Is there a limit to how fast a game can be speed run? Like in super metroid how much faster could they really do ittbhfam
>>3239010
Yes
>>3239010
you can't break the speed of light so yes
>>3239030
>you can't break the speed of light
>>3239010
Speedrunners are human and thus prone to human error. But there's always a theoretical limit that cannot be broken. SMB for example, you could imagine replacing the levels with blank level with no holes or obstacles except for the ones that make up shortcuts, and then running these blank levels to find the theoretical limit.
>>3239535
Such limit can still be broken via glitching.
>>3239552
The SOTN speedrun is like black magic or something if you've seen it
>>3239535
Even in that case, you fill a blank room up to that point, then use a TAS to find the minimum frames needed to execute the glitch and anything else required.
>>3239010
Imagine a speedrun route that is represented by a TAS-like list of inputs and timings associated with each one. Every game will have an optimal route, defined by such a list, that plays out in a minimum number of frames, making no distinction between glitches and intended strategies. That's just common sense.
However, it is probable that such routes cannot be computed in polynomial time -- the rules governing a game's behavior are complex. Human-brand pattern recognition and "creativity" if you will is still better than an algorithm at exploiting game behavior to invent fast routes.
So the takeaway is that there is a point where a run is optimal because games are deterministic. But it's hard to compute and hard to verify, so ultimately it doesn't matter all that much. We're probably only approaching that point in a few relatively simple games.
>>3239987
>However, it is probable that such routes cannot be computed in polynomial time -- the rules governing a game's behavior are complex. Human-brand pattern recognition and "creativity" if you will is still better than an algorithm at exploiting game behavior to invent fast routes.
People have already shown that SMB and Zelda are capable of producing NP-complete problems. Granted, most levels could prolly be solved perfectly, but in general you get some nasty pathing issues.
Moreover, if you can glitch the game to inject code that runs the ending, arguably you have beaten the game. This can really mess with finding what is truly an optimal solution, and you are treading Halting Problem territory.
>>3239987
>>3241613
This is a great explanation.
The actual limit is a hypothetical "perfect run" of inputs.
The effective limit is the state-of-the-art of TAS routing at the time.
The practical limit is how much of the perfect TAS route can be executed in real time, unless that takes longer for a human than some alternate route to the TAS.
>>3241613
woah dude mind blown
>>3241613
You can go pretty deep down the rabbit hole when you start talking about stuff like arbitrary code execution. There are probably tons of games with undiscovered exploits to execute arbitrary code or at least reach a totally broken state that allows the player to trigger the ending (I'm reminded of the Super Mario Land 2 run that has Mario literally navigate through the game's RAM/ROM, represented as garbage tiles, to change a value that skips to the end).
The fact that many optimal routes might boil down to turning the game inside-out just illustrates how difficult it is to say "well, we found it, you can beat this game in 372 frames and no fewer" or whatever, because it reveals that the system you're dealing with is much much more complex than just guiding a little man to the flagpole at the end of the level or whatever.
>>3241679
>I'm reminded of the Super Mario Land 2 run that has Mario literally navigate through the game's RAM/ROM, represented as garbage tiles, to change a value that skips to the end).
https://www.youtube.com/watch?v=24XHkmOJebA
I forgot about this, shit's awesome.
>>3241742
Video games are so surreal. I love em, bagus and all.
https://www.youtube.com/watch?v=sTJRZTj-kDs&list=PLUCcl7X553EBtCQJmJHqgLwAjjJm7y0MI
>>3239010
Even if you make no mistakes you can only move as fast the game will let you so there is a minimum time limit a game can be beaten
>>3241769
Too bad that game is awful.
>>3239010
I don't know, but i did a tool assisted speedrun of your mom
TIL speedrunning is for faggots.
>>3241813
was it a wrong warp?
>>3241810
It's not as good as Circle of the Moon, but it's better than all the other shit Castlevanias on GBA
>>3242002
CotM is GOAT, but AoS is arguably the best iteration of the metroidvanias in terms of production and design, and was pretty much SotN done right. Poor HoD just sucked.
>>3242035
have you
actually played it though
it fails on the most fundamental requirement of being fun and interesting to play as a game
>>3242057
Off topic:
I've played CotM through 20+ times. Just all around kick ass game, and one of my all time favorites. Mechanics are simple, but the cards add a lot of depth. It's actually a very challenging game, a rarity today. Graphics are good (when backlit, a problem because it was a launch title) and music great. Good replay value with the different game modes.
Played AoS probably 10+. I think AoS has great variety in mechanics and enemies, I think it's got good challenge, the graphics and music are both top notch, the level design is fun (and like CotM does away with the overused two-worlds mechanic), and the over theme of the game is really great. Too bad you don't like it, it's your loss. What didn't you like? Did it feel like a collectathon to you or something?
The only thing HoS got right was the main castle music. Really a chill song that could have worked in any of the other games. The rest was just off, which was a huge let down after CotM; the level design sucked ass, the graphics and music were incredibly bland, the mechanics sucked with the item/book system being all around gay. It really was a train wreck.
On topic:
SM is actually really interesting to see run because the choice of routes and items is more important than execution. You need a minimum number of missles to get through the Zeebites and hurt Mother Brain, so you have a variant of the difficult Traveling Salesman type optimization problem. I always am interested in how people chart their routes and the rational behind choices made, especially when it comes to managing resources and saving.
Actually watching speed runners has made me think "wouldn't it be interesting if in Metroid a missle pack gave you 5 missles, that once used, are gone forever?" You'd have to be really careful with ammo unlike the normal endgame where it takes 2 minutes to unload a fully loaded Samus.
>>3242101
AoS is very very easy and yes, it is a collectathon.
The enemies tend to be very slow and boring, and for all the variety of attacks and souls and equipment, it really doesn't matter all that much what you choose. Every problem is a nail and you could hammer nails with your fists if you wanted to.
I also have a problem with the the level design, which is almost entirely
>series of long hallways filled with monsters to farm
or
>large rectangular area with platforms and also monsters to farm
The best part of the game is either fighting Julius or being Julius, and both are not actually as fun as being Richter.
The game looks really great and the characters are designed real pretty and cool lookin, and all of them have nothing but worthless garbage to talk about and the plot is complete nonsense.
Playing the MMZ games yields a very similar experience except the games are actually challenging and a lot less patronizing in terms of content and design.
>>3242101
>Actually watching speed runners has made me think "wouldn't it be interesting if in Metroid a missle pack gave you 5 missles, that once used, are gone forever?" You'd have to be really careful with ammo unlike the normal endgame where it takes 2 minutes to unload a fully loaded Samus.
there's an entire genre about this anon
>>3242225
That's what I was saying man.
AoS is the one that sucks.
>>3241813
The tools you'd need to speedrun my mom are a shovel and a crowbar.
