periwinkle
[TAS] (Obsolete) Super Mario Bros. coinless in 4:59.407 (4:56.129 RTA timing)
updated
I actually got this run offline a month ago (2024-09-05) but didn't immediately record and upload it because I thought I might try to beat it again later down the road. Well...I haven't gotten around to doing that yet, so I figured I may as well upload this one anyways in the meantime.
This run was done on the freeware version; the equivalent time in Cave Story+ (which runs at 60 FPS instead of 50) would be 1'44"1.
This was almost a perfect run for me until I messed up the boost-jump after the dog and lost 0.5s. I briefly considered continuing to grind this category until I got a similar run that fixes that mistake, but certain aspects of this run are so inconsistent (for me currently) that I feel like it would take me a long time to get another run on this pace again. Someday I'll probably still come back to fix it...but for now I want to go back to running Best Ending.
Twitch VOD with commentary: twitch.tv/videos/2175808838
#cavestory #洞窟物語
I was rewatching my (now former) Boss Attack PB and thought "Wow, this looks really sloppy", so I decided to clean it up a bit.
This run is pretty decent aside from the XP situation before Toroko, the missed 1-cycle on Puu Black, the movement hiccup heading to Undead Core, and then losing ~5 seconds on phase 2 Ballos because I thought I could skip the refill station before Heavy Press (because I had full health) but totally forgot that I needed a missile refill too, so I ended up having to not use missiles on Ballos' second phase. (Also I messed up the Blade strat for that which cost me even more time.)
I'd imagine that 5'2x" is probably possible for me with a perfect run...but we'll see if (when) I ever decide to try for that.
Twitch VOD with commentary: twitch.tv/videos/2163060732
Cave Story+ time attack PB playlist: youtube.com/playlist?list=PLeKYCW86fVk2dI8dIZcXTp9VySj56eZAj
New Best Ending PB! It's been a while since the last one (over 1 year and 8 months, to be precise), and a lot of strats have improved since then. I can still take this down to at least a 54:3x, so follow me at twitch.tv/periwinkle9 if you want to see the grind for that live.
Also, I'm very disappointed that my "Best Pace Ever" component wasn't working during this run. :(
Twitch VOD with commentary: twitch.tv/videos/2106232095
#cavestory #speedrun #洞窟物語
Good run, minor time losses here and there (Lab M drops were also pretty bad) but I'm quite happy with this for the time being.
Twitch VOD with commentary: twitch.tv/videos/2091822849
This was supposed to be a Best Ending run until I failed to grab the Tow Rope in time. It's very sloppy so I wouldn't be surprised if 29:xx is possible, but this category is very low on my current priority list so it'll be a while before I get around to actually grinding for such a time.
Twitch VOD with commentary: twitch.tv/videos/2058195201
• The framerate of the game on PC was discovered to not actually be 60 FPS. It's more precisely 1/0.0166 ≈ 60.24 FPS, which means the old TAS which was done at 60 FPS skipped a few frames and the final time wasn't calculated precisely.
• The old TAS missed a couple of instances where holding Z/X sped up the cutscene.
• I used a brute-forcer to optimize the Start Point movement, so I'm much more confident now that there are no more improvements to that room.
• This video is now uploaded at a higher resolution, which means YouTube should give you the option to play it at 60 FPS whereas the old video is limited to 30 FPS because I didn't realize at the time that videos needed to be at least 720p to have a 60 FPS option. Also, I did away with the holiday graphics, because yeah.
The TAS timing of 13.164s in the title is calculated from opening the game to the last input (pressing jump towards the spike), so it's pretty meaningless (IMO) but I included it anyways because that's how TASes are typically timed, as well as for consistency with my Mario TASes which also always include the TAS timing in the title. This movie also assumes the existence of a blank save file with the language set to Japanese; I considered starting from a clean slate which would include the time spent changing the language setting, but TASVideos guidelines permit either way so I went with the faster option. (Not that I'm planning on submitting this to TASVideos anyways.)
The RTA timing of 10.973s follows the speedrun.com category rules which begin timing from the first textbox frame and end on the frame that the player explodes when touching the spike.
Very sloppy run throughout, but mid-Last Cave through the end were pretty okay and carried the run.
Twitch VOD with commentary: twitch.tv/videos/1979288079
Here's another thing that can happen if you go too fast. Though, I think at least in this case, I could have avoided death by shooting the missiles earlier...? There should be some point, though, at which I'm forced to shoot the missiles later in order to hit the Polish at all, but by shooting later it won't die in time before I run into it.
