🤓 Japan's grid system is split east/west when it comes to mains frequency (60 Hz east, 50 Hz west), but broadcasting and receiving equipment is designed to deal with that difference while using NTSC-J.
I grew up playing everything on 50 Hz and i literally had no idea something was wrong because i've had no reference to how it was supposed to be. Now i finally understand why every rendition of green hill zone ost sounds wrong compared to what i remember from the childhood.
Kosmic: I want to make a video about this cool SMB achievement Niftski: But that achievement hasn’t actually been done yet Kosmic:… Niftski: Fine, give me a week
That's like when Skilloz asked Pianist for a few recorded runs of 48 track items on MK8 so he could have more data for his RNG table and he just casually breaks 1:29 before the video could get finished. "Never ask me for anything again."
As an Australian gamer it wasn’t until I started emulating a lot of the games I used to play as a kid and seeing just how slow some were ape escape is a good example of just a sluggish game, then there were things like the battle mode in resident evil code Veronica on the pal version it ran slower but the in game timer doesn’t so getting the highest ranks wasn’t possible
It seems unfortunate for old school Australian gamers that your country is PAL. Your are relatively close to Japan and would presumably have had better access to Japanese NTSC imports than the USA.
Japan actually uses both 60Hz power and 50Hz power depending on the region. It turns out that the refresh rate of a TV signal actually doesn't need to match the frequency of the power. Interestingly, Color NTSC is actually not 60Hz, it's 60/1.001 (about 59.94) Hz. Black and white TV in the USA was originally 60Hz, but the field rate was decreased slightly when color was added to avoid interference between the color and black and white portions of the signal.
However, the (NTSC) NES strangely runs at about 60.0988 FPS. Perhaps another case of interference avoidance, or some such? (The hardware uses a ratio of exactly 39,375,000 / 655,171 FPS to be precise)
@@R2Bl3ndi do find it coincidental there are quartz oscilators around with 13.828 and 14.400 mhz when 576 pal lines times 24 fps = 13828, and 14400 for 25 fps
I only know Rodd Togers, and he has an astonishing 4:49 But on a different timeline where Billy Mitchell speedruns Super Mario Brothers, he's achieved a time of 3:59, putting all competitors to shame. He just needs to find the tapes with the recordings, and proof that his oddly shaped controller is in fact a genuine Super Mario Brothers controller.
Very interesting stuff. Someday someone will get to 4:53. It may take a decade or more, but it’ll happen! If it’s physically possible, it will be done. I think, sort of like you’re saying, once the PAL version record gets under 5, there will be more momentum to try and break the overall game’s record (as long as NTSC doesn’t go lower first!)
I've always found it interesting that PAL has got little interest, I remember vividly when the PAL TAS got rejected from TASVideos and how unusual it was for the fastest completion of a game to be ignored. Brilliant video - I am bewildered that there is even more to come after this (though I am glad to be someone who knew about all of these techniques before watching!)
from what I remember, TASVideos won't accept something if it's too similar even if it's a time improvement, and the TAS for PAL SMB was more or less played the same.
Commodore 64 gamers living in the states suffered the NTSC/PAL debacle in reverse - many of the best games were PAL-only, and attempts at conversion had mixed results similar to PAL NES games.
At worst, it could even screw up fastloaders and tape loaders that used the VIC-II for sensitive timings. Thankfully, a lot of revisions of the Amiga's Agnus chip can swap between NTSC and PAL easily, and depending on which ROM's inserted, you can even do so by holding the mouse buttons at boot and clicking a GUI button.
great video as usual man! one note: when you're comparing two things (like the music from two different version of the same game), it helps to put a tiny bit of the other version in for comparison. I know it's obvious to you because you've probably heard that mario theme _millions_ of times by now, but I haven't heard it quite that many times so i could use the reference 😂
@@Kosmicd12last time I had something weird. I can't remember in wich X-4 level it was, but i didn't kill Bowser, and I didn't jump "on" the axe either. I jumped "over" it, behind the axe. And instantly Mario walked towards toad, level was completed. So no fireballs to kill Bowser, no collapsing bridge. And the axe was still there, intact. Is that a random glitch or...? If not, can this save some frames? UPDATE: I made a short vid right after it happened again recently (it's on my channel), you can still see the axe intact.
@@sander230edisculpa no hablo inglés pero me llegó a suceder eso pero no en nes sino en SNES pero prácticamente lo que dijiste es real no hay puentes derrumbados ni bolas de fuego al hacer el glitch no suena la música de mundo completado solo Mario va directo hacia toad y avanza al siguiente mundo
This explanation was really good. If anyone is interested in some more detail, the earliest TVs used the power lines to approximately control the vertical hold; in other words, they had no timing circuits and just used the frequency of the alternating current to control the speed of the flyback transformers that bend the electron beam. That means a 60 Hz signal causes the TV to draw fields at a rate of 60 per second. This still doesn't synchronize the receiving TV with the signal; to do that, it had to latch onto the vertical synchronization signal (VSYNC), which for early TVs usually meant manually adjusting the vertical hold until it was very close to synchronization before the circuitry in the TV could lock on. And in 50 Hz countries, it worked the same way at 50 Hz. Once color standards were introduced, this was no longer a real problem, as the electronics in a color TV are far more complex anyway. However, the new TVs still had to be compatible with existing monochrome signals. That way, new TVs could pick up old signals and old TVs could pick up new signals. The first "compatible color" standard was NTSC, invented for the American market. This is technically not at 60 Hz, but just very close to it, for rather complicated reasons that you can read about on Wikipedia (or find in some good videos by Matt Parker or Technology Connections. The frame rate is actually multiplied by 1000/1001 in order to make the horizontal frequency an odd multiple of half the audio subcarrier frequency. (The carrier frequencies had already been assigned and could not be adjusted, so the field rate was adjusted instead). Otherwise, the new chrominance signal (which carries all the color information) would interfere with the audio subcarrier or else produce a visible dot pattern on monochrome sets. So the field rate was technically 60000/1001 ≈ 59.9401 fields/sec. This change of less than 0.1% caused no problems for existing sets, since they already had to handle slop in the power line, whose frequency is not guaranteed to be at precisely 60 Hz anyway but varies with power production and consumption. The second compatible color standard was PAL, invented for the European market. In that case, no adjusting was needed, because the horizontal frequency was already an odd multiple of half the audio frequency. Also, PAL alternates phase with every horizontal line drawn, which minimizes interference anyway. So PAL was still at exactly 50 Hz. Even today, US broadcast television has a 59.94 Hz frequency (or 29.97 frames/sec for interlaced TV), while PAL still has 50 (25) Hz. Early video game consoles therefore had to roughly match the 59.94 or 50 Hz frequencies of the TV sets they connected to. Again, they didn't have to be exact, but they had to be close. The Family Computer (Famicom, called the NES internationally) went with a 240-line signal at a frequency of 39375000/655171 ≈ 60.0988 fps progressive scan (it was not powerful enough to support interlacing), which is calculated from the nominal master clock frequency. Since this clearly won't work on PAL TVs, a different PAL version was sold in those countries, which runs at 3325214/66495 ≈ 50.0070 frames per second. Unfortunately, because mostly the same hardware was used, this could not benefit from PAL's increased vertical resolution, so the PAL version was effectively weaker in a sense, sacrificing some temporal frequency and getting no gain in vertical resolution. The lower frame rate often doesn't matter much for television, but it matters a lot for video games, since a game made in one region must be reprogrammed to work properly in another region. Games were typically first released for the Japanese or American market before being released internationally, so in practice this meant that the original design was for NTSC and it was then converted to PAL. Sometimes the conversion was pretty good, and at other times it was sloppy and created a worse experience in PAL countries. For Super Mario Bros., the conversion was pretty good, but it can never be perfect. This means there are slight discrepancies between the physics in American and European versions of the game. And in speedrunning, these small discrepancies can make a huge difference. For instance, in order for Mario to have approximately the same real speed in both versions, he has to move farther each frame on the PAL version. As a result, some clips that are impossible on NTSC become possible on PAL. As a result, most games have separate speedrunning leaderboards for the NTSC and PAL versions, and SMB1 is no exception. For most games, the PAL times are a bit slower, since in practice, converting a game from 60 to 50 Hz often meant slowing at least some parts of it down slightly. But for SMB1, the ability to get a full FPG on most levels along with the other tricks Kosmic mentioned ends up making its optimal time slightly faster instead. Thus, a 4:52 turns out to be theoretically possible on PAL, whereas the optimal time on NTSC is a 4:54.
