Talk:PDIF

Model Review

At the moment there are 2~3 models for melee pDIF listed on the main page. I'm going to fall back on the old "All models are wrong, but some models are useful" truism at this point. There are reasons to prefer Motenten's model and there are reasons to prefer pchan's model. In practice, I would say that using Motenten's model all the time is pretty much the best course of action.

However, pchan's 1H model does have some attractive points:

• It uses similar constants as the ranged pDIF equations, which have long been assumed to be entirely accurate though their original source is unclear.
• It is very simple.

Motenten's model's advantages are:

• It matches the data more accurately (especially for critical hits).
• It works for both 1H and 2H weapons.

Neither model fully accounts for:

• How is the final value actually calculated? - There is a frequency difference between high and low values at any Ratio that has yet to be satisfactorily explained.
• What is the average pDIF for a given cRatio? - This is the primary thing that anyone who visits this page will want to know. Motenten explicitly has not determined this and pchan essentially hardcodes the 1.0 cRatio frequency spike into his data with further processing that isn't included on the main page then makes up the rest with very little data.

That's all that I have at the moment. --Byrthnoth 14:17, 11 January 2013 (EST)

Pchan's Post-cRatio Model

In this model, the two types of melee damage (1H and 2H weapons), use fundamentally different pDIF minimum and maximum equations. To determine the minimum and maximum, you must use the equation corresponding to your cRatio.

1. Minimum and Maximum pDIF

1H Weaponry

cRatio UL Equation 0 ≤ cRatio < 0.5 (cRatio - 0.5)×(10/9) + 1 0.5 ≤ cRatio < 0.75 1 0.75 ≤ cRatio < 2 (cRatio - 0.75)×(10/9) + 1 cRatio LL Equation 0 ≤ cRatio < 0.5 (1/6) 0.5 ≤ cRatio < 1.25 (cRatio - 1.25)×(10/9) + 1 1.25 ≤ cRatio < 1.5 1 1.5 ≤ cRatio < 2 (cRatio - 1.5)×(10/9) + 1 Min Max (4/9) 1 1 1 1 2.389 Min Max (1/6) (1/6) (1/6) 1 1 1 1 (14/9)

pchan 1H Model min and max pDIF curves

2H Weaponry (It is worth noting that pchan primarily focused his efforts on 1H weapons)

cRatio UL Equation 0 ≤ cRatio < 0.5 (cRatio×1.2) + 0.4 0.5 ≤ cRatio < 5/6 1 5/6 ≤ cRatio < 10/6 cRatio × 1.25 10/6 ≤ cRatio < 2.25 cRatio × 1.2 cRatio LL Equation 0 ≤ cRatio < (5/12) 0 (5/12) ≤ cRatio < 1.25 1.2×cRatio - 0.5 1.25 ≤ cRatio < 1.5 1 1.5 ≤ cRatio < 2 1.2×cRatio - 0.8 Min Max 0.4 1 1 1 1.042 2.083 2 2.7 Min Max 0 0 0 1 1 1 1 1.9

pchan 2H Model min and max pDIF curves

2. Critical pDIF

• After finding the minimum and maximum, a pDIF between these two values is selected as your pDIF
  • If the pDIF is less than zero, change the pDIF to zero
  • If the pDIF exceeds three, set to three
• Once the pDIF has been selected, determine whether the hit was a critical hit or normal hit
  • If normal hit, add zero to pDIF
  • If critical hit, add 1 to pDIF

ex: If your pDIF turned out to be 1.9 and you critical hit, your new pDIF is 2.9

3. Final pDIF

• After calculating critical pDIF, a final multiplier applies
  • Multiply pDIF by a random value between 1 and 1.05

Masamune

These are the fits obtained from Masamune's "Distribution Study 1.1.xls":

• Critical hit: pDIF Max:
  • if 0 ≤ cRatio < 1.63 → Maximum Crit pDIF = 1.007 × cRatio + 1.328
  • if 1.63 ≤ cRatio → Maximum Crit pDIF = 3
• Critical hit: pDIF Min:
  • Minimum Crit pDIF = 1.007 × cRatio + 0.51
    
    

• Normal hit: pDIF Max:
  • Maximum Normal pDIF = 1.019 × cRatio + 0.485
• Normal hit: pDIF Min:
  • if 0 < cRatio < 1.3 → Minimum Normal pDIF = 1.129 × cRatio - 0.430
  • if 1.3 ≤ cRatio < 1.5 → Minimum Normal pDIF = 1.035
  • if 1.5 ≤ cRatio < 1.5 → Minimum Normal pDIF = 1.13 × cRatio - 0.700