Anyways, it's a slow day today and this idea randomly popped into my head so I made it for some reason. I have nothing further to say on this matter. :P
Sources:
youtu.be/cqCSJyFOaaw
youtu.be/WIKKyrGGaDk
#meme #gdq #リコリコed #リコリコ
The first big test of my TRUE abilities now that I have a keyboard adapter and therefore can now play on an input device that I'm more accustomed to. 34 soon!
Twitch VOD with commentary: twitch.tv/videos/1938498756
Would've been a 34 if I hadn't failed Core quick-kill. We'll get 'em next time...
Twitch VOD with commentary: twitch.tv/videos/1938498757
I got (won) a keyboard.gg adapter for Switch recently, so I decided to do some runs of various CS+ Switch categories to give it a whirl. Expect more Switch runs in the future. :)
This run is still quite improvable, even without implementing 3 HP Chako skip. I'll do more hard mode runs eventually...not sure when that'll be exactly, though.
Twitch VOD with commentary: twitch.tv/videos/1921491154
I got (won) a keyboard.gg adapter for Switch recently, so I decided to do some runs of various CS+ Switch categories to give it a whirl. Expect more Switch runs in the future. :)
This run is super sloppy with a lot of mistakes, but my main goal was to be slightly better than 8th place on the Switch leaderboard (which was my standing before this attempt session). I achieved that with this run, so it's all good. :)
Twitch VOD with my commentary: twitch.tv/videos/1921491155
Time attack challenge PB playlist: youtube.com/playlist?list=PLeKYCW86fVk2dI8dIZcXTp9VySj56eZAj
I got (won) a keyboard.gg adapter for Switch recently, so I decided to do some runs of various CS+ Switch categories to give it a whirl. Expect more Switch runs in the future. :)
5'3x" should be relatively doable with this route, but I'm not sure if I want to go for it just yet. For now I'm just here to reclaim my top spot on the Switch leaderboard (sorry Jayden) :P
Note that this is not the fastest known Boss Attack run across all Cave Story+ versions; for that, see youtu.be/PB8IHG2IkSc .
Twitch VOD with commentary: twitch.tv/videos/1921491153
Cave Story+ time attack PB playlist: youtube.com/playlist?list=PLeKYCW86fVk2dI8dIZcXTp9VySj56eZAj
IT'S OVER
Twitch VOD with commentary: twitch.tv/videos/1902352644
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
(This run was literally perfect except for getting 4 cycles on Balfrog, which loses 5 seconds. Grr...)
Twitch VOD with commentary: twitch.tv/videos/1900613084
I skipped another Life Capsule but only saved 1 second in the end :/
Twitch VOD with commentary: twitch.tv/videos/1863700564
This run was done on the freeware version; the equivalent time in Cave Story+ (which runs at 60 FPS instead of 50) would be 1'46"3.
Kind of a sloppy run, but I don't think cleaning it up would result in getting a 2'06" so I think I'll just leave it at this for the time being.
The bottom left corner of the video was supposed to have an input display, but it must've been minimized or something because OBS didn't pick it up. Sorry!
Twitch VOD with commentary: twitch.tv/videos/1838659645
#cavestory #洞窟物語
Trying to get sub-18...just 4 seconds away now...follow the journey at twitch.tv/periwinkle9 if you want to see THE RUN happen LIVE...
Twitch VOD with commentary: twitch.tv/videos/1835220275
Now, most of the timesave here is actually just at the very end (bottom path turns out to be faster than middle path), but even before the path split you can see that the video on the right is slightly ahead of the current TAS. I estimate the time difference based purely on the fan boost approach to be around 6 frames in favor of the RTA route, give or take a couple.
(Here's a comparison of just that: youtu.be/MeeMgYvc-L4)
#cavestory #tas
On Pixel's birthday (April 29), Cave Story+ has an Easter egg that replaces the Cave Story main theme (i.e. the Plantation theme) with the main theme of Ikachan. However, in the Switch port of CS+, Nicalis messed up and made it so every track is the Ikachan theme, even the "silence" track. I thought this was super funny so I decided to try to set a speedrun PB with this, just for the lolz. Here's the result!
Twitch VOD with commentary: twitch.tv/videos/1807291442
Up to now, I've always played on the Humble Bundle version of the game, but recently I decided to finally redeem the Steam key that came with it, and so I figured it'd be fun to speedrun getting all of the achievements!