PAL representative here and SMB1 was one of my first ever videogames. The NES I played actually had guite an adventure. It was originally my godfather's brother's who then gave (or sold, IDK which) it to my grandfather I think in early to mid 90's or so. It was at their home for some time but around early 2000's it was moved to our family log cabin accompanied with a TV from the 80's. It was in the cabin when first time I cleared SMB1 after about 10 years of trying (I was 13 at the time). Later when the cabin was sold because grandparents were too old and nobody else wanted to keep it the NES went to my cousin (godfather's kid) and couple years later in mid 2010's it was sold to an unknown person.
One of my first big experiences with game glitches was Pokémon, and the more I learned about why missingno (and later, other glitches) worked gave me such appreciation for how the developers slammed so much game into such a small ROM
Hobbyist game developer here, I thought some people might be interested in what I can say about framerate. Dealing with different framerates is still difficult today, and will always be difficult. Computers and game engines tend to handle things like audio quite well now, but if you make your own logic, you have to think about FPS too. Normally the game engine gives you a game loop that runs every frame, so it is FPS-dependent. You get the time passed since the last frame (delta), and you have to use that for any logic you have. Standard practice is to NOT use this for gameplay. The reason for this is, even if you consider the delta in your calculations, the effects can be different. Consider a jump that puts you at 60px/frame at first, and at 60 FPS, you lose 10px of upward velocity every frame. At 30 FPS, you could say, well, double the deceleration to compensate, so you lose 20px every frame. But 60+50+40+30+20+10 = 210 px, and 60+60+40+40+20+20 = 240px high jump. So at low FPS, you'd jump higher, even though you tried to account for it. Because of this, there is also a physics loop that tries its best to remain at a fixed framerate. Gameplay logic should go there. This physics loop is ready for all kinds of sacrifices to remain at 60 FPS (or a different framerate if you so choose), although even it isn't guaranteed to always remain like that. So it's still considered a good practice to use its delta there. Like most people, I personally don't (I have better things to worry about, like not taking 10 years to finish my game). So if my game's physics loop were forcibly changed to a different FPS, its gameplay would break.
Just a curious note: some game engines run the physics / game logic at a fixed rate decoupled from the actual framerate, IIRC Doom used 35 "ticks" per second and I'm pretty sure the Track Mania games also have a fixed tick rate, but I have no idea what it is. Speedrunners also love to exploit how rendering times affect game logic / character speed / collision detection, but that's a whole other can of worms. Anyways, thanks for sharing your knowledge and I wish you the best of luck with your game dev endeavor.
Great example of gameplay loop running every frame is Spyro Reignited Trilogy. If the game runs above 70-80fps, some intended jumps become impossible. As well as running it at 25fps lets you do some massive skips because of how high you jump
Nowadays only beginners tend to put their important logic in the game loop, and I believe more or less everything that can run games supports this "stable framerate loop". But this probably hasn't been the case in the past, so developers had to deal with it firsthand, putting everything in the FPS-dependent loop?
Also hobbyist game dev, and also computer science graduate. The reason why the time-delta game loop approach does not work well is down to numerical integration. Physics simulations, such as games, at their core are a bunch of differential equations. Not all of those are solvable analytically, think about e.g. the three body problem (it is not possible to create a mathematically perfect prediction of movement of 3 or more planets). This comes down to solutions needing to solve integrals that do not have analytic solutions. However since integrals have a geometric interpretation (the area under a functions curve) you can approximate every integral just by approximating the area under the function. You can do so by cutting it into small rectangles. In games where most often you're dealing with functions in time, you're essentially doing exactly that. Say you have a x,y position, then the position after some time delta t, is just the integral of the velocity over time + the current position. You approximate that by multiplying the current velocity with the time delta x = x+vx*t and y = y+vy*t. Then you update the velocity according to some game logic in a very similar manner. This approach has its limitations. It's scientific name is "Euler method", and one of the most blatant flaws it has it is unstable especially if the time slice is changing for every step. If you try to implement a simulation of a pendulum or a rope from scratch you will quickly realize it is not possible using that approach. Most physics engines use other approaches like the "Backward Euler method". It's usually not that serious in games as the results don't need to be physically accurate and small errors don't matter much. BUT if you ever wanted to improve your game physics, there is a whole science discipline revolving around numerical integration. Backward Euler, especially, is not much harder to implement and can have visible improvements resulting in slicker game play.
The lengths that speedrunners will go to just to save the the tiniest bit of time never ceases to amaze me. I really want to do more speedrunning stuff too.
Kosmic, I clicked onto this video with high hopes. And let me tell you, you met those expectations. The cinematography, the story telling. It truly is something special. Thank you for showing me how to complete Mario Bros.
The music at 10:24 got me messed up (Good Weather from Gimmick!), I found it in description and listened to it for 15 minutes before continuing the video, thank you for showing us this fire
This just puts into perspective about what you can find from trying from a different angle, eventually youll be able to land on something new! Cant wait for someone insane to preform this
yeah minecraft, mario oddysey, and mario 64 (and a multitude of other games) are impossible to play perfectly, unless you go into second gear like todd rogers did in his mom's basement in 1996, then of course you can get 4 seconds on any game.
@@matheuscabral9618 To be fair, a perfect run of Minecraft already has to deal with the statistics of world generation to make it impossible before any human skill or RNG comes into play
@@lol-gb5vt Nobody knows what the future will bring. With physical and mental modifications or even genetic manipulation, humans will probably be able to perfect these games at some point. At some point it will also be possible to dramatically increase the lifespan. And if a person can live for several centuries, someone will have enough time.
Shout out to old school Naughty Dog, their PAL conversions of Crash Bandicoot played fantastically well (movement slightly too fast, if I remember correctly, but that hardly feels bad!). That franchise deserved all the success it had in Europe purely because of that
Pokémon has an ACE and it’s very simplistic as well. The secret is finding a way to escape the jail of values you’re supposed to modify in RAM, but SMB1 doesn’t really have you write anything other than coins, score, lives, position, and level
I have to say, this video made me appreciate speedrunning a lot more. I never understood it. Utilizing bugs and glitches, which are basically an overlook on the developers part, to cut some corners to clock a better completion time? Where's the fun in that? But now I realize it's an impressive feat of engineering (figuring out the glitches, down to studying the source code, and learning to utilize them), fine motor skills and insane amount of reaction time and muscle memory. It's not so different from an IRL sports like football or F1 racing, where people impose rules and restrictions, artificial challenges to overcome. Thanks for the insight.
I've been waiting for this follow up. I've been wondering lately what the future of SMB speedruns would look like. I feel like we can't be too far off from a "perfect" run in standard any%. I was wondering if we might see new categories, such as an emulator category that allows left+right, just to give players something to shoot for. I'd never even considered the PAL version. I love these sort of details. Thanks, Kosmic, for always doing what you can to push the game to its limits.
