a very beautiful 2017 platformer game about utilising conservation of angular momentum to make a slug-spider-octopus hybrid reach and skeletonise rabbits as efficiently as possible
unfortunately, it's iOS-only, was a commercial failure and (following an iOS update breaking compatibility) was pulled from the App Store. A later version since fixed it, but it hasn't returned yet.
however, it is a game that was designed to be speedran, and since competitive speedrunning started in about 2020 we have been very busy!
modern iPads seem to be the ideal platform for high-level speedrunning, since when the ocmo is displaced vertically from the screen's centre, there is more margin for precise vertical aiming downward at the bottom or upward at the top
however, i have 1 WR and 2 former WRs and am an iPhone-only player!
the limitations of iPhones can be mitigated by playing the game in guided access mode to prevent accidentally catching the app switcher/notification centre triggers with stray fingers mid-run, but note that the touch bar still leaves a 'dead zone' where it's meant to be
i play the game with two fingers (using one to grapple (usually keeping it held down for the entire duration) and another to detach, removing the interval between lifting the contract finger and pressing again to detach), though SXY3 (who is evidently much better than me) does not; theoretically, this implies their real-time runs (which are most present WRs) are all suboptimal
currently only SXY3 has the ability to produce TAS runs, via an apparent bug in their iPad's iOS version which doesn't blur the background when the screen mirroring dialogue is entered, and allows for play at ≈1/17th speed
as mentioned in
i used to suspect them of cheating but trust them now, having seen their handcam footage
attempting to document all levels; currently contains WR speedruns posted to the Discord server, as well as the complete videographies of Cyraph, SXY3, 2h40m, ssssyzy and MY牧毅
note that to view videos uploaded to Discord you must be a member of the server; invite permalink
hover over any run to see its author; when one is uploaded to multiple places, the second one is linked for redundancy with an initial (D for Discord, B for Bilibili)
in cases where runs were uploaded specifying the date of achievement, that is used as the official time by this page; in fact, its creation has led SXY3 to realise there were a few they'd forgotten to upload, sometimes even with a >4-year gap
i know of speedrunners in the anglosphere and China, but nowhere else currently! it appears no such videos were uploaded to niconico
the historical record we do have is large enough (and hidden away in enough obscure places) that this page will almost certainly never be complete
i have also created a speedrun.com page, which is more complete than this in some ways (leaderboards runs that were not the WR when published; here they're included but commented out in the HTML unless they imply unpublished WRs in between), while less in others (does not separate runs on alternate/challenge routes or include implied WRs, which are documented here in magenta)
WR history: before 3.69 2020-01-22 3.68 2020-07-12 3.65 2021-01-25 3.62 2026-01-18 3.59 2026-01-25
WR history: before 6.03 2020-01-22 6.26 2020-07-01 6.07 2020-07-19 6.02 2020-11-28 5.97 2021-08-13
WR history: before 1.99 2020-01-22 2.01 2020-08-26 2.02 between 1.98 2020-08-26 2.01 2026-01-20 1.94
