How to DI (almost) Every Move in Melee

[Sycorax]

This thread is obsolete. Read this instead: http://smashboards.com/threads/the-specifics-of-trajectory-di.437171/

 

[Kadano]

Nice work!

You should change the download link from the mediafire view link to its proper download link, though—the values as mediafire reads it from the xlsx are completely wrong and might confuse people.

“It is my suggestion that you round this angle to the nearest registerable angle, e.g. 10 degrees rounds to 17.3 degrees and 359 degrees rounds to 0 degrees.”

10° would be rounded to 0°, and so would 17.2°. The game is really unforgiving in that regard.

Maybe you might want to add that reverse hits are determined by the attacker’s and target’s base positions. To be exact, reverse hits are the same as hitting from the other side.
For example, if a Marth hits a “reverse fair”, where the fair is executed facing to the right, but the target flying to the left, it’s the same calculation as if Marth had been facing to the left (and hit with the same hitbubble).
In both cases, Marth’s base position is on the right side of the target, so the pre-DI trajectory is [180°−(base trajectory)]. (Unlike @Magus420, I can’t read Melee code directly, so I’m not 100% sure the game calculates it exactly like this, but fairly certain.)

 

[-ShadowPhoenix-]

This is sick

 

[Sycorax]

You should change the download link from the mediafire view link to its proper download link

I used MediaFire for the first time today so I'm not really sure how it works Let's see if I can make it better.

10° would be rounded to 0°, and so would 17.2°. The game is really unforgiving in that regard.

Hmm, I guess I was unclear. The human should round. E.g. Falco's Up B sends you at 80° so you should DI at 350°, but you can't. Rounding to the nearest allowable angle would be about 342°. Since 342° is closer to the optimal DI angle, 350°, it would yield greater directional influence than 0° (which would come from any DI between about 343° and 360°).

The specifics of reverse hits is outside the scope of this post, imo. I just included the tidbit in case someone felt compelled to ask a question about it. I'll copy and paste some of the stuff you just wrote if you don't mind.

 

[Dapplegonger]

You said that Popo's are probably the same as Nana's, but I'm not sure if that's true. I don't know about all of their moves, but I believe I saw somewhere that Popo's f-air is a meteor smash while Nana's is a spike. Probably not the highest priority if you didn't do every character, but it could be something to look into.

 

[Sycorax]

I believe I saw somewhere that Popo's f-air is a meteor smash while Nana's is a spike.

This is unequivocally false. I went in and checked the game data.

 

[NFMLH2]

That's a lot of values and stuff. I can't do this..

 

[Espi]

I didn't know that I was Di'ing so many moves wrong...

 

[Stratocaster]

I believe the "dead zone" turns out to be pretty insignificant, because the difference between close angles is pretty negligible do to of the nature of a circular input which is converted into a component of the perpedicular distance from the launch angle. What I'm trying to get at it's the sine/cosine of the angle that determines the angle like if the launch angle was 0 then to get an angle of 18 it'd be:

18 * sin ( 90 deg ) = 18
and thus for other angles...
18 * sin ( 89 deg ) = 17.997
18 * sin ( 82 deg ) = 17.8
18 * sin ( 70 deg ) = 16.9
18 * sin ( 45 deg ) = 12.7
18 * sin ( 20 deg ) = 6.2
18 * sin ( 0 deg ) = 0

What I want you to notice is that the values don't change much for the partial DI just because you are off a few degrees. Being off by one degree will change your launch angle by 0.003. Being off by 8 degrees only makes a difference of 0.2 in your launch angle. You have to be off by about 20 degrees to see real significant effects in your DI at which point your angle will be effected 1 degree.

Since You can hit 0 and 17.3, in the case you needed to DI 8.45 to be perfect, choosing either of those options will only effect your DI by 0.2 degrees, so you shouldn't concern yourself with it. It's why we've never noticed you can't DI some moves the full 18 degrees difference. You can DI 17.8 at least so you're not going to miss the small drop-off.

I hope this makes sense as I feel I'm having a little trouble explaining it thoroughly.

Also another thing to keep in mind is that you cannot DI fully in most directions because the octagonal shape prevents you from holding the stick fully extended. The consequence of this is that DIing in the notches instead of the exact "ideal" angle around the notches may just as good or so because it's really the distance from the axis of the launch angle that matters rather than the launch angle (on a circle this would be the fully extended perpendicular angle, but on an octagon it is not so straight-forward).

And one more thing concerning survival DI: The best DI is not always the "ideal" DI shown here. The issue is that you really want your trajectory to send you to the corner as much as possible. The "ideal" DI is a great guide and about as good as is realistic, but for an example I was testing when dthrow -> knee is a guaranteed kill on Puff and I found that using the exact angle that is perpendicular to the knee's launch angle was not always best because that DI actually sent Puff off the top blastzone while DI'ing "less" could help Puff survive by shooting Puff right into the corner. There are a lot of factors to the optimal survival DI, but the difference between the convential wisdom survival DI and "perfect" survival DI is about 5-10% depending on the stage and move and such. It's not likely to be the difference in winning games very often.