I love these sorts of videos - people just assume that PAL is just the worst version (sometimes it is), but it is good to see you discuss the in-game differences. It may also explain why I could never get the 1-2 wall clip to get to the minus world to work properly on PAL following US guide books. Also another small point that you didn't mention is that PAL televisions used more vertical pixels than NTSC which may explain why Mario's jump height/fall rate differs from the Japanese & North American versions.
it works the same on pal haha, i tested it out while making this because I was curious. It might be a little different with different jump heights. Different screen size doesn't have a bearing on Mario's jump/falling, only the manual values they set for how many pixels / frame he should move.
I've known of Kosmic for what feels like forever but recently started checking out his videos. Great information, concise and coherent. I have no idea why speedrunning content is so exciting but I can't stop watching and this is one of the best channels for information! + Not to brag, but I use the same microphone as the world greatest Speedrunner 😏
As I've always said, PAL doesn't suck. There's a reason there's this one quote "Good going, PAL". NTSC there doesn't make much sense anyway. Except with "Nice try, PAL", but doesn't automagically mean NTSC is better, it's only one of the many options at that situation. For those nice PALs that read this, good going, PAL. As we can see from these examples, PAL can be your friend.
Oh, I'm so glad for this. I'm coming from PAL and played SMB1 on a NES here when I was a kid. Recently I attempted casually going back to the game (using NTSC on an emulator) and I thought I'm playing the game wrong for lacking the jump height and stuff like that. I didn't thought of a regional difference. Gotta switch revisions.
I think what many people are forgetting in these days is that this entire full game with all assets, sprites, pixels and sounds, is 40 KB! yeah, there's gonna be some glitches and exploits in a game that's less data than most 28x28 pixel emojis.
@@WilliamAndrews0 It is possible that some compressed formats like JPEG or PNG take more space than uncompressed bitmaps for sufficiently small images, due to the overhead of their representation. Image files can also include a bunch of metadata. Anyway, your comparison is not that far off. An uncompressed 128x128 bitmap with three color channels (no opacity channel) and 256 levels (one byte) in each channel gives just short of 50 kB. SMB 1985 is smaller.
Tape on the fingers to avoid blisters is smart. Would recommend investing in silicone thimbles used in sewing. They're more comfortable, last longer, stay on better. You can get packs of them in multiple sizes pretty cheap.
When Kosmic said that Niftski wasn’t able to do the last 2 levels, I was sure a « just kidding » moment was happening. I watched the other video 👀 Great explaination. Can’t wait for the next one 😊
Man, I’ve been waiting for this for months. It is so wonderful finally learning how the 4:53 can be achieved. Another fun fact is that there are actually 2 different PAL versions of this game. The first PAL version is essentially just the NTSC version but in slow motion, being completely unoptimised complete with slow sound. The second PAL version is the one discussed in the video where Nintendo put in some effort to optimise it. This is the version Nintendo would continue to re-release for many years, until they just decided to give PAL regions the NTSC versions.
Todd Rodgers got a 4:47 on the Duck Hunt/Super Mario Bros. PAL Version in either 1992 or 1993. He plugged the NES Zapper in the 2nd controller port, which makes Mario start out in 2nd gear.
People have claimed that we've hit the lowest possible time every 6 months for the last 10 years. At this point I'm never gonna believe anyone when they say "okay but this time for real"
I sometimes think that the only thing speedruners do is go fast, but then I remember that for them it's like a science. The amount of knowledge they have of the game they play is absurd.
Another technique for dealing with PAL is to do two frames worth of action on every fifth vertical refresh. On the Atari 2600, this could often work out pretty well if a game didn't need to use sprite collision hardware and all of the game logic was performed during the vertical blanking interval, since a PAL game would have about twice as many blank lines as NTSC. The NES can run game logic during the displayed portion of a frame, and if there's a lot of game logic running it twice in one frame might not be practical, but I would think for many games this would be the simplest way of adjusting speed without introducing quirkiness.
Guys, really, I need to shout out to smb speedrunning community. I am not speedrunning at all but enjoying your videos so much. Kosmic, Niftski and of course Bismuth. Just great job guys! Keep on grinding and making those fascinating and tense videos ❤
Another interesting tidbit about PAL, when television shows got ported over the programme had to be sped up which results in a higher pitch audio output. It's fairly noticable when you watch live TV and then look up a video of the same content. Somehow in the digital age the PAL versions of shows still get broadcast as well
As far as I remember, the main difference between pal and secam for consoles was colour rendering, meaning if you used pal consoles on secam tvs, you played black and white. Framerate is still 50. However, there is "or something". Famiclones for secam/pal region which played with ntsc cartridges and converted 60fps to 50 on the fly. No idea about performance though
9:40 I had a feeling that this was why I sometimes fall straight through the final moving platform in D-4, so it's cool to have that randomly confirmed in a video about something else.
5:41 (context Super Metroid) "You of course have to adjust all the sounds and music so they sound normal" The SNES SPC-700 coprocessor ran completely independent of the main CPU, and the v-sync timing was irrelevant to audio clock, which was identical on PAL and NSTC.
Alright now I have to ask a question, I’ve actually asked this a while ago and was told me it’s impossible but with this floor clip information I wanna ask is it possible to clip through the floor in the warp room in 4-2 to save time going through the pipe? Also is it possible to do a wrong warp in said pipe to reach the minus world?
You're talking about the warp zone room with warps to world 6, 7, and 8 right? You can't clip through stairs, there's no angle you can jump at that allows it. You need to be near the peak of your jump to clip into a wall, and with stairs, you cant get close enough. So you have to be already falling pretty fast before you reach the wall. You can see some clips through stairs in the "Minimum A Presses TAS"! A rare opportunity, utilizing a low gravity glitch. Speed aside, there's a way to get down the pipes with the warp zone not fully loaded. It just resets the room though. Minus world happens only in the 1-2 warp zone, because it sort of defaults to thinking you're in the end of level warp zone from 4-2, which only has 1 pipe in the middle to take you to world 5. So the side pipes take you to "nothing" - 1.
For protecting the fingers there's this finger tip glove thing I found on wish a while back, meant to be for touch screens but should do great for smooth dragging with the fingers. And yes, they are the same size as that blue tape or whatever that was shown in vid.
Ummm ACKSHUALLY it is super mario BROTHERS, not “bros” if you grew up in the 80 like me you would know this “Bros.” is an abbreviation. See the PERIOD?? kids these days try to “speedrun”, ha, back in my day that was called PLAYING. ;)
The lengths some speedrunners will go to just to save the tiniest bit of time never ceases to amaze. I wanna get into some speedrunning stuff of other games too.
Great video. The PAL version is certainly interesting, and I've only done 1 TAS with it. I rarely play it, and I'm not sure if it has more potential. :D
Ok, thanks for showing Sonic, cuz when i was playing, i thought "wish it was a bit faster, like on TV", and online ... i thought my memory wasn't that good(not great, but still), but it wasn't boosted emulators, my foggy memories, or me being slow, it was the difference between PAL and NTSC, the latter of which was shown on TV! Thanks to Kosmic for showing yet another great video, and Niftski and Miniland for saving our European a55es!
modded controllers aren't allowed in any of the categories discussed in this vid (and within these categories, seeing inputs for both left and right on the same frame is a dead giveaway for TAS gameplay), but this DOES have me wondering that with this knowledge, how much faster could a TAS go if allowed to double up on both left and right on the same frame
As a kid of parents working in US and Europe in the 1990s, NTSC/PAL was my worst enemy. Back then I thought it's just some stupid local format. Never could have imagined the technical marvels they had to pull to get the games working semi-accurately in different zones with different types of electric systems.
couldnt have the music for NTSC before/after your examples for PAL for all us PAL people who grew up with PAL and think thats how it should be and have no memory to compare it to? other wise good vid
This reminds me of when Darbian got a WR, and said he was only a stepping stone in the middle of the game's history. It feels like even 4:54 might be far from the end.