TAS history: 2023-07-25 1.94; the TAS has been met in real time but after many many slowed-down attempts 1.94 has not been subseded
the dummy has been reached in 2.89 on 2025-02-09
WR history: before 5.45 2020-01-22 5.79 2020-08-05 5.41 2021-01-25 5.17 2021-02-01 5.00 2021-08-13
WR history: before 4.74 2020-01-22 4.81 2020-08-25 4.77 2020-08-31 4.73 2020-09-03 4.71
WR history: before 5.29 2020-01-22 5.31 2020-03-22
WR history: before 7.33 2020-01-22 7.66 2020-03-11 7.33 2023-07-21 7.21 (
on 2023-07-21, SXY3 also uploaded a non-WR 7.25 with a different route
WR history: before 7.40 2020-01-22 7.69 2020-08-05 7.69 2020-08-12 7.55 2021-02-13 7.39,7.38 2022-01-13 7.35 2026-01-12 7.33,7.31 2026-01-19 7.29
dummy WR history: 2022-02-17 8.38
WR history: before 7.72 2020-01-22 7.91 2020-03-11 7.70 2020-05-24 7.63 2020-06-02 7.53 2024-12-28 7.31,7.29
upper route history: 2021-08-22 7.74
WR history: before 3.42 2020-01-22 3.57 2020-03-21 3.30 (
SXY3 also shared
WR history: 2020-01-22 2.93 between 2.80 2020-04-18 2.93 between 2.79 2020-07-15 2.80 2021-01-02 2.76 2022-02-03
WR history: 2020-01-22 13.78 2020-04-24
WR history: 2019-03-20 8.42 between 8.21 2020-03-11 8.25 between 8.18 2020-06-17 8.27 2020-08-14 8.08 2020-08-18 7.93 2020-12-06 7.86 2026-01-29 7.72
WR history: 2020-03-23
WR history: before 5.76 2020-01-22 5.87 2020-08-05 6.36 2020-10-06 5.66
WR history: 2020-01-22 3.44 2020-03-23
WR history: 2020-03-11 8.67 2020-03-23
WR history: 2019-03-20 10.80 2020-05-21
WR history: 2020-01-22 6.03 2020-03-11 5.86 2020-03-22 5.66 (
the shortcut (that all WRs use)
Joe And I guess Crack was not meant for speedrunners to complete it that way
Salty Crack was made on purpose. I wanted it to seem impossible though
WR history: before 10.69 2020-01-22 11.58 between 9.94 2020-06-10 11.05 2020-07-05 9.77 2020-08-16 9.66 2020-11-12
WR history: before 4.17 2020-01-22 4.20 2020-04-20 4.16 2020-05-31
dummy WR history: 2023-06-28 5.95
WR history: 2020-01-22 7.69 2020-03-11 7.42 2020-05-19 7.36 2020-07-26 7.09 2020-12-06 7.08 2021-01-25 6.99 2021-08-16 6.95 2026-01-26
dummy WR history: 2022-08-24
WR history: before 9.13 2020-01-22 9.32 2020-03-11 9.12 2020-04-24
WR history: before 12.76 2020-01-22 13.75 2020-03-11 12.33 2020-10-04 12.25 2020-10-08 11.87,11.74 2025-01-26
SXY3 has achieved fast times in a plethora of other routes; their WR was done in what they term route 1.1, see also route 1.2 (11.50), route 2.1.1 (11.73), route 2.1.2 (11.67), route 2.1.3 (11.65), route 2.2 (11.90), route 2.3 (11.70)
WR history: 2020-03-11 8.28 between 8.13 2020-05-08 9.23 2021-01-25 7.86 2025-07-23 7.60
dummy WR history: 2026-01-16 9.43
WR history: 2020-03-11 6.37 between 6.26 2020-05-26 6.84 2021-01-25 6.19 2021-02-02 6.16,6.13,6.12,6.09 2021-02-03 5.97 2022-02-03
WR history: 2020-03-11 6.19 between 6.16 2020-08-05 6.23 2021-02-24 5.28 2022-08-24
WR history: 2020-03-11 10.62 2020-03-21 10.07 2020-03-23
WR history: 2020-02-01 11.51 2020-06-20 11.05 2021-01-07 10.48 2021-03-12 10.35 2021-08-13
the 9.60 was via an entirely new route that ignores the second vertical platform
it is possible to complete the level by only touching objects attached to the inside chain, however only SXY3 has done so, on 2023-05-02 in
honey Lol. I did not know this was even possible.
We could have made the dark world level just that !
SXY3 You can. And it will become the hardest Dark World level.