I had a route planned out beforehand, but I wasn't quite sure whether it would work out as planned. (For example, I wasn't sure whether the Mimiga Mask ending would count for both that achievement and the normal achievement for beating the game.) As you'll see in this run, I did end up having to make some on-the-fly route changes, a couple of which were maybe not the best decision in terms of speed. Stick around for the ending plot twist...
I've lightly edited the beginning of the video to trim out some of the extra fluff and empty space before the start of the run. Here's the unedited Twitch highlight: twitch.tv/videos/1736763611
Some timestamps just to see if the chapters feature works:
0:00 Intro
1:43 Easy mode run
1:00:46 Normal mode run
1:54:52 Hard mode run
3:26:46 Wind Fortress
3:31:19 Sanctuary Time Attack
3:35:00 Boss Attack
3:43:15 lol
I was only going for a 1'48", but I'll take a 1'47" any day :D
FAQ:
Q: What's the point of unequipping and re-equipping the Booster?
A: Unequipping the Booster allows you to preserve the speed from the boost. Otherwise, if you just release the button to stop boosting (or run out of fuel), you'll lose all of your speed. Note that the in-game timer doesn't count up while in the inventory, and so this technique can be done without any time penalty.
If you don't like the use of this technique, I also have a 1'51"5 that doesn't pause at all except to use the Life Pot: youtu.be/M6ZixKELfZk
Q: I saw that previous records used the Nemesis; is Blade faster?
A: It is now. :P
Nemesis was used previously because it allows for a phase 3 "ultra" cycle that starts the fight much closer to the left side of the arena, saving about 2-3 seconds over the usual strat that starts him all the way on the right (see youtu.be/vD8q9y42ifw?t=185). But sometime last year, I discovered that an equally fast (and actually even slightly faster) phase 3 ultra was possible with the Blade. Moreover, Blade is better than Nemesis for phase 1.
If you're not going for a world record-level time, though, I would recommend using either Spur/Blade or Snake/Nemesis, as this will make it easier to activate the last two Deletes at the same time (by using the Spur or Nemesis), whereas with the Snake/Blade combo you have to use missiles for them, which is quite precise and probably more difficult than I would bother learning initially.
Q: Do you use a turbo button?
A: No, I don't.
Q: Is this category dead?
A: No, of course not! There's always going to be more time to save. In particular, there's a decent chunk of time that can be saved in B2 (by doing more Booster unequips), and the movement in both B1 and B3 can be tightened up a bit in various places. For Heavy Press, I swapped to Blade near the end of the fight to conserve some missiles, but if you got luckier with missile drops from enemies then you could spam missiles for the entire fight and probably save a few tenths there. Then, of course, the Ballos fight can always be mashed faster...
All that being said, I'm quite happy with this time so I don't have any plans to return to this category anytime soon. Maybe if a new strat is found...
Twitch VOD with commentary: twitch.tv/videos/1740229808
Here's my best time with pauses (1'47"9): youtu.be/BLuOhNW4wJ4
Sorry, Altair...I was only trying for 1'53", but I got 1'51" instead...
Twitch VOD with commentary: twitch.tv/videos/1737467448
Watching this run back, it was actually kind of sloppy in several places...there were also a few damage-boosts that are common in high-level Hell runs that I didn't even do. Either way, I'm happy that I finally have a sub-2 of my own. I wonder what kind of time I could get with the Snake?
Twitch VOD with commentary: twitch.tv/videos/1728578143
The time trial mode of this mod is unlocked after completing the normal quiz and features only the "questions" that have non-trivial execution (i.e., more than just "pick the correct answer").
The spikes are _supposed_ to hurt you, but with subpixel-perfect precision (1/512th pixel) you can slip in between them without taking damage. (Don't try it at home, you'll never get it.)
I deliberately answered Q45 incorrectly because it would have been much slower to get the correct answer. Thankfully, it's not required to complete all of the questions perfectly in order to finish the time trial. :P
Fun fact: I realized after making this TAS in Hourglass and then trying to encode the video that there were some...issues with using Hourglass for this (mostly with the weird framerate of the game, and that I had mistakenly specified the wrong framerate for the movie). So I had to make a Linux port so I could dump the video through libTAS instead.
btw Enlight, wtf is that MOD command, it makes no sense-
Although maybe you can still make it work if you mash fast enough...?