It's always super interesting to look at version differences. I think one of the more interesting examples is in the Super Monkey Ball franchise. The PAL version of Super Monkey Ball on Gamecube has both a 50Hz mode (PAL50) and an additional 60Hz mode (PAL60) that's accessible at startup by holding B. The physics of collisions remains unchanged between NTSC and PAL, so PAL60 feels normal, but PAL50 gains very slightly less speed from bounces and hits off of edges. By contrast, in Super Monkey Ball 2, the PAL version has updated collision physics that makes PAL50 collisions roughly equivalent to NTSC's 60Hz collisions; however, this different version of the collision physics is also used in PAL60, so PAL60 gains very slightly more speed in the situations where PAL50 lost speed in the original. I believe that Super Monkey Ball Deluxe fully resolves the issue, using different collision calculations between the PAL50 and PAL60 modes.
Ah, so that's why I found it strange that soundtracks from the NES games were slower based on my memories. The OSTs on TH-cam made them sound "faster", so I dropped it to 0.9x speed and found that this was the tempo I have been hearing as a child, which explains the "5/6" speed they were running at.
So if anyone was wondering why values are always modified as a function of DeltaTime in game programming, this is why. To program in real seconds instead of processor cycles as a unit of time, and thus make games framerate-independent. So much less work, and you just have to keep track of the total runtime and keep working out the time elapsed during every frame.
2:28 That reminds me how when I first saw the game, I was so little I didn't know how to read it correctly, so I pronounced it "super maREE-o bross". I didn't know what a "bross" was, but I knew this one was pretty super. Then for a brief time I thought it was about Marie's bras.
One of the advantages of pal game consoles/computers is, that because the lower framerate meant more time to render, this resulted in european amiga games not being able to run in america
Ummm ACKSHUALLY it is Super Mario BROTHERS, not "bros" if you grew up in the 80's like me you would know this "Bros." is an abbreviation. See the PERIOD?? kids these days try to "speedrun", ha, back in my day that was called PLAYING. ;)
I had no idea about the PAL version jumping higher. That would explain why I could jump the flagpole in 1-1 without the turtle glitch. I have seen videos explaining it isn't possible, but I know I did it. Finally an explanation :)
@@Kosmicd12 I have heard that so often, but I have a memory of consistently doing it on my PAL NES when I was a kid. I wish I had the console to try and repeat it since everything I hear says it is impossible
I just discovered this, but not only is PAL better for Any%, it really soars in Warpless. You calculated that the flagpole glitch saves 2.8 seconds... per flagpole. That means that if you add up all 24 levels that end with a Flagpole in a Warpless run, it'll potentially save as much as 67 seconds! Even if it doesn't save that many and make sub 18 possible, that trick alone would still catapult PAL far ahead of NTSC even just for making it a lot easier.
This title reminds me of CS surf. Some of the best routes are slower visually. But they give you more units due to your smoothness. So you wind up benefiting from not taking crazy tight angles. Has nothing to do with this game or game mechanics. But its pretty fascinating how we all achieve top speed in speedrunning.
![BOWKYLION - วิงวอน (ex-change) [Official MV]](http://i.ytimg.com/vi/cLdnZWDaO-w/mqdefault.jpg)
🤓 Japan's grid system is split east/west when it comes to mains frequency (60 Hz east, 50 Hz west), but broadcasting and receiving equipment is designed to deal with that difference while using NTSC-J.
Ur stupid.
ruined the 69 likes sorry
A temporary solution becoming permanent
🤓
Thank you for this genuinely interesting information
I grew up playing everything on 50 Hz and i literally had no idea something was wrong because i've had no reference to how it was supposed to be. Now i finally understand why every rendition of green hill zone ost sounds wrong compared to what i remember from the childhood.
Same
Same but i played this on bootleg console back in 2010
Same but I definitely always thought Mario kart seemed slow. We used to always go 150cc just so it played somewhat normal
It's crazy how I immediately noticed something weird with the physics in the first few seconds
Yeah that's called being a nerd😂 Love your stuff man!❤
The graphics on this game kinda sucks
I didn't.
@@tired6919the recent ray tracing patch made a huge difference with the lighting engine though.
nice, I was not the first to notice also
I love how Niftski decided to just "get" the ffpgs. Too awesome.
Kosmic: I want to make a video about this cool SMB achievement
Niftski: But that achievement hasn’t actually been done yet
Kosmic:…
Niftski: Fine, give me a week
That's like when Skilloz asked Pianist for a few recorded runs of 48 track items on MK8 so he could have more data for his RNG table and he just casually breaks 1:29 before the video could get finished.
"Never ask me for anything again."
@@MecceldorfLMAO that's a great way to see it
As an Australian gamer it wasn’t until I started emulating a lot of the games I used to play as a kid and seeing just how slow some were ape escape is a good example of just a sluggish game, then there were things like the battle mode in resident evil code Veronica on the pal version it ran slower but the in game timer doesn’t so getting the highest ranks wasn’t possible
I think nearly every simulator supports both PAL and NTSC. It was such a common problem.
What a mess
Ff9 had the same problem imagine trying to get the Excalibur secret sword with 2/3 the time
It seems unfortunate for old school Australian gamers that your country is PAL. Your are relatively close to Japan and would presumably have had better access to Japanese NTSC imports than the USA.
Ape escape mentioned W
Japan actually uses both 60Hz power and 50Hz power depending on the region. It turns out that the refresh rate of a TV signal actually doesn't need to match the frequency of the power.
Interestingly, Color NTSC is actually not 60Hz, it's 60/1.001 (about 59.94) Hz. Black and white TV in the USA was originally 60Hz, but the field rate was decreased slightly when color was added to avoid interference between the color and black and white portions of the signal.
and there is also PAl with 59.94 in Brazil
However, the (NTSC) NES strangely runs at about 60.0988 FPS. Perhaps another case of interference avoidance, or some such?
(The hardware uses a ratio of exactly 39,375,000 / 655,171 FPS to be precise)
@@Muhahahahaz Things don't need to match perfectly with analogue video signals, as long as they're "close enough".
@@Muhahahahazthat could have been a result of the precision limit of the timing components used; that's my guess at least.
@@R2Bl3ndi do find it coincidental there are quartz oscilators around with 13.828 and 14.400 mhz
when 576 pal lines times 24 fps = 13828, and 14400 for 25 fps
4:51 is also possible, but only Todd Togers can do it.
@ayo3291 The absolute legend!
I only know Rodd Togers, and he has an astonishing 4:49
But on a different timeline where Billy Mitchell speedruns Super Mario Brothers, he's achieved a time of 3:59, putting all competitors to shame. He just needs to find the tapes with the recordings, and proof that his oddly shaped controller is in fact a genuine Super Mario Brothers controller.
@@marcohueber5130 bro wdym billy mitchell already got a 1:54
Ddor Goters got a :10 though
@@Lugiamasterbrony And that was done while he also broke the Super Mario 64 record by getting 121 stars instead of 120.
Very interesting stuff. Someday someone will get to 4:53. It may take a decade or more, but it’ll happen! If it’s physically possible, it will be done. I think, sort of like you’re saying, once the PAL version record gets under 5, there will be more momentum to try and break the overall game’s record (as long as NTSC doesn’t go lower first!)
