Accounting for DI?

[SSS]

Trying to make some percent lists for different characters. If I test kill moves on non-DIing dummy CPUs, how much percent should i add to account for DI?

 

[shairn]

Try using a second controller to test the DI, or get someone to help.

 

[SSS]

Try using a second controller to test the DI, or get someone to help.

there's not a rule of thumb?

 

[ECHOnce]

there's not a rule of thumb?

If only...I'd imagine that there's just too much variation between DI'ed directions and the KB received at different %s to account for using a single, easy-to-calculate rule of thumb.

 

[SSS]

Hmmmm. Gotta start testing. find them equations.

does anybody know the knockback scaling equation? or is it different for each move?

 

[Rizner]

Hmmmm. Gotta start testing. find them equations.

does anybody know the knockback scaling equation? or is it different for each move?

Depends on each moves kbg (knockback growth)

 

[Chesstiger2612]

The only thing it really works against are upwards killing moves, otherwise its even more:
KB=((Knockback Growth)/100*(14*((opponent's percentage before the move hits)+(damage of the hitbox))*((damage of the hitbox)+2)/(100+(weight)))+(base knockback)
Then vertical knockback obviously is sin(trajectory)*KB. Note that value. Now you can add/substract 18 from trajectory DI for perpendicular (maximum effect) DI. Obviously you will choose the one which lowers sin(trajectory), so the one farer away from 90°, so to speak.
Now solve for (opponent's percentage before the move hits). This is the percentage for maximum DI. Also rather round the value down if you need to round, because if it KOd at a certain percent without DI that just means it KOs between ((percentage)-1) and (percentage).

Overall, testing is probably easier.

 

[SSS]

The only thing it really works against are upwards killing moves, otherwise its even more:
KB=((Knockback Growth)/100*(14*((opponent's percentage before the move hits)+(damage of the hitbox))*((damage of the hitbox)+2)/(100+(weight)))+(base knockback)
Then vertical knockback obviously is sin(trajectory)*KB. Note that value. Now you can add/substract 18 from trajectory DI for perpendicular (maximum effect) DI. Obviously you will choose the one which lowers sin(trajectory), so the one farer away from 90°, so to speak.
Now solve for (opponent's percentage before the move hits). This is the percentage for maximum DI. Also rather round the value down if you need to round, because if it KOd at a certain percent without DI that just means it KOs between ((percentage)-1) and (percentage).

Overall, testing is probably easier.

This is still godlike. i don't really understand how to quantify something like base knockback, and figuring out the knockback growth. . .would i just have to see the knockback for each percent and find the slope of that? is knockback growth always linear?

 

[Chesstiger2612]

Knockback growth and base knockback are dependent on the hitbox only. One hitbox has a value for both that doesn't change. I could search for the values if you have only questions on a few moves, otherwise you need to work with PSA

 

[DrinkingFood]

Your KB formula is wrong, chesstiger. It's:
KB=(((((damage after hit / 10) + ((damage after hit * damage of the move when fresh) / 20)) * (200 / (100+weight)) * 1.4) + 18) * (KBG/100)) + BKB
Rainbow color coded parenthesis for viewing pleasure

Guide: hitstun frames are determined by multiplying the above result by 0.4. determining launch speed works by multiplying the above result by like .03, but that just gives you units per frame which is useless to our understanding how far that is.
32 frames of hitstun cause knockdown, which is 80 KB.

 

[Chesstiger2612]

Oh I was using the Melee one from here thank you for correcting me.
EDIT: see next post

 

[tauKhan]

Your KB formula is wrong, chesstiger. It's:
KB=(((((damage after hit / 10) + ((damage after hit * damage of the move when fresh) / 20)) * (200 / (100+weight)) * 1.4) + 18) * (KBG/100)) + BKB
Rainbow color coded parenthesis for chewing pleasure

It actually is exactly same function, the one @ Chesstiger2612 posted is just a "simplification" of the formula you posted made by kadano.

 

[DrinkingFood]

Oh yeah, redoing the work, I see how he got that simplified version.
Except it doesn't seem all that simplified, I find the other one easier to do in my head.
That one also doesn't tell you how to account for damage staling. Since knockback uses damage after the hit to determine the final result, but otherwise uses fresh damage when multiplying, simplifying it that way requires you differentiate between the two "d" variables, the first of which is staled damage and the second is fresh. The extended formula corrects this by using damage after the hit and only ever using damage once, where it only ever means one thing- fresh damage.

 

[tauKhan]

The significant difference is that the one formula chess posted doesn't have the redundant decimal/large coefficients like the old one.