Watch my continued suffering at twitch.tv/periwinkle9
This is not the Cave Story+ Boss Attack challenge, but rather a mod of the original freeware game by SeriousFace (which did later become the basis of the CS+ Boss Attack challenge): cavestory.org/forums/threads/boss-rush-again.336
This beats my time from almost 3 years ago (youtu.be/56vZ0JoMmNk) by over 40 seconds, thanks to newer and improved strats on almost every boss. In particular, better use of the Spur allows for a cycle skip on Monster X, Core, and Undead Core; the Sisters and Ballos fights have also been vastly improved since the previous run as well.
There's definitely still much room for improvement, but I'm pretty happy with this run for now. (In particular, a different missile routing could allow for using missiles on the Doctor, which would save a bunch of time)
Twitch VOD with commentary: twitch.tv/videos/1658518006
Someone sent me this hack a few months ago asking for help TASing a few levels. Afterwards, I ended up redoing the entire thing on my own, but never shared the movie file until now. I couldn't find their TAS uploaded anywhere, so I don't have a comparison for how much time I saved or lost compared to their TAS. Hopefully I didn't lose any time anywhere...
(It's very possible that there's an entirely different route that I never knew existed.)
FM2 download: http://dehacked.2y.net/microstorage.php/info/714028546/Air_2.fm2
So, let me tell you a little bit about that mushroom bounce...
- It's actually more like a "mushroom jump", in that I land on the mushroom's head and jump off of it, as opposed to simply using the upward boost from taking damage. The normal jump height is higher than the height from a damage-boost, and allows me to make it directly onto the higher ledge, whereas otherwise I would have to land on the lower edge and do an extra jump to get onto the block, which ends up being about 5-6 frames slower.
- I used a brute-force program to find the best RNG state that would allow the mushroom to walk off as early as possible. After that, I used another brute-force tool to help me figure out which frames to shoot the chinfish so that I get that RNG. What you see here is not the theoretically fastest solution, but it's the best that I was able to manipulate without wasting tons of time (I still had to waste 2 frames in the end). There may still be a faster solution that I wasn't able to find.
- The weapon shots after entering the Cemetery are just to pass the time. They don't affect the RNG at all. (The one thing in that room that is necessary to complete the RNG manipulation is hitting my head on the ceiling, which has to be done on that particular frame that you see in the video.)
Combined with the previous 9-frame First Cave improvement (youtu.be/Q9cMBbEi7l4), which is actually now 11 frames (although I didn't post the video because the change is very minor), this means that I am now 279 frames (5.58s) ahead of nitsuja's TAS heading into Egg Corridor.
It's not quite the low/mid :3x that I was hoping for, but it's still a pretty okay run...
Twitch VOD with commentary: twitch.tv/videos/1593906814
Decided to forgo skipping the Egg Corridor Life Capsule so that I could get more runs past Grasstown. I'll start skipping it again from now on and go for low/mid 38:3x next.
Twitch VOD with commentary: twitch.tv/videos/1582980535
I've been wondering whether I've been invisibly losing time due to slower-than-usual text mashing, so I setup my autosplitter to track the "in-game time" of how long is spent in control of the player—that way I could see how much time I'm losing or gaining due to differences in text mashing speed. (For example, if I were to save 1 second in-game time but the real-time split shows me losing 2 seconds, then I would know that I had played that segment 1 second faster but then lost 3 seconds to bad text mashing.)
I had intended to do a bunch of runs to gather some data, but instead, my first test run ended up being a PB, so now I don't have enough data to draw any meaningful conclusions. What do I do now?!
Twitch VOD with commentary: twitch.tv/videos/1574539523
It's still hard to believe that this kind of time could ever be achieved by human hands...I still remember when, not even that long ago, the record was magmapeach's 56:32 and the consensus was that 55:xx was "possible" but "wouldn't be fun to go for". Then 55 happened and I thought "perhaps this could be the last minute-barrier for this category". But now...how are we looking at a 54:xx?!? What times we live in...
The splits that I was comparing against in this run are not my PB splits exactly, but rather my "Balanced PB". Balanced PB is a LiveSplit feature that tries to make the splits more pleasant to run against by taking the mistakes made in the PB and spreading out that time across all of the other segments, in theory easing the pressure of running against bad beginning splits followed by extremely good late-game splits (as an example). My actual PB splits comparison can be seen after the run when I switch back to them (around 54:57 in the video).
