I was measuring the distance at which Haohmaru uses close.HP instead of far.HP. According to Jesuszilla's Data Tool it was about 91.5 pixels. This, when translated to MUGEN, made cl.HP whiff at maximum range (the problem was me using P2BodyDist instead of P2Dist, but in the end it this lead to something).

Weird translations in horizontal distances have been noted in the past. extravagant pointed out how Fausto's Claw, a character that used datamined data from Jesuszilla's tool, was traveling more horizontal distance than it should when superjumping back.

You can see a direct comparison between my Terry (Red) and CvS2 Terry (Blue) jumping back.



My Terry clearly travels more horizontal distance. What's up with that? I used datamined data to make Terry, so what happened? There's a catch. Since CvS2 coordinate space values are 358x224 there's a 1 : 0.8938547486 ratio between CvS2 units of horizontal distance and real pixels (MUGEN pixels)


According to the data tool, Haohmaru uses a "PosAdd" of 26 on that frame, but when you measure the distance in real pixels, he only moved 23
which follows the 1 : 0.9 (or 1 : 0.88461538461 to be more specific) ratio I refer earlier.

This means that all horizontal displacement controllers (PosAdds, VelAdds, VelSets) made with data from Jesuszilla's data tool should be multiplied by 0.88461538461 to get a more accurate translation. Jesuszilla will update his CT table soon to match this discovery, but if your character was made previous to this, you'll have to update the velocities.

Note that vertical velocities are fine, there's a 1 : 1 ratio between CvS2 units of vertical distance and real pixels, so yaccels, and all sorts of vertical movement made with JZ's tool is accurate.