 

[Sycorax]

@ Stratocaster , that's some good input.

I believe the "dead zone" turns out to be pretty insignificant, because the difference between close angles is pretty negligible do[sic] to of the nature of a circular input which is converted into a component of the perpedicular[sic] distance from the launch angle...I hope this makes sense as I feel I'm having a little trouble explaining it thoroughly.

This makes sense to me because I know trigonometry, but I can't speak for the rest of smashers :/

What I want you to notice is that the values don't change much for the partial DI just because you are off a few degrees...You have to be off by about 20 degrees to see real significant effects[sic] in your DI at which point your angle will be effected[sic] 1 degree.

I think this is a good point for anyone out there who wants to memorize these values and try to minutely adjust their DI to get an extra .003 extra degrees of trajectory. It may be worth it in the case of a medium-strong hit like Marth's Fsmash at 100% that won't kill you but set you up to be edge guarded. With such a long hypotenuse from the hitstun of that move, a fraction of a degree could make a difference. It's worth testing, which I don't have the technical setup to do

Also another thing to keep in mind is that you cannot DI fully in most directions because the octagonal shape prevents you from holding the stick fully extended.

This is really interesting. I never thought of that. I'd love to see some testing and data on that if anyone is able to.

The best DI is not always the "ideal" DI shown here. The issue is that you really want your trajectory to send you to the corner as much as possible.

This is very true and an excellent point. Though implicit, I hoped I made it clear in my bold red note.

P.S. Are you going to upload those pictures of all the frames of Marth's moves that you have for EVERY other character? I don't need them for anything, but I'd like to have a complete set T_T Please and thank you :D

 

[dude it's raining]

The change in KB Trajectory due to TDI is:
<Control Stick Displacement>*sin(<DI Angle relative to KB Trajectory>)

Control Stick Displacement is:
18 at the eight corners of the octagon
OR
18*sin(67.5 degrees)/sin(112.5 - <angle between control stick position and a corner>)
(Thanks to ED for the math)

If the angle between the control stick position and a corner is 17.2 degrees or less, it is input as zero, thus:

For TDI down, the new angle is:
<KB Angle> - <18>*sin(<DI Angle>)

For Perfect Double Stick DI, the new angle (without ASDI calculated) is:
<KB Angle> - <18*sin(67.5 degrees)/sin(112.5 - <90 degrees - KB Angle>)*sin(<DI Angle>)
OR if the KB Angle is 72.8 degrees or higher, it is either:
<KB Angle> - <18*sin(67.5 degrees)/sin(112.5 - <17.3 degrees>)*sin(<DI Angle>)
OR (whichever is bigger):
<KB Angle> - <18*sin(<KB Angle>)

Note: if the KB Angle is greater than 90 degrees, just translate it so it's in quadrant 1.
Note: it doesn't matter for these calculations, but max DI Angle is:
72.7 + (90 - <KB Angle>)

Note: So far (I've only tested Sheik on Fox), it seems that going for the 17.3 degrees is never better than going for a cardinal direction due to the octagonal displacement.

 

[Sycorax]

The change in KB Trajectory due to TDI is:
<Control Stick Displacement>*sin(<DI Angle relative to KB Trajectory>)

Control Stick Displacement is:
18 at the eight corners of the octagon
OR
18*sin(67.5 degrees)/sin(112.5 - <angle between control stick position and a corner>)
(Thanks to ED for the math)

If the angle between the control stick position and a corner is 17.2 degrees or less, it is input as zero, thus:

For TDI down, the new angle is:
<KB Angle> - <18>*sin(<DI Angle>)

For Perfect Double Stick DI, the new angle (without ASDI calculated) is:
<KB Angle> - <18*sin(67.5 degrees)/sin(112.5 - <90 degrees - KB Angle>)*sin(<DI Angle>)
OR if the KB Angle is 72.8 degrees or higher, it is either:
<KB Angle> - <18*sin(67.5 degrees)/sin(112.5 - <17.3 degrees>)*sin(<DI Angle>)
OR (whichever is bigger):
<KB Angle> - <18*sin(<KB Angle>)

Note: if the KB Angle is greater than 90 degrees, just translate it so it's in quadrant 1.
Note: it doesn't matter for these calculations, but max DI Angle is:
72.7 + (90 - <KB Angle>)

Note: So far (I've only tested Sheik on Fox), it seems that going for the 17.3 degrees is never better than going for a cardinal direction due to the octagonal displacement.

Thanks for the math, but Magus posted something recently that would change your formulae. Flerbert_ is Magus.

 

[TastyCarcass]

Hi, it seems the file is no longer available. Can anyone post a mirror?

 

[Sycorax]

Hi, it seems the file is no longer available. Can anyone post a mirror?

This thread is obsolete. Read this instead: http://smashboards.com/threads/the-specifics-of-trajectory-di.437171/