Twitch VOD with commentary: twitch.tv/videos/1546586043
(If I put Japanese text here, will people be able to find this video by searching for "洞窟物語RTA"?)
In summary: The left side strat sucks, don't do it. Also, skipping the damage-boost is *technically* faster, but in practice is difficult to execute in a way that actually saves time. (I spent about 20 minutes getting an attempt that fast that also simultaneously got enough XP for Level 2 missiles. I almost uploaded a video featuring a very similar-looking segment that was slower than the traditional strat, but decided to re-record it only because it didn't actually get enough XP and I wanted to be a perfectionist about it. :P )
SIMPLY GOLD HELL
(but also don't fail Waterway skip twice in the future)
For those curious, I'm estimating that Hell to be around a 2'25"6 on the 290 Counter (equivalent to a 2'01"4 if this were CS+).
Twitch VOD with commentary: twitch.tv/videos/1533853793
Just another dead Best Ending run...except this time, it died not because I failed Core quick-kill, but because I failed to grab the Tow Rope in time afterwards. :P
Lost a lot of time in Egg Corridor? and Plantation so I'll have to come back at some point to clean this up for a 48:4x (or :3x?).
Twitch VOD with commentary: twitch.tv/videos/1497555222
Pretty sloppy run but it still achieves a goal I had of getting a sub-6 time with the Machine Gun. (Is this the first known Machine Gun sub-6?)
The Spur route is probably still faster for RTA, but Machine Gun has a couple of advantages...that I didn't really effectively utilize in this run. Oh well :P
Twitch VOD with commentary: twitch.tv/videos/1315254568
Cave Story+ time attack PB playlist: youtube.com/playlist?list=PLeKYCW86fVk2dI8dIZcXTp9VySj56eZAj
This challenge is officially dead. Not in the sense that the time is unimprovable (it most certainly still is), but rather that skipping basically 2/3rds of the challenge completely ruins any entertainment value this challenge may have had, from both the player's and the viewers' perspective. Sorry if you clicked on this video expecting any semblance of Professional Gameplay™. (But also, it was inevitable that it would eventually come to this.)
PB progression playlist: youtube.com/playlist?list=PLeKYCW86fVk27qUz8wpoGRWuc8ljWMXhP
Cave Story+ time attack PB playlist: youtube.com/playlist?list=PLeKYCW86fVk2dI8dIZcXTp9VySj56eZAj
This is a TAS of the Machine Gun Challenge in Cave Story+, made using libTAS with the Linux port of the game. It aims for fastest in-game time (290 Counter time) completion, and thus pauses in ways that saves in-game time but not real time. However, it does NOT use the Booster unequip technique (unequipping the Booster mid-boost to conserve the speed). Here's why you don't want to watch a TAS that does use Booster unequips: youtu.be/nBrejdxk6A0
Prefer the original soundtrack? Here is an encode with that music setting: youtu.be/WNKS4HC8mXk
Q&A:
• Why the goal of "no Booster unequips", rather than "no pauses" or "minimize real time"?
I wanted to strike a balance between optimizing for in-game time (which is what most people care about) versus making the video actually watchable (see video linked in first paragraph above). Now, pausing *does* actually save real time in several places, so eliminating all pauses would make the run *slower* in every way. So, given that a TAS optimized for real time would still use pauses, I figured I'd go one step further and pause everywhere where it saves in-game time, but just not to unequip.
• So, what do the pauses do, exactly?
I go more into depth about this in the description of my Wind Fortress TAS (youtu.be/V6IWKvAdRRk), but in short, most pauses are used to release the jump button during the pause, in order to do things like first-frame boost jumps or switching boost directions without stopping the boost. There's also a couple of instances where I pause to hold jump for lower gravity without starting a boost.
• How much (in-game) time could be saved by incorporating Booster unequips?
A couple of seconds, I think. (At the expense of maybe 20 minutes of extra real time. Yes, I said minutes, not seconds.)
Stay tuned...
This run was done on the freeware version; the equivalent time in Cave Story+ (which runs at 60 FPS instead of 50) would be 1'47"7.
B2 had a bit of an oopsie near the end, but otherwise the run was pretty clean. 2'08" is definitely possible by cleaning that up, but I'm satisfied enough with this for the time being.