WR history: 2020-01-22 9.16 between 8.99 2020-04-18 9.66 2020-06-14 8.85 2023-07-03
WR history: earlier 5.62 2020-01-22 5.95 2020-08-09 5.49 2020-09-25 5.46 2023-06-26 5.42
WR history: 2020-01-22 6.98 earlier 6.46 2020-05-21 6.74 2022-01-28
WR history for alternate route (going underneath the vertical slime wall at start): 2020-08-20 6.85 2025-11-03
since it took me so long, i considered my 6.60 very optimised executionwise (and doubt more than .05 of timesave could be achieved), but SXY3's run chains a swing off the first pivot with a swing on the circular drum at the start!
WR history: before 13.43 2020-01-22 13.86 2020-03-11 13.78 between 13.28 2020-05-06 14.18 2020-10-03 13.20 2021-01-27 12.92 2021-01-29 12.86 2022-08-24
the 12.86 and the 12.70 both go underneath the starting platform anticlockwise instead of over it clockwise; SXY3's
WR history: 2020-03-11 11.73 2020-09-05 11.33 2020-10-04 11.26 2021-05-03
SXY3's record on the intended route (bouncing off the wall after going around the S) is 2024-08-09 11.41
WR history: before 12.07 2020-01-22 13.89 2020-03-22
my present WR grabs the very top of the pipe while coming out of it, which is more consistent but suboptimal; replicating it with about the same precision elsewhere but fixing that seems the easiest way to achieve timesave!
the WR without taking the shortcut apparently remained at 11.61 until 2021-08-07
shortcutless TAS history: 2024-06-16
WR history: before 3.83 2020-01-22 3.96 2021-01-17 3.86 2021-01-25 3.82 2021-02-09 3.81 (
WR history: before 5.22 2020-01-22 5.39 between 5.21 2020-05-19 7.36 2020-08-23 5.19 2021-01-09 5.15 2021-01-25 5.14 2022-02-13 5.12 2023-08-12 5.10 2026-01-18 5.07
on 2022-02-07 SXY3 published a demonstration of a possible new route in 5.18; this 'route 2' became the WR on 2026-01-18
TAS history: 2023-08-14 4.97
WR history: before 12.76 2020-01-22 14.86 2020-01-24 12.42 2020-04-22
after the 11.92s, which avoided the mushrooms entirely (a trick which is unfortunately much more annoying to do on iPhone than iPad), 2h40m said
2h40m Apart from this, and the one in Long Route, I haven't discovered any other exploits in the game.
Salty It is very cool. I did not know it was possible
Do you mean that you have not found them first
it seems the 12.42 (which also did this) was unknown to the wider community, since it wasn't posted to the Discord
WR history: 2020-01-22 4.19 between 2020-03-23
TAS history: 2026-02-05 3.72
WR history: before 6.24 2019-03-20 6.27 between 6.03 2020-07-08 6.23 2020-07-20
WR history: 2020-03-11 6.31 2020-03-21
WR history: 2020-03-11 14.60 between 14.49 2020-05-16 15.22 2021-02-01 14.15 2021-02-03 13.98 2021-08-01 13.84 2025-01-31 13.81
WR history: 2020-03-11 7.08 2020-03-22 6.47 (
WR history: 2020-01-22 15.10 some intermediate time 12.49 2020-04-22
my times:
the discovery of a skip at the end seems to be by Cyraph and 2h40m independently, since the former achieved their WR with it first then published it after
the 11.67->11.56->11.51 improvements were both from route changes; the former from landing on the right-facing wall to jump off it, the latter from grabbing the square at the bottom-left instead of top-right during the skip (which is something i'd learned from 2h40m's 11.99 after noticing it broke even with my 11.98 after falling behind; implementing it immediately saved me .17, then i figured SXY3's run must have omitted it for a reason and followed suit for my next two, then reimplemented it and immediately saved .15 again)
under the assumption that it saves .15, with SXY3's execution a future WR could theoretically be about .1 lower still!
after the 11.99,
Salty I was actually laughing out loud. That should not have been possible 😄
2h40m Was this not by design? haha no wonder why it felt really difficult 😅
Salty When we made the level we did not think of the possibility that you could get to the last running platform and skip the middle
note that SXY3's message with the 11.56 contains a 13.01 which is done without the skip, suggesting that it only saves about 1.45 if played optimally; however, such optimised execution of the non-skip route as they did is somewhat harder than the skip!
skipless WR history: 2021-08-16
WR history: 2020-03-11 7.40 2020-04-24
WR history: 2020-05-11 11.48 2021-01-25 10.55 2021-02-15 10.52 2023-04-29
TAS history: 2023-08-20 7.76
WR history: before 3.27 2020-01-22 3.27 2020-06-07 3.17 2020-12-12 3.15 2021-03-28
TAS history: 2.97
WR history: 2020-02-01 14.46 2020-03-11 13.50 2021-01-13
WR history: 2019-03-20 17.89 2020-03-11 16.51 2020-05-12 15.39 2021-08-16
WR history: 2020-02-01 5.75 2020-04-27 5.59 2026-01-16 5.56
WR history climbing the left wall instead: 2026-01-16 5.97
WR history: 2020-03-11 14.53 2020-09-13 13.38 2021-01-25 13.30 2021-01-31 13.13 2023-07-23 12.97
WR history: 2020-03-11 12.59 between 12.26 2020-04-29 16.34s 2020-07-21
assumedly, the .77 timesave was due to a very difficult shortcut at the beginning ()
WR history: 2020-01-22 10.36 2020-06-03
as a challenge run, it is possible to complete the level without touching any surface besides the one you start on (the first wall being reasonably difficult/tedious, the second just slightly slower, but tricky when braking high upward speed); for that, the timeline is 2025-02-10
WR history: before 14.03 2020-01-22 14.22 2021-02-18 2021-03-06 13.89 2021-03-12 12.97
the artist/level architect Salty - Boy held back from setting good times post-release (to avoid demotivating the players) on all levels except this one, upon which they competed with Routalanttu. This means the very early (pre-external-upload) WR history here is better than everywhere else, so the WR runs since with surviving recordings are sparser because they had a harder baseline to surpass
WR history: 2020-01-22 5.89 between 5.70 2020-04-18 6.06 2020-09-05 5.44 2023-05-11
WR history: before 23.00 2020-01-22 23.30 2020-04-24
SXY3 also shared a 22.83 route PB
WR history: before 18.55 2020-01-22 18.58 2020-05-11 17.45 2021-08-24
WR history: before 16.95 2020-01-22 18.41 2020-02-01 16.92 2020-11-25
WR history: before 14.93 2020-01-22 15.10 2021-02-24 12.86 2023-08-01 12.73
note that SXY3 created a LOTAD on 2023-08-05 demonstrating that with an early cycle and going above the last spinner,
WR history: 2020-01-22 33.56 2020-03-11 33.05 2020-05-11 29.62 2021-01-25 29.29 2023-05-11
WR history: 2020-03-11 4.03 2021-01-31 4.00 2023-08-24 3.83 2023-08-25 3.72
WR history: 2020-01-22 15.47 2020-03-11 13.96 2020-03-22 13.82 (
WR history: 2020-01-22 14.26 <2020-04-20
WR history: 2022-01-22 9.52 2020-03-23
WR history: 2020-01-22 8.92 2020-02-01 8.62 2020-03-11 8.41 2020-06-21 8.38 2020-08-29 8.08 2021-02-24 8.04
WR history: before 7.29 2020-01-22 7.12 between 5.97 2020-05-14 6.24 2022-08-24
all known WRs with videos were done by going through the fudge at the start; MY牧毅 also shared 2020-05-14 7.22 from going entirely underneath at the start, and i shared 2021-11-15
WR history: 2019-03-20 10.11 2020-12-04 9.80 2021-01-26 9.67 2025-02-13 9.33
it is possible to go over the revolving door here, albeit tightly, as demonstrated in 2024-08-11
WR history: 2020-01-22 13.75 2020-04-27 13.26 2020-10-03 12.90 2020-10-05 12.76,12.19 2020-11-28 11.94 2021-02-16 11.71 2021-03-06 11.33 2021-08-24
WR history: 2020-01-22 11.24 2020-03-11 10.78 2021-01-07 10.25 2021-02-20 10.22 2024-08-02 10.11
the level has also been completed remarkably quickly without substantially moving the chain; 2020-03-22 12.56 2024-08-09 11.57
WR history: 2020-01-22 26.10 2020-03-11 24.83 2020-03-23
this one is interesting
drone is this one of the intended shortcuts (it feels like it is) or another happy accident like Work Around
Salty It is kind of both [intended and not].