You say years.
My money?
6 months.
(someone will say 3 months and win).
3 months
@@ThunderChunky101 nah lol quadruple ffpg is horrible
@@Jplays-Games dammit! 😂
give niftski two hours
I've always found it interesting that PAL has got little interest, I remember vividly when the PAL TAS got rejected from TASVideos and how unusual it was for the fastest completion of a game to be ignored. Brilliant video - I am bewildered that there is even more to come after this (though I am glad to be someone who knew about all of these techniques before watching!)
from what I remember, TASVideos won't accept something if it's too similar even if it's a time improvement, and the TAS for PAL SMB was more or less played the same.
Commodore 64 gamers living in the states suffered the NTSC/PAL debacle in reverse - many of the best games were PAL-only, and attempts at conversion had mixed results similar to PAL NES games.
Yeah, i remember there being something about the C64 CPU being clocked lower but because it had to render less images it was practically faster.
At worst, it could even screw up fastloaders and tape loaders that used the VIC-II for sensitive timings. Thankfully, a lot of revisions of the Amiga's Agnus chip can swap between NTSC and PAL easily, and depending on which ROM's inserted, you can even do so by holding the mouse buttons at boot and clicking a GUI button.
The progress this game has shown in just a few years is honestly pretty insane.
great video as usual man!
one note: when you're comparing two things (like the music from two different version of the same game), it helps to put a tiny bit of the other version in for comparison. I know it's obvious to you because you've probably heard that mario theme _millions_ of times by now, but I haven't heard it quite that many times so i could use the reference 😂
The PAL version is noticeably faster than NTSC, although not dramatically so. It's slower than the hurry up NTSC version, but imo not by much.
that's true, good point. Thanks :)
@@Kosmicd12 Point's been made already but came here to say that too... I'm in the UK and everything here sounds like it always did 😆
@@Kosmicd12last time I had something weird. I can't remember in wich X-4 level it was, but i didn't kill Bowser, and I didn't jump "on" the axe either. I jumped "over" it, behind the axe. And instantly Mario walked towards toad, level was completed. So no fireballs to kill Bowser, no collapsing bridge. And the axe was still there, intact. Is that a random glitch or...? If not, can this save some frames? UPDATE: I made a short vid right after it happened again recently (it's on my channel), you can still see the axe intact.
@@sander230edisculpa no hablo inglés pero me llegó a suceder eso pero no en nes sino en SNES pero prácticamente lo que dijiste es real no hay puentes derrumbados ni bolas de fuego al hacer el glitch no suena la música de mundo completado solo Mario va directo hacia toad y avanza al siguiente mundo
Looking forward to a video next week after Nifski ends up getting a 4:53 over the weekend
This explanation was really good. If anyone is interested in some more detail, the earliest TVs used the power lines to approximately control the vertical hold; in other words, they had no timing circuits and just used the frequency of the alternating current to control the speed of the flyback transformers that bend the electron beam. That means a 60 Hz signal causes the TV to draw fields at a rate of 60 per second. This still doesn't synchronize the receiving TV with the signal; to do that, it had to latch onto the vertical synchronization signal (VSYNC), which for early TVs usually meant manually adjusting the vertical hold until it was very close to synchronization before the circuitry in the TV could lock on. And in 50 Hz countries, it worked the same way at 50 Hz.
Once color standards were introduced, this was no longer a real problem, as the electronics in a color TV are far more complex anyway. However, the new TVs still had to be compatible with existing monochrome signals. That way, new TVs could pick up old signals and old TVs could pick up new signals. The first "compatible color" standard was NTSC, invented for the American market. This is technically not at 60 Hz, but just very close to it, for rather complicated reasons that you can read about on Wikipedia (or find in some good videos by Matt Parker or Technology Connections. The frame rate is actually multiplied by 1000/1001 in order to make the horizontal frequency an odd multiple of half the audio subcarrier frequency. (The carrier frequencies had already been assigned and could not be adjusted, so the field rate was adjusted instead). Otherwise, the new chrominance signal (which carries all the color information) would interfere with the audio subcarrier or else produce a visible dot pattern on monochrome sets. So the field rate was technically 60000/1001 ≈ 59.9401 fields/sec. This change of less than 0.1% caused no problems for existing sets, since they already had to handle slop in the power line, whose frequency is not guaranteed to be at precisely 60 Hz anyway but varies with power production and consumption.
The second compatible color standard was PAL, invented for the European market. In that case, no adjusting was needed, because the horizontal frequency was already an odd multiple of half the audio frequency. Also, PAL alternates phase with every horizontal line drawn, which minimizes interference anyway. So PAL was still at exactly 50 Hz. Even today, US broadcast television has a 59.94 Hz frequency (or 29.97 frames/sec for interlaced TV), while PAL still has 50 (25) Hz.
Early video game consoles therefore had to roughly match the 59.94 or 50 Hz frequencies of the TV sets they connected to. Again, they didn't have to be exact, but they had to be close. The Family Computer (Famicom, called the NES internationally) went with a 240-line signal at a frequency of 39375000/655171 ≈ 60.0988 fps progressive scan (it was not powerful enough to support interlacing), which is calculated from the nominal master clock frequency. Since this clearly won't work on PAL TVs, a different PAL version was sold in those countries, which runs at 3325214/66495 ≈ 50.0070 frames per second. Unfortunately, because mostly the same hardware was used, this could not benefit from PAL's increased vertical resolution, so the PAL version was effectively weaker in a sense, sacrificing some temporal frequency and getting no gain in vertical resolution.
The lower frame rate often doesn't matter much for television, but it matters a lot for video games, since a game made in one region must be reprogrammed to work properly in another region. Games were typically first released for the Japanese or American market before being released internationally, so in practice this meant that the original design was for NTSC and it was then converted to PAL. Sometimes the conversion was pretty good, and at other times it was sloppy and created a worse experience in PAL countries. For Super Mario Bros., the conversion was pretty good, but it can never be perfect. This means there are slight discrepancies between the physics in American and European versions of the game. And in speedrunning, these small discrepancies can make a huge difference. For instance, in order for Mario to have approximately the same real speed in both versions, he has to move farther each frame on the PAL version. As a result, some clips that are impossible on NTSC become possible on PAL.
As a result, most games have separate speedrunning leaderboards for the NTSC and PAL versions, and SMB1 is no exception. For most games, the PAL times are a bit slower, since in practice, converting a game from 60 to 50 Hz often meant slowing at least some parts of it down slightly. But for SMB1, the ability to get a full FPG on most levels along with the other tricks Kosmic mentioned ends up making its optimal time slightly faster instead. Thus, a 4:52 turns out to be theoretically possible on PAL, whereas the optimal time on NTSC is a 4:54.
PAL representative here and SMB1 was one of my first ever videogames. The NES I played actually had guite an adventure. It was originally my godfather's brother's who then gave (or sold, IDK which) it to my grandfather I think in early to mid 90's or so. It was at their home for some time but around early 2000's it was moved to our family log cabin accompanied with a TV from the 80's. It was in the cabin when first time I cleared SMB1 after about 10 years of trying (I was 13 at the time). Later when the cabin was sold because grandparents were too old and nobody else wanted to keep it the NES went to my cousin (godfather's kid) and couple years later in mid 2010's it was sold to an unknown person.
Sounds so fun and nostalgic to play SMB in a log cabin on an 80s TV
This video had me from the beginning and never let go. Well done sir. Looking forward to what's next!
It's kind of ironic that whenever I learn about game glitches, I'm always struck with admiration for the incredible work done by game devs.
