isg notes (and bank clip notes)

----
frame perfect shield drop isg:

C-Down (bombs)
2 nothing
R 4 times
B + R
R 6 times
R + A
19 nothing

34 frames total
======
set down isg:

C-Down (bombs)
1 nothing
A
12 R
R + B
R 6 times
R + A
19 nothing

42 frames total
-------------------
setting down a bomb

A
11 nothing

12 frames total
---

with set down bomb isg, you have 9 frames to do things for bank clip, but with shield isg you have 17 frames

setting down the bomb takes 12 frames, which would leave us with 5 frames to do things


-----

backflip (y-velocity pattern 2):

0
-1
0
-1
-2

js (y-velocity pattern 1):

0
-1
-2

make a script that starts at some initial position on the banker, then it performs some permutation of movements (permutations will be some sequence of consistent movements FOLLOWED BY a walking direction; perform the permutations without the walking direction before the savestate), then create a savestate of this position, set 'Sword Active' = 1 for isg, then read the walking direction and walk into the ledge in that direction, then it tests if it has y-velocity pattern 1 or y-velocity pattern 2, then it loads the savestate once it has determined the pattern, then ...... ACTUALLY, instead of making a decision based off of pattern 1 and pattern 2, just record which pattern it is as extra information; what you should really do is record the number of visual frames it takes to reach -2 y-velocity (AFTER DROPPING THE BOMB, DO NOT FORGET TO INCLUDE THIS;;;; ALSO CHECK IF BOMB Y POSITION IS LESS THAN 133) (this is equivalent to telling us when we reach -14 y-velocity, but it is obviously 12 frames faster to test), then make a decision on how long to wait before walking after that

when testing the clip itself, be sure to time it so that you hold R for two visual frames, starting at the start of the visual frame where link has -11 y velocity and then 1 visual frame of no inputs, then advance 1 input frame with no inputs AND RECORD WHETHER OR NOT LINK'S VELOCITY VALUES, ETC. UPDATE AFTER ADVANCING THIS INPUT FRAME, record this information with a True/False ('Midframe Change'), then advance frames to see if link clips. if link does clip, then test the JS (after advancing the single input frame, advance two input frames while holding B + Left to test it IF YOU GET THAT Midframe Change == True, else if Midframe Change == False, just advance an entire visual frame while holding left + B)

Possibly useful to also record the bomb's Y Position each time (after it has settled, of course)

DO NOT FORGET to record HOW MANY FRAMES YOU WAITED BEFORE WALKING because if you determine that you have to wait at least 8 visual frames before walking, then you can do set down bomb isg instead of shield bomb isg; so just record this for each trial

Making Sure Link has a Valid Position and Valid Angle:
Obviously, assert that link's Y Position must be equal to 133 (the height of the banker) before spending time making a savestate (i.e. if link's y position is not equal to 133, then just skip this permutation entirely). Also, assert that link's angle (after walking) is within some angle range (maybe the angle should be roughly 180 degrees to just make sure link isn't facing the wall or something; also when testing how many frames link needs to reach -2 y velocity with isg, make sure to stop trying if it takes more than X frame (X might be 17, but it might also be higher... not entirely sure because of pattern 1/pattern 2 thing confusing me; 25 is safe but look into it more)


Obviously, check if the clip/js succeed using the same criteria (involving y position) as before

---

actually, depending on link's angle, use that to determine the set of possible walking directions maybe? and test each case at the same time and record information for each case as if they are separate


------

Permutations and initial positions/angles to consider?

Initial Positions:
- Target banker + JS onto banker
- Target banker + JS onto banker + Turn Down
- get into corner and do a hold backflip onto banker
- target wall across from bomb shop, use corner, hold backflip onto banker

Can perform set of slashes, including jump slash, but be aware that targeting the banker is a thing and that can mess with angles/turns potentially....... is there a memory address to see whether or not you are targeting something? if so, use that or try hold targeting

adapt "turn" functions to not target... PAY ATTENTION TO CAMERA ANGLE... NOT ANGLE



----------------- Updated idea:
(Note: can add in ESS turns during the waiting times to create more setups *** figure out how these work before implementing this)

Initial Position: Target wall across from bomb shop, get into corner, hold backflip onto banker

Movement_Options = ['Untargeted Horizontal Slash', 'Untargeted Horizontal Slash, Shield', 'Targeted Vertical Slash', 'Targeted Vertical Slash, Untarget ASAP', 'Untargeted Right Slash', 'Untargeted Right Slash, Shield', 'Targeted Forward Thrust', 'Targeted Forward Thrust, Untarget ASAP', 'Nothing']

If 'Nothing' is chosen, then all following options will also be 'Nothing' to reduce the number of redundant permutations.

Perform sequence of consistent movements, get shield isg, set down the bomb, create a new savestate in this process (saves time to do it now instead of earlier), check if bomb y position is <133 AND sword active == 1 (i.e., make sure you both have isg and the bomb is below 133 y position), input Up on control stick (start holding target while setting down the bomb, perhaps) and count how many visual frames it takes to reach -2 y velocity (and thus we know how many frames it takes to reach -11 and -14 y velocity). Use this to determine how many frames to wait before walking forward, then reload the newly created savestate, wait however many frames it determined it should wait, then walk forward, do bank clip (as noted above)


-- idea

for each savestate created (i.e. corresponding to each permutation) test the following: Hold Up (targeted) and figure out how many frames until -2 y velocity, test Hold Up (untargeted) and figure out how many frames until -2 y velocity. Then try various numbers of ESS turns before testing each case (targeted/untargeted)

when accounting for ESS turns, try to keep your angle between -21000 and -4000; this corresponds to at most 6 ESS turns left and at most 3 ESS turns right


--- write a function to set down a bomb
---- write two functions: set_down_() and set_down_target() (the latter which sets down the bomb and targets in the process???)

idea: try 3 cases: walk forward without targeting, target while setting down bomb before walking forward, walk forward AND target at the same time (all 3 cases can give different angles)