Very often when we were designing a level we spent a lot of time fooling around and finding out. I think originally there was no intention of creating an alternative route. The landing platform was meant as a resting ground. But when there is an opportunity for a secret shortcut we wanted to keep it and tried to make it less obvious with visual guides and demanded more accurate execution for movement. This one is not too difficult to perform but the actual route looks so obvious that not many people ever try the shortcut...
much of this history is lost media!
here is
honey Out of the 500k downloads we know less than ten players who have beaten everything including both devs. I believe around 50 players have beaten all dark world basic levels, but there might be more as we stopped collecting data after the GDBR .
Joe It’s 53 players who triple starred Pull Up so far
Joe 42 in Wheel of Unfortune
Joe 38 in Revolving Doors
Joe Wow I checked most Dark World levels now, least people with 3 star are in Revolving Doors making it the hardest level as I remember saying that before as well when I first started
honey Interesting. I always thought that "Pull Up", "Long and Hard" and "Escalator" were the 3 most difficult level.
It might still be that Escalator is more difficult than Wheel of Unfortune but it gets more attempts as it might be more interesting as the level chances depending the route
Salty What I mean is that if a level is not as interesting it might receive less 3 skull times as people get bored. Interesting level gets more attempts and thus more people would get the three skulls even if the level was harder.
Salty It is hard for neither of us to know what motivates other players to try each level. All I am saying completion is not the only motivation for every player, even in Dark World and that might skew the statistics that revolve around game center times.
Also not all levels behave equally on what percentage of WR is correct time limit for 3 stars/skulls. Long levels ("long and hard") and levels with dynamic variation ("escalator") in my opinion should have more slack in time limits compared to a level where the level is short ( "W", "hight jump") or the level does not allow much variation (wheel of fortune).
The 3 star times are our best guesses on what would be fair time limit but still an interesting challenge.
consider also that (as i hope the historical record above shows) we are standing on the shoulders of giants
i think
I do not know if anyone has finished the game with thubms 😄
which might be fun to try recording myself doing if i ever do full-game runs
excluding unintended shortcuts, which are covered in levels
there are surprisingly few critical bugs affecting the gameplay itself
generally (oversimplified; see The pupil as a measure of emotional arousal and autonomic activation) human pupils dilate in response to happy stimuli and constrict to bad ones.
quite a few games have monocular entities that convey information about their surroundings; Terraria has the suspicious looking tentacle, which has both a pupil and iris, dilating at treasure and constricting at enemies; this was added in 1.3.0.1, which was released on 2015-06-30 (long before Ocmo)
however, ocmo's eye is maximally open by default; the eyelids only become apparent when narrowed while moving quickly (which also causes the pupil to shrink so its centre can be further towards the edge), so pupil dilation is used in lieu of further widening, which i find still effective!
upon approaching its prey, the pupil grows horizontal (like that of a toad), then upon catching it, oscillates wildly through various sizes and eccentricities of ellipse, until it's done feeding and begins to glow, the ocmo inert
the inert state is immune to slime, but in some levels causes the ocmo to fall offscreen to its death. maybe death (trisection) is how ocmos reproduce, like the segmented cacti whose dropoffs can also function as independent cacti
the name has been
honey I have no idea what we were thinking when the decoy bunny sound was made. Soundboys propably made some suggestions and we went with that.
But a fun fact. The name Ocmo which comes from "Octopus monster" is pronounced Okumo in Japanese which can be understood 御蛛 meaning "venerable/royal" spider