One of my first big experiences with game glitches was Pokémon, and the more I learned about why missingno (and later, other glitches) worked gave me such appreciation for how the developers slammed so much game into such a small ROM
Hobbyist game developer here, I thought some people might be interested in what I can say about framerate. Dealing with different framerates is still difficult today, and will always be difficult. Computers and game engines tend to handle things like audio quite well now, but if you make your own logic, you have to think about FPS too.
Normally the game engine gives you a game loop that runs every frame, so it is FPS-dependent. You get the time passed since the last frame (delta), and you have to use that for any logic you have. Standard practice is to NOT use this for gameplay. The reason for this is, even if you consider the delta in your calculations, the effects can be different. Consider a jump that puts you at 60px/frame at first, and at 60 FPS, you lose 10px of upward velocity every frame. At 30 FPS, you could say, well, double the deceleration to compensate, so you lose 20px every frame. But 60+50+40+30+20+10 = 210 px, and 60+60+40+40+20+20 = 240px high jump. So at low FPS, you'd jump higher, even though you tried to account for it.
Because of this, there is also a physics loop that tries its best to remain at a fixed framerate. Gameplay logic should go there. This physics loop is ready for all kinds of sacrifices to remain at 60 FPS (or a different framerate if you so choose), although even it isn't guaranteed to always remain like that. So it's still considered a good practice to use its delta there. Like most people, I personally don't (I have better things to worry about, like not taking 10 years to finish my game). So if my game's physics loop were forcibly changed to a different FPS, its gameplay would break.
Just a curious note: some game engines run the physics / game logic at a fixed rate decoupled from the actual framerate, IIRC Doom used 35 "ticks" per second and I'm pretty sure the Track Mania games also have a fixed tick rate, but I have no idea what it is. Speedrunners also love to exploit how rendering times affect game logic / character speed / collision detection, but that's a whole other can of worms.
Anyways, thanks for sharing your knowledge and I wish you the best of luck with your game dev endeavor.
Great example of gameplay loop running every frame is Spyro Reignited Trilogy.
If the game runs above 70-80fps, some intended jumps become impossible. As well as running it at 25fps lets you do some massive skips because of how high you jump
Delta time in general is "cool", but it is a terrible way to deal with lower framerate, because of how inconsistent your game can become.
Nowadays only beginners tend to put their important logic in the game loop, and I believe more or less everything that can run games supports this "stable framerate loop". But this probably hasn't been the case in the past, so developers had to deal with it firsthand, putting everything in the FPS-dependent loop?
Also hobbyist game dev, and also computer science graduate. The reason why the time-delta game loop approach does not work well is down to numerical integration. Physics simulations, such as games, at their core are a bunch of differential equations. Not all of those are solvable analytically, think about e.g. the three body problem (it is not possible to create a mathematically perfect prediction of movement of 3 or more planets). This comes down to solutions needing to solve integrals that do not have analytic solutions.
However since integrals have a geometric interpretation (the area under a functions curve) you can approximate every integral just by approximating the area under the function. You can do so by cutting it into small rectangles.
In games where most often you're dealing with functions in time, you're essentially doing exactly that. Say you have a x,y position, then the position after some time delta t, is just the integral of the velocity over time + the current position. You approximate that by multiplying the current velocity with the time delta x = x+vx*t and y = y+vy*t. Then you update the velocity according to some game logic in a very similar manner. This approach has its limitations. It's scientific name is "Euler method", and one of the most blatant flaws it has it is unstable especially if the time slice is changing for every step. If you try to implement a simulation of a pendulum or a rope from scratch you will quickly realize it is not possible using that approach. Most physics engines use other approaches like the "Backward Euler method".
It's usually not that serious in games as the results don't need to be physically accurate and small errors don't matter much. BUT if you ever wanted to improve your game physics, there is a whole science discipline revolving around numerical integration. Backward Euler, especially, is not much harder to implement and can have visible improvements resulting in slicker game play.
That slow sonic hurt my soul bro 4:44
The lengths that speedrunners will go to just to save the the tiniest bit of time never ceases to amaze me. I really want to do more speedrunning stuff too.
Kosmic, I clicked onto this video with high hopes. And let me tell you, you met those expectations. The cinematography, the story telling. It truly is something special. Thank you for showing me how to complete Mario Bros.
was not expecting you to be here lol, used to watching you beat extremes but im happy to see u in diff communities ngl
Trick?
Yo what
Trick is here?
The music at 10:24 got me messed up (Good Weather from Gimmick!), I found it in description and listened to it for 15 minutes before continuing the video, thank you for showing us this fire
This just puts into perspective about what you can find from trying from a different angle, eventually youll be able to land on something new! Cant wait for someone insane to preform this
This might be your best video. This game's community of players and researchers is unbelievable
bro got sponsored by netflix
🤓”bro got sponsored by netflix”
Bro visited his friend
@@alternateaquathis so much this. reddit gold for you sir
I farted
If it is somehow possible, it will be done at some point. That is a fundamental rule in speedrunning. Even if it takes 100 years.
Not necessarily, a perfect run of Minecraft will never happen
yeah minecraft, mario oddysey, and mario 64 (and a multitude of other games) are impossible to play perfectly, unless you go into second gear like todd rogers did in his mom's basement in 1996, then of course you can get 4 seconds on any game.
@@lol-gb5vt oh yeah I forgot about Todd Rogers my bad, obviously possible tho we would have to get him out of jail for him to do that
@@matheuscabral9618 To be fair, a perfect run of Minecraft already has to deal with the statistics of world generation to make it impossible before any human skill or RNG comes into play
@@lol-gb5vt Nobody knows what the future will bring. With physical and mental modifications or even genetic manipulation, humans will probably be able to perfect these games at some point. At some point it will also be possible to dramatically increase the lifespan. And if a person can live for several centuries, someone will have enough time.
Shout out to old school Naughty Dog, their PAL conversions of Crash Bandicoot played fantastically well (movement slightly too fast, if I remember correctly, but that hardly feels bad!). That franchise deserved all the success it had in Europe purely because of that
might be one of the reasons the playstation was always more beloved in europe
Also Rare's games were optimized well for 50Hz because it was a British developer. Donkey Kong Country trilogy for example
Huh, who would have thought that these tiny differences mean so much!?
Hey new TH-cam feature cool
@@Thegoatone23Hey new TH-cam feature cool
@@Thegoatone23 the subscriber badge?
1~1.20* speed difference does that, because it affects *everything* down to the minute.
@@Nitosa yes, that makes sense. I guess I just never thought about it before :)
Kosmic 20 years later be like: “sub 1 minute is possible”
Someone needs to find an ACE
@@CrushedAsian255yeah, but where? the game may be too simplistic for an ace.
There is a tas that do that if i understand what you mean@@CrushedAsian255
@@starleaf-luna most likely
Pokémon has an ACE and it’s very simplistic as well. The secret is finding a way to escape the jail of values you’re supposed to modify in RAM, but SMB1 doesn’t really have you write anything other than coins, score, lives, position, and level
I have to say, this video made me appreciate speedrunning a lot more. I never understood it. Utilizing bugs and glitches, which are basically an overlook on the developers part, to cut some corners to clock a better completion time? Where's the fun in that? But now I realize it's an impressive feat of engineering (figuring out the glitches, down to studying the source code, and learning to utilize them), fine motor skills and insane amount of reaction time and muscle memory. It's not so different from an IRL sports like football or F1 racing, where people impose rules and restrictions, artificial challenges to overcome. Thanks for the insight.
Love your vids! Tysm for getting me into speedrunning again!
