Skip to content

Commit

Permalink
super experimooper | 1
Browse files Browse the repository at this point in the history
  • Loading branch information
MinaciousGrace committed Jul 1, 2020
1 parent df1f412 commit 1233b6d
Show file tree
Hide file tree
Showing 4 changed files with 13 additions and 13 deletions.
20 changes: 10 additions & 10 deletions src/Etterna/Globals/MinaCalc.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -506,7 +506,7 @@ Calc::Chisel(float player_skill,
}
if (ss == Skill_Technical) {
gotpoints -= fastsqrt(
jackloss(player_skill * 0.625F, *this, hi) * 0.75F);
jackloss(player_skill * 0.651F, *this, hi) * 0.75F);
}
}
}
Expand Down Expand Up @@ -562,14 +562,14 @@ Calc::Chisel(float player_skill,
/* the new way we wil attempt to diffrentiate skillsets rather than using
* normalizers is by detecting whether or not we think a file is mostly
* comprised of a given pattern, producing a downscaler that slightly buffs
* up those files and produces a downscaler for files not detected of that
* type. the major potential failing of this system is that it ends up such
* that the rating is tied directly to whether or not a file can be more or
* less strongly determined to be of a pattern type, e.g. splithand trills
* being marked as more "js" than actual js, for the moment these modifiers
* are still built on proportion of taps in chords / total taps, but there's
* a lot more give than their used to be. they should be re-done as
* sequential detection for best effect but i don't know if that will be
* up those files and produces a downscaler for files not detected of that
* type. the major potential failing of this system is that it ends up such
* that the rating is tied directly to whether or not a file can be more or
* less strongly determined to be of a pattern type, e.g. splithand trills
* being marked as more "js" than actual js, for the moment these modifiers
* are still built on proportion of taps in chords / total taps, but there's
* a lot more give than their used to be. they should be re-done as
* sequential detection for best effect but i don't know if that will be
* necessary for basic tuning if we don't do this files may end up
* misclassing hard and polluting leaderboards, and good scores on overrated
* files will simply produce high ratings in every category */
Expand Down Expand Up @@ -836,7 +836,7 @@ MinaSDCalcDebug(const vector<NoteInfo>& NoteInfo,
}
}

int mina_calc_version = 408;
int mina_calc_version = 1;
auto
GetCalcVersion() -> int
{
Expand Down
Original file line number Diff line number Diff line change
@@ -1,7 +1,7 @@
#pragma once
#include "Etterna/Globals/MinaCalc/Dependent/HD_MetaSequencing.h"

static const float rma_diff_scaler = 1.2F;
static const float rma_diff_scaler = 1.21F;

enum rm_behavior
{
Expand Down
2 changes: 1 addition & 1 deletion src/Etterna/Globals/MinaCalc/SequencedBaseDiffCalc.h
Original file line number Diff line number Diff line change
Expand Up @@ -106,7 +106,7 @@ struct techyo
// we definitely don't want to average here because we don't want tech
// to only be files with strong runningman pattern detection, but we
// could probably do something more robust at some point
return max(weighted_average(get_tc_base(calc), nps_base, 5.5F, 9.F),
return max(weighted_average(get_tc_base(calc), nps_base, 4.F, 9.F),
rm_itv_max_diff);
}

Expand Down
2 changes: 1 addition & 1 deletion src/Etterna/Globals/MinaCalc/UlbuAcolytes.h
Original file line number Diff line number Diff line change
Expand Up @@ -12,7 +12,7 @@
* patterns have lower enps than streams, streams default to 1 and chordstreams
* start lower, stam is a special case and may use normalizers again */
static const std::array<float, NUM_Skillset> basescalers = {
0.F, 0.93F, 0.89F, 0.82F, 0.925F, 0.91F, 0.815F, 0.83F
0.F, 0.93F, 0.89F, 0.82F, 0.925F, 0.91F, 0.815F, 0.84F
};

static const std::string calc_params_xml = "Save/calc params.xml";
Expand Down

0 comments on commit 1233b6d

Please sign in to comment.