Speedruns thread

X-SOLDIER

Harbinger O Great Justice
AKA
X
One: In case you're wondering why he shows his console and everything at the end of that, this video will provide context:

Kosmic's pretty great, and his reactions for Speedruns are always so interesting, especially when he hits a run like this one when he's just not really expecting it, and everything just works out. He does an absolutely AWESOME breakdown of the entire run here that's well worth checking out if you've got the time:




X :neo:
 
Only tangentially related to speedrunning, but here is this amazing video about how the JP and NA versions of Super Mario 64 play slower in certain segments compared to PAL.


The fact that Super Mario 64 has been so meticulously de-compiled is just so goddamn awesome.

EDIT: The opinion of a different analyst is that Nintendo did not mess up or forget to add the O2 compiler optimizer unintentionally.
 
Last edited:

X-SOLDIER

Harbinger O Great Justice
AKA
X
Oh man, I wish I knew about the ESA event! I'd totally have had that on. Also, Karl Jobst did a really good breakdown of the Any% WR of OOT, and what makes it important to have it split out a different category.




X :neo:
 

X-SOLDIER

Harbinger O Great Justice
AKA
X

One of those things found out on the Internet that was shared to our work Slack (because this is the software that our company makes). Figured that the folks in this thread would appreciate it and maybe be able to share some other new speed runs since the threads been quiet for a while. :awesomonster:




X :neo:
 

X-SOLDIER

Harbinger O Great Justice
AKA
X
I love this, because it is full of the same sorts of pressures, difficulties, and weird intuitive-counterintuitive-intuitive sort of thinking problems that speedrunning is, but just in a different format that's got the same sort of opportunity for optimizations. While at face value, the concept it sounds boring, because every. single. level. is all about running the clock to the absolute limit, it's actually got a really good balance of tension to keep it interesting for the games where it's possible.




X:neo:
 

It's funny how I've watched so many videos about speedrunning Paper Mario TTYD but I've never actually played the game or watched a normal playthrough of it. :monster:

EDIT: I realize now we have an old thread specifically for tool-assisted speedruns but I'd much rather use this current one for all speedruns (since that's how it has been used anyway for the last couple of years). I vote to have this thread's title changed to "Speedruns thread" and to include a note of this in the opening post.
 
Last edited:

Rydeen

In-KWEH-dible
Something I've always been curious about is the appeal of a TAS over a regular speedrun? Is it just because it makes it less tedious? I've always seen regular speedruns as being better for a record breaking perspective, since it's just a human interfacing with the game and not intermediary code (I'm not throwing shade or anything, I just don't really get it and I'd like to understand it more).
 
Something I've always been curious about is the appeal of a TAS over a regular speedrun?
To me the fundamental joy of tool-assisted speedruns is that they provide a theoretical best possible time based on current knowledge of the game. Like a glimpse into the perfectionist's dream of "the best possible of worlds".

Every human-performed real-time speedrun suffers from the awareness of just how many mistakes were performed in the run, which is why it's so common for world-record holders to not be pleased with their own speedruns: They are just far too aware of how much better it could be if just everything happened exactly the way they want it to. Watching a TAS becomes escapism from the world of human flaws.

Because a TAS is meticulously choreographed there are so many more sick executions to enjoy. There is such beauty in the movement even when it's not performed in real-time. Not impressive in the same way as when a human plays in real-time but still enormously pleasing to look at on an aesthetic level. Creating a TAS can take hundreds of hours so there is definitely a lot of hard work to appreciate as well.


I can't say which I enjoy more: Real-time action or tool-assisted speedruns. Both are enjoyed in similar ways but have nuances that set them apart. Real-time executions are more impressive knowing that it was done within the confines of human limitation and then there's the parasocial and emotional element of seeing a pro speedrunner getting "THE RUN" and reacting in joy to the achievement.


I've been watching Super Metroid speedrunner zoasty a lot lately and it's so wholesome how supportive his wife is. When a really great run ends they often hug (just so it's barely visible on the cam) and you can hear the kisses and laughter and shared joy. Might be considered inappropriate on some level but it definitely adds to that emotional aspect of real-time speedruns.


I'm not throwing shade or anything
I sure hope so since NOBODY TOSSES A DWARF POTATO SHADEMP! :awesome:
 
Last edited:

X-SOLDIER

Harbinger O Great Justice
AKA
X
Something I've always been curious about is the appeal of a TAS over a regular speedrun? Is it just because it makes it less tedious? I've always seen regular speedruns as being better for a record breaking perspective, since it's just a human interfacing with the game and not intermediary code (I'm not throwing shade or anything, I just don't really get it and I'd like to understand it more).

The people putting in the work to make TAS runs are also very frequently the same people closely involved in understanding how a lot of the exploits and other things function in the games being speedrun. This means that they are involved with a lot of work done on discovering glitches and determining whether something is human-executable or not. Additionally, they also do a lot of investigating around things that runners stumble upon accidentally. Oftentimes, once something's discovered, that means that you'll get a bunch of TAS runs to test out parameters on what's happening. That work gets used, so that the community can reach a level of understanding around why particular parts of the game work the way that they do, & see if things are reasonably exploitable for run attempts or not based on what factors lead to success. All of that helps runs, but particularly Any% records to have sudden shifts in how quickly you can beat a game that make big impacts in their respective communities.

Bismuth has a really excellent explanatory video about TAS stuff in general:


The other thing that's neat about TAS is that they can do things that are either literally impossible for people to execute, or that use a degree of constant extreme precision that you could never have from a human player. That type of entertainment is usually appreciated most by the specific game's speedrunning audience because of the understanding of how difficult all of those things are for human players, and they wouldn't always be as impressive for a general audience without a good degree of context. A REALLY great example of that is TASBot playing Super Mario Maker 2 during AGDQ 2020:





X :neo:
 

X-SOLDIER

Harbinger O Great Justice
AKA
X
Related, Kosmic just ran all 18 mainline Mario games in JUST under 24 hours over the weekend. I started watching around SMB2 until I went to sleep, woke up, and watched him finish the run. It was an absolute blast!



X :neo:
 
Four months ago, zoasty achieved a world record time of 40m 55s on Super Metroid any%.

Earlier this week he beat his own PB by five seconds...and now he cut it down with an additional four seconds!

zoasty said:
I think this is the end for the time being.

Sure, MAYBE I could eek out a couple more seconds by grinding for some more months. You could keep saying maybe to that forever though, because the perfect run doesn't exist of this game.

There are so many things (RNG and otherwise) that went really well in this run, and it's very difficult to clutch the ending on this kind of pace. My hands were wanting to shake and grip harder during the escape sequence and I'm really lucky I didn't make any costly mistakes there.
 
Top Bottom