At 4:33 I feel so much nostalgia just from this pitch, it always reminds me of watching old SM64 content back in the day ;)
I've been waiting for this follow up. I've been wondering lately what the future of SMB speedruns would look like. I feel like we can't be too far off from a "perfect" run in standard any%. I was wondering if we might see new categories, such as an emulator category that allows left+right, just to give players something to shoot for. I'd never even considered the PAL version. I love these sort of details. Thanks, Kosmic, for always doing what you can to push the game to its limits.
I love these sorts of videos - people just assume that PAL is just the worst version (sometimes it is), but it is good to see you discuss the in-game differences. It may also explain why I could never get the 1-2 wall clip to get to the minus world to work properly on PAL following US guide books. Also another small point that you didn't mention is that PAL televisions used more vertical pixels than NTSC which may explain why Mario's jump height/fall rate differs from the Japanese & North American versions.
it works the same on pal haha, i tested it out while making this because I was curious. It might be a little different with different jump heights.
Different screen size doesn't have a bearing on Mario's jump/falling, only the manual values they set for how many pixels / frame he should move.
In my experience, most games on most consoles letterbox their output rather than making the display area larger or anything like that.
I've known of Kosmic for what feels like forever but recently started checking out his videos. Great information, concise and coherent. I have no idea why speedrunning content is so exciting but I can't stop watching and this is one of the best channels for information!
+ Not to brag, but I use the same microphone as the world greatest Speedrunner 😏
Your speedrunning content is way more interesting than most. I love how you fully lay out everything important. Can't wait for part 3!
"If you can play it slowly, you can play it quickly" -a music speedrunner
Ah, a man of lingling culture.
As I've always said, PAL doesn't suck. There's a reason there's this one quote "Good going, PAL". NTSC there doesn't make much sense anyway. Except with "Nice try, PAL", but doesn't automagically mean NTSC is better, it's only one of the many options at that situation.
For those nice PALs that read this, good going, PAL. As we can see from these examples, PAL can be your friend.
Of _course_ Niftski wouldn't let you make a video while those times hadn't yet been completed! Great job with this video!
Oh, I'm so glad for this. I'm coming from PAL and played SMB1 on a NES here when I was a kid. Recently I attempted casually going back to the game (using NTSC on an emulator) and I thought I'm playing the game wrong for lacking the jump height and stuff like that. I didn't thought of a regional difference. Gotta switch revisions.
PAL is like the minecraft bedrock of smb speedruns
I think what many people are forgetting in these days is that this entire full game with all assets, sprites, pixels and sounds, is 40 KB! yeah, there's gonna be some glitches and exploits in a game that's less data than most 28x28 pixel emojis.
40 kB is not much, but to be fair, even an uncompressed 28x28 bitmap with four channels and four bytes per channel is less than that.
@@EliasHasle I usually see 16x16 emojis with 25kb, i just went with the next biggest size and estimated.
@@WilliamAndrews0 It is possible that some compressed formats like JPEG or PNG take more space than uncompressed bitmaps for sufficiently small images, due to the overhead of their representation. Image files can also include a bunch of metadata. Anyway, your comparison is not that far off. An uncompressed 128x128 bitmap with three color channels (no opacity channel) and 256 levels (one byte) in each channel gives just short of 50 kB. SMB 1985 is smaller.
Always a good day when Kosmic uploads
Fr fr
Tape on the fingers to avoid blisters is smart. Would recommend investing in silicone thimbles used in sewing. They're more comfortable, last longer, stay on better. You can get packs of them in multiple sizes pretty cheap.
Kosmic please 2.2 just came out I can’t take another revolution
lmao
When Kosmic said that Niftski wasn’t able to do the last 2 levels, I was sure a « just kidding » moment was happening. I watched the other video 👀
Great explaination. Can’t wait for the next one 😊
concise enough to be understandable, detailed enough to be interesting, fantastic video kosmic :)
Man, I’ve been waiting for this for months. It is so wonderful finally learning how the 4:53 can be achieved. Another fun fact is that there are actually 2 different PAL versions of this game. The first PAL version is essentially just the NTSC version but in slow motion, being completely unoptimised complete with slow sound. The second PAL version is the one discussed in the video where Nintendo put in some effort to optimise it. This is the version Nintendo would continue to re-release for many years, until they just decided to give PAL regions the NTSC versions.
I’m glad I have gotten to see the progression of this series early on.
Todd Rodgers got a 4:47 on the Duck Hunt/Super Mario Bros. PAL Version in either 1992 or 1993. He plugged the NES Zapper in the 2nd controller port, which makes Mario start out in 2nd gear.
so would this theoretical record be the true perfect run? or can we go even faster?
People have claimed that we've hit the lowest possible time every 6 months for the last 10 years. At this point I'm never gonna believe anyone when they say "okay but this time for real"
Maybe even faster in a decade or something :)
It's the perfect run until another exploit is found.
It's a different game, so no
@@ParadiseDB7that's not how that works
I sometimes think that the only thing speedruners do is go fast, but then I remember that for them it's like a science. The amount of knowledge they have of the game they play is absurd.
Part of me was expecting the trick to be playing the PAL version on an NTSC console to speed up everything lol
Another technique for dealing with PAL is to do two frames worth of action on every fifth vertical refresh. On the Atari 2600, this could often work out pretty well if a game didn't need to use sprite collision hardware and all of the game logic was performed during the vertical blanking interval, since a PAL game would have about twice as many blank lines as NTSC. The NES can run game logic during the displayed portion of a frame, and if there's a lot of game logic running it twice in one frame might not be practical, but I would think for many games this would be the simplest way of adjusting speed without introducing quirkiness.
"Guys it's real but only in the European version I swear"… but real
8:00 the moment you said "the length of a framerule" I said 'oh my god' out loud on a public train
Nice to see some PAL love here!
Summoning Salt quivering rn
Guys, really, I need to shout out to smb speedrunning community. I am not speedrunning at all but enjoying your videos so much. Kosmic, Niftski and of course Bismuth. Just great job guys! Keep on grinding and making those fascinating and tense videos ❤
If I have learnt anything is that anything a TAS can do a human seems to eventually be able to do too
Another interesting tidbit about PAL, when television shows got ported over the programme had to be sped up which results in a higher pitch audio output. It's fairly noticable when you watch live TV and then look up a video of the same content. Somehow in the digital age the PAL versions of shows still get broadcast as well
Its insane that a thing as simple as how many hertz your outlets are can help you get a world record
It is... beautiful to see people with such passion about ANYTHING.
My guess for the next video is that there is a SECAM version prototype or something that has the theoretical fastest time if it was playable.
As far as I remember, the main difference between pal and secam for consoles was colour rendering, meaning if you used pal consoles on secam tvs, you played black and white. Framerate is still 50.
However, there is "or something". Famiclones for secam/pal region which played with ntsc cartridges and converted 60fps to 50 on the fly. No idea about performance though
@@dinocat8 That's interesting
@@dinocat8I had one of those and it runs a bit slow otherwise there are no major differences
9:40 I had a feeling that this was why I sometimes fall straight through the final moving platform in D-4, so it's cool to have that randomly confirmed in a video about something else.
21:00 I am so angry at this video being clickbait. Now I am going to write an angry comment.
I can't believe I never even thought about the PAL version of SMB or heard anyone talk about it. This is awesome.
It’s insane seeing all of these things happening for such a old game like SMB1
What you need to do is pizz. really fast on a string instrument to deal with the pain. Playing double bass is really god for this.
That "Sorry guys, I click baited you" scared the sh*t out of me 😅😅
5:41 (context Super Metroid) "You of course have to adjust all the sounds and music so they sound normal"
The SNES SPC-700 coprocessor ran completely independent of the main CPU, and the v-sync timing was irrelevant to audio clock, which was identical on PAL and NSTC.
