redcometchar
Smash Journeyman
Theoretical infinite
Didn't see this anywhere else
Tested on samus with no di
So this is pretty cool, also pretty unfeasible but hey 20xx. It works by using the interupt mechanics of shellshift to shellshift again earlier and then exploiting moonwalks to shorten the distance squirtle travels between consecutive shell shifts. Since shellshift doesnt really scale much with percent, it works well. It abuses hard landing lag as well as hitstun, and requires a frame perfect dash out and moonwalk.
I would like to point out that the opponent can di out at any time with in or down, but away will not work. after about 30 percent, you then can follow all possible di options except up for floaties at around 70.
1 dash
2 dash
3 dash First 8 frames dont matter but there must be at least 8 frames of dash before shellshift can begin
4 dash
5 dash
6 dash
7 dash
8 dash
9 (1) begin shellshift
10 (2)
11 (3)
12 (4)
13 (5)
14 (6)
15 (7) Window for first frame of no input begins
16 (8)
17 (9) hit connects, first frame of hitlag
hitlag
hitlag about 3 frames across the board
hitlag
18 (10) (1) First frame of opponent hitstun (about 21 frames)
19 (11) (2)
20 (12) (3)
21 (13) (4)
22 (14) (5)
23 (15) (6) Window for first frame of no input ends
24 (16) (7) (1) first frame of dash in current trajectory
25 (end) (8) (2)Window for first frame of input backwards begins
26 (9) (3) Window for first frame of input backward ends
27 (10) (4)Input backward is held until next direction
28 (11) (5)
29 (12) (6)
30 (13) (7) Window for first frame of input diagonal forward begins
31 (14) (8) Window for first frame of input diagonal forward ends
32 (1) (15) (9) Shellshift begins
33 (2) (16) (end)
34 (3) (17)
35 (4) (18)
36 (5) (19)
37 (6) (20)
38 (7) (21) Last frame of hitstun
39 (8) (end) (1) First frame of hard landing lag (4 frames)
40 (9) Second shellshift will hit in the next 3 frames, now repeat above procedure
41 (10) (2)
42 (11) (3)
43 (12) (4) last frame of both hitbox and hard landing lag
Slightly unrelated but since I have your attention
By the way there is a lot of misinformation about shellshift floating around, specifically hydropivot in the current version. Im going to clear it in case it wasnt already. It looks like shell shift got nerfed again at some point before 3.6 full
Im using this video for misinformation comparison
https://www.youtube.com/watch?v=LTleUo52zOE a good video to be sure but it didnt line up with my numbers in testing.
I am running a 3.6 mirror though so maybe its that
Frames 15-22 of shellshift can be canceled with pivot but the slide lengths dont change in this window, extra distance is provided by extended time in shellshift.
If direction of original dash is inputted on frames 10 to 14 and the last frame of the input a boosted slide is the result with frame 12 being the longest slide.
Also and shell stall input frames last until frame 9 not frame 6 if only stalling for a few frames.
Here is an updated version of his (daftatt's) graphic
Didn't see this anywhere else
Tested on samus with no di
So this is pretty cool, also pretty unfeasible but hey 20xx. It works by using the interupt mechanics of shellshift to shellshift again earlier and then exploiting moonwalks to shorten the distance squirtle travels between consecutive shell shifts. Since shellshift doesnt really scale much with percent, it works well. It abuses hard landing lag as well as hitstun, and requires a frame perfect dash out and moonwalk.
I would like to point out that the opponent can di out at any time with in or down, but away will not work. after about 30 percent, you then can follow all possible di options except up for floaties at around 70.
1 dash
2 dash
3 dash First 8 frames dont matter but there must be at least 8 frames of dash before shellshift can begin
4 dash
5 dash
6 dash
7 dash
8 dash
9 (1) begin shellshift
10 (2)
11 (3)
12 (4)
13 (5)
14 (6)
15 (7) Window for first frame of no input begins
16 (8)
17 (9) hit connects, first frame of hitlag
hitlag
hitlag about 3 frames across the board
hitlag
18 (10) (1) First frame of opponent hitstun (about 21 frames)
19 (11) (2)
20 (12) (3)
21 (13) (4)
22 (14) (5)
23 (15) (6) Window for first frame of no input ends
24 (16) (7) (1) first frame of dash in current trajectory
25 (end) (8) (2)Window for first frame of input backwards begins
26 (9) (3) Window for first frame of input backward ends
27 (10) (4)Input backward is held until next direction
28 (11) (5)
29 (12) (6)
30 (13) (7) Window for first frame of input diagonal forward begins
31 (14) (8) Window for first frame of input diagonal forward ends
32 (1) (15) (9) Shellshift begins
33 (2) (16) (end)
34 (3) (17)
35 (4) (18)
36 (5) (19)
37 (6) (20)
38 (7) (21) Last frame of hitstun
39 (8) (end) (1) First frame of hard landing lag (4 frames)
40 (9) Second shellshift will hit in the next 3 frames, now repeat above procedure
41 (10) (2)
42 (11) (3)
43 (12) (4) last frame of both hitbox and hard landing lag
Slightly unrelated but since I have your attention
By the way there is a lot of misinformation about shellshift floating around, specifically hydropivot in the current version. Im going to clear it in case it wasnt already. It looks like shell shift got nerfed again at some point before 3.6 full
Im using this video for misinformation comparison
https://www.youtube.com/watch?v=LTleUo52zOE a good video to be sure but it didnt line up with my numbers in testing.
I am running a 3.6 mirror though so maybe its that
Frames 15-22 of shellshift can be canceled with pivot but the slide lengths dont change in this window, extra distance is provided by extended time in shellshift.
If direction of original dash is inputted on frames 10 to 14 and the last frame of the input a boosted slide is the result with frame 12 being the longest slide.
Also and shell stall input frames last until frame 9 not frame 6 if only stalling for a few frames.
Here is an updated version of his (daftatt's) graphic
Last edited: