumble. The game also relocates shells to specific places on the track once one goes off screen. Easiest one to see it is Wario Stadium because its in the dip underneath the big jump
You earned my sub! I love videos like this. Love seeing all the replies in the comments too. Excited to see what cover in the future!
Thanks for sticking around, should be a new video out next week!