Brazil is also 60 despite being "pal". its a rare case where it made the best choice possible.
I lost my mind when i heard waterworld music in the background. That ost is a hidden gem. Banger video
Alright now I have to ask a question, I’ve actually asked this a while ago and was told me it’s impossible but with this floor clip information I wanna ask is it possible to clip through the floor in the warp room in 4-2 to save time going through the pipe? Also is it possible to do a wrong warp in said pipe to reach the minus world?
You're talking about the warp zone room with warps to world 6, 7, and 8 right? You can't clip through stairs, there's no angle you can jump at that allows it. You need to be near the peak of your jump to clip into a wall, and with stairs, you cant get close enough. So you have to be already falling pretty fast before you reach the wall. You can see some clips through stairs in the "Minimum A Presses TAS"! A rare opportunity, utilizing a low gravity glitch. Speed aside, there's a way to get down the pipes with the warp zone not fully loaded. It just resets the room though. Minus world happens only in the 1-2 warp zone, because it sort of defaults to thinking you're in the end of level warp zone from 4-2, which only has 1 pipe in the middle to take you to world 5. So the side pipes take you to "nothing" - 1.
@@Kosmicd12 Would that be a water level with no defined exit, perchance?
For protecting the fingers there's this finger tip glove thing I found on wish a while back, meant to be for touch screens but should do great for smooth dragging with the fingers. And yes, they are the same size as that blue tape or whatever that was shown in vid.
Ummm ACKSHUALLY it is super mario BROTHERS, not “bros” if you grew up in the 80 like me you would know this “Bros.” is an abbreviation. See the PERIOD?? kids these days try to “speedrun”, ha, back in my day that was called PLAYING. ;)
i wasn’t born in the 80’s btw, i just said this for the memes.
The lengths some speedrunners will go to just to save the tiniest bit of time never ceases to amaze. I wanna get into some speedrunning stuff of other games too.
Great video. The PAL version is certainly interesting, and I've only done 1 TAS with it. I rarely play it, and I'm not sure if it has more potential. :D
Ok, thanks for showing Sonic, cuz when i was playing, i thought "wish it was a bit faster, like on TV", and online ... i thought my memory wasn't that good(not great, but still), but it wasn't boosted emulators, my foggy memories, or me being slow, it was the difference between PAL and NTSC, the latter of which was shown on TV!
Thanks to Kosmic for showing yet another great video, and Niftski and Miniland for saving our European a55es!
Just wanted to point out that 4:51 is possible with an L+R modded controller (the TAS is 4:51.7)
modded controllers aren't allowed in any of the categories discussed in this vid (and within these categories, seeing inputs for both left and right on the same frame is a dead giveaway for TAS gameplay), but this DOES have me wondering that with this knowledge, how much faster could a TAS go if allowed to double up on both left and right on the same frame
@@thag_simmons Yeah, I know that. I was saying that *theoretically* it's possible to get 4:51 without the restriction of stock controllers.
As a kid of parents working in US and Europe in the 1990s, NTSC/PAL was my worst enemy. Back then I thought it's just some stupid local format. Never could have imagined the technical marvels they had to pull to get the games working semi-accurately in different zones with different types of electric systems.
so pal version isn't just a joke for SiIvaGunner rips?
8:06 i love the 21 against 18 polyrhythm
Awesome video!! :D Always learn a lot
Oh hi
I love your usage of Soleil music at 7:16. It had its problems, but I still love that game.
couldnt have the music for NTSC before/after your examples for PAL for all us PAL people who grew up with PAL and think thats how it should be and have no memory to compare it to? other wise good vid
This reminds me of when Darbian got a WR, and said he was only a stepping stone in the middle of the game's history. It feels like even 4:54 might be far from the end.
druga w nocy idealna godzina na obejrzenie nowego odcinka
dfsdfsdfsdfd
ye, but 3 am. is better for clickbait
ttest
Cześć! Pozdrawiam z Stanów Zjednoczonych! Uczę się języka polskiego bo uwielbiam Polskę! 🇵🇱
It's always super interesting to look at version differences. I think one of the more interesting examples is in the Super Monkey Ball franchise. The PAL version of Super Monkey Ball on Gamecube has both a 50Hz mode (PAL50) and an additional 60Hz mode (PAL60) that's accessible at startup by holding B. The physics of collisions remains unchanged between NTSC and PAL, so PAL60 feels normal, but PAL50 gains very slightly less speed from bounces and hits off of edges. By contrast, in Super Monkey Ball 2, the PAL version has updated collision physics that makes PAL50 collisions roughly equivalent to NTSC's 60Hz collisions; however, this different version of the collision physics is also used in PAL60, so PAL60 gains very slightly more speed in the situations where PAL50 lost speed in the original. I believe that Super Monkey Ball Deluxe fully resolves the issue, using different collision calculations between the PAL50 and PAL60 modes.
Man obtains a doctorate as an electronic engineer to improve the world record of Super Mario Bros. by 0.1 seconds.
Time well spent
Ah, so that's why I found it strange that soundtracks from the NES games were slower based on my memories. The OSTs on TH-cam made them sound "faster", so I dropped it to 0.9x speed and found that this was the tempo I have been hearing as a child, which explains the "5/6" speed they were running at.
Wow, I had no idea that SMB1 speedrunning had such a history and dedicated community. And now after watching this video, I want to catch up to it
I hope you've got a free month or so, there are volumes of information. The detail on world 4-2 alone could take up an evening.
So if anyone was wondering why values are always modified as a function of DeltaTime in game programming, this is why. To program in real seconds instead of processor cycles as a unit of time, and thus make games framerate-independent. So much less work, and you just have to keep track of the total runtime and keep working out the time elapsed during every frame.
Sponsored by NETFLIX? bro nice iob
2:28 That reminds me how when I first saw the game, I was so little I didn't know how to read it correctly, so I pronounced it "super maREE-o bross". I didn't know what a "bross" was, but I knew this one was pretty super. Then for a brief time I thought it was about Marie's bras.
Me that used SECAM: what about me?
One of the advantages of pal game consoles/computers is, that because the lower framerate meant more time to render, this resulted in european amiga games not being able to run in america
Ummm ACKSHUALLY it is Super Mario BROTHERS, not "bros"
if you grew up in the 80's like me you would know this
"Bros." is an abbreviation. See the PERIOD?? kids these
days try to "speedrun", ha, back in my day that was called PLAYING. ;)
I had no idea about the PAL version jumping higher. That would explain why I could jump the flagpole in 1-1 without the turtle glitch. I have seen videos explaining it isn't possible, but I know I did it. Finally an explanation :)
you can't jump over it in 1-1 without glitches or cheats. Can't even clear it in 2-1 afaik and the platform to jump from is 2 blocks higher
@@Kosmicd12 I have heard that so often, but I have a memory of consistently doing it on my PAL NES when I was a kid. I wish I had the console to try and repeat it since everything I hear says it is impossible
I just discovered this, but not only is PAL better for Any%, it really soars in Warpless. You calculated that the flagpole glitch saves 2.8 seconds... per flagpole. That means that if you add up all 24 levels that end with a Flagpole in a Warpless run, it'll potentially save as much as 67 seconds! Even if it doesn't save that many and make sub 18 possible, that trick alone would still catapult PAL far ahead of NTSC even just for making it a lot easier.
This title reminds me of CS surf. Some of the best routes are slower visually. But they give you more units due to your smoothness. So you wind up benefiting from not taking crazy tight angles. Has nothing to do with this game or game mechanics. But its pretty fascinating how we all achieve top speed in speedrunning.