Twitch VOD with commentary: twitch.tv/videos/1466088189
Alternate title: "nitsuja was right all along"
Compared to my previous TAS (youtu.be/6L8xFwMJJEg):
+4f in Start Point (intentional delay for better bat patterns in First Cave)
-5f on the turnaround after the first bat
+Heart drop for Egg Corridor :D
Compared to nitsuja's TAS (tasvideos.org/1917M):
-0f in Start Point
-9f in First Cave.
The turnaround is still a bit sketchy so I might have to mess around with that a little more.
An 8-frame improvement to the beginning of nitsuja's Cave Story Best Ending TAS (tasvideos.org/1917M). The breakdown of those time saves is as follows:
• –4f in Start Point by not delaying going through the door (nitsuja stalled to get better bat patterns in First Cave)
• +3f lost to the turnaround after the first bat (which I had to do differently due to the bat being in a different position)
• –7f throughout the rest of it from slightly better speed manips.
This was made using libTAS on CSE2-portable compiled for Linux. The loading frames at the start are different, but in theory, the rest of the gameplay should sync in Hourglass (I should probably double-check that).
This 40s or so of gameplay took almost 4 and half hours to TAS (and that's not even including the time spent writing a brute-forcer to optimize Start Point). Imagine how long it would take me to do the rest of the game... (In other words, don't expect a full TAS from me anytime within the next few years.)
(For those curious, the Start Point brute-forcer did not find a timesave over what was already known. The closest it got was about 1/3 of a pixel away from saving a frame.)
Also, I will have to go back at some point to manipulate a heart drop from one of the bats. Otherwise I won't have any health left for damage-boosting in Egg Corridor....
Individual level notes:
1-2: Still 2 frames away from saving a framerule. It's so close, yet so far...
4-1: I finish the level with 8 frames to spare in the framerule, but it's possible to finish with 9 frames remaining by delaying 1 frame to grab the flag on the first frame of 340 instead of 341. This would save a frame overall because you'd lose 1 frame slowing down to grab the flag later, but then gain that frame back by the timer being 1 less, and then gain another frame from not getting fireworks. (Yes, 1 firework does technically lose 1 frame.)
I opted to get the firework anyways for entertainment purposes, though.
4-2: A better subspeed on the movement up to the vine allows for getting speed on the vine platform earlier. That means something something and then you can grab the vine a pixel higher, saving a frame on the climb. I also saved a frame in the warp zone by being a better TASer than I was 3 years ago. :)
In terms of pellsson/Sockfolder ROM scores, this level would be 4-9 to 8-D (11 frames to spare).
8-2: The bullet shot I get at the end is fast enough for the TAS 8-2 framerule. Too bad it's a glitchless TAS...
8-3: I kick a koopa shell and it goes straight through a hammer bro without killing it. Nice glitch in a glitchless TAS :D
Also, apparently I hadn't manipulated the judges properly in my last glitchless TAS, so this TAS now finishes with 1 extra frame to spare in this level (8 frames compared to 7 in the last TAS).
8-4: B 9 C 1+2 (45.458s). The frame save in room 2 comes from faceplanting harder into the block, which makes you get pushed out of the block to the left much more than you otherwise would, giving you more room to accelerate and get up to speed quicker. You do land a frame later from that jump than the previous strat (which is why I didn't consider that possibility; shoutouts to Kriller37 for having a bigger brain than me), but I guess the extra wiggle room combined with doing the running speed jump strat (which didn't previously save time) more than makes up for it.
Total rerecords: 8888 (nice)
FM2 movie download: http://dehacked.2y.net/microstorage.php/info/726294582/glitchless_v6.fm2
This run was done on the freeware version; the equivalent time in Cave Story+ (which runs at 60 FPS instead of 50) would be 1'53"3.
The last time I sat down to try to improve my fastest Hell time was almost 2½ years ago (youtu.be/IsifYhofFZk), so I figured I'd take some time to set a new PB. I wasn't comfortable with Snake strats, so I decided to go with my good ol' trusty Spur for these runs. Here's the result of that effort!
To my knowledge, this is the fastest recorded freeware time (as of this writing) that uses the Spur instead of the more-optimal Snake. The previous fastest Spur time (that I know of) is PPU's 2'19"6 from 2013: nicovideo.jp/watch/sm20332815 .
I'm starting to learn Snake strats now so let's see if I can eventually beat the WR of 2'13"8 (nicovideo.jp/watch/sm21443486). :)
Twitch VOD with commentary: twitch.tv/videos/1447443169