mirror of
https://github.com/Relintai/pandemonium_engine.git
synced 2024-12-28 22:57:15 +01:00
324 lines
10 KiB
C++
324 lines
10 KiB
C++
|
/**
|
||
|
* Adapted from Penner Easing equations' C++ port.
|
||
|
* Source: https://github.com/jesusgollonet/ofpennereasing
|
||
|
* License: BSD-3-clause
|
||
|
*/
|
||
|
|
||
|
#include "scene/animation/tween.h"
|
||
|
|
||
|
const real_t pi = 3.1415926535898;
|
||
|
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
// linear
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
namespace linear {
|
||
|
static real_t in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return c * t / d + b;
|
||
|
}
|
||
|
|
||
|
static real_t out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return c * t / d + b;
|
||
|
}
|
||
|
|
||
|
static real_t in_out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return c * t / d + b;
|
||
|
}
|
||
|
|
||
|
static real_t out_in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return c * t / d + b;
|
||
|
}
|
||
|
}; // namespace linear
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
// sine
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
namespace sine {
|
||
|
static real_t in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return -c * cos(t / d * (pi / 2)) + c + b;
|
||
|
}
|
||
|
|
||
|
static real_t out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return c * sin(t / d * (pi / 2)) + b;
|
||
|
}
|
||
|
|
||
|
static real_t in_out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return -c / 2 * (cos(pi * t / d) - 1) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out_in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return (t < d / 2) ? out(t * 2, b, c / 2, d) : in((t * 2) - d, b + c / 2, c / 2, d);
|
||
|
}
|
||
|
}; // namespace sine
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
// quint
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
namespace quint {
|
||
|
static real_t in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return c * pow(t / d, 5) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return c * (pow(t / d - 1, 5) + 1) + b;
|
||
|
}
|
||
|
|
||
|
static real_t in_out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
t = t / d * 2;
|
||
|
if (t < 1) return c / 2 * pow(t, 5) + b;
|
||
|
return c / 2 * (pow(t - 2, 5) + 2) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out_in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return (t < d / 2) ? out(t * 2, b, c / 2, d) : in((t * 2) - d, b + c / 2, c / 2, d);
|
||
|
}
|
||
|
}; // namespace quint
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
// quart
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
namespace quart {
|
||
|
static real_t in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return c * pow(t / d, 4) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return -c * (pow(t / d - 1, 4) - 1) + b;
|
||
|
}
|
||
|
|
||
|
static real_t in_out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
t = t / d * 2;
|
||
|
if (t < 1) return c / 2 * pow(t, 4) + b;
|
||
|
return -c / 2 * (pow(t - 2, 4) - 2) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out_in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return (t < d / 2) ? out(t * 2, b, c / 2, d) : in((t * 2) - d, b + c / 2, c / 2, d);
|
||
|
}
|
||
|
}; // namespace quart
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
// quad
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
namespace quad {
|
||
|
static real_t in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return c * pow(t / d, 2) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
t = t / d;
|
||
|
return -c * t * (t - 2) + b;
|
||
|
}
|
||
|
|
||
|
static real_t in_out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
t = t / d * 2;
|
||
|
if (t < 1) return c / 2 * pow(t, 2) + b;
|
||
|
return -c / 2 * ((t - 1) * (t - 3) - 1) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out_in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return (t < d / 2) ? out(t * 2, b, c / 2, d) : in((t * 2) - d, b + c / 2, c / 2, d);
|
||
|
}
|
||
|
}; // namespace quad
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
// expo
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
namespace expo {
|
||
|
static real_t in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
if (t == 0) return b;
|
||
|
return c * pow(2, 10 * (t / d - 1)) + b - c * 0.001;
|
||
|
}
|
||
|
|
||
|
static real_t out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
if (t == d) return b + c;
|
||
|
return c * 1.001 * (-pow(2, -10 * t / d) + 1) + b;
|
||
|
}
|
||
|
|
||
|
static real_t in_out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
if (t == 0) return b;
|
||
|
if (t == d) return b + c;
|
||
|
t = t / d * 2;
|
||
|
if (t < 1) return c / 2 * pow(2, 10 * (t - 1)) + b - c * 0.0005;
|
||
|
return c / 2 * 1.0005 * (-pow(2, -10 * (t - 1)) + 2) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out_in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return (t < d / 2) ? out(t * 2, b, c / 2, d) : in((t * 2) - d, b + c / 2, c / 2, d);
|
||
|
}
|
||
|
}; // namespace expo
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
// elastic
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
namespace elastic {
|
||
|
static real_t in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
if (t == 0) return b;
|
||
|
if ((t /= d) == 1) return b + c;
|
||
|
float p = d * 0.3f;
|
||
|
float a = c;
|
||
|
float s = p / 4;
|
||
|
float postFix = a * pow(2, 10 * (t -= 1)); // this is a fix, again, with post-increment operators
|
||
|
return -(postFix * sin((t * d - s) * (2 * pi) / p)) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
if (t == 0) return b;
|
||
|
if ((t /= d) == 1) return b + c;
|
||
|
float p = d * 0.3f;
|
||
|
float a = c;
|
||
|
float s = p / 4;
|
||
|
return (a * pow(2, -10 * t) * sin((t * d - s) * (2 * pi) / p) + c + b);
|
||
|
}
|
||
|
|
||
|
static real_t in_out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
if (t == 0) return b;
|
||
|
if ((t /= d / 2) == 2) return b + c;
|
||
|
float p = d * (0.3f * 1.5f);
|
||
|
float a = c;
|
||
|
float s = p / 4;
|
||
|
|
||
|
if (t < 1) {
|
||
|
float postFix = a * pow(2, 10 * (t -= 1)); // postIncrement is evil
|
||
|
return -0.5f * (postFix * sin((t * d - s) * (2 * pi) / p)) + b;
|
||
|
}
|
||
|
float postFix = a * pow(2, -10 * (t -= 1)); // postIncrement is evil
|
||
|
return postFix * sin((t * d - s) * (2 * pi) / p) * 0.5f + c + b;
|
||
|
}
|
||
|
|
||
|
static real_t out_in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return (t < d / 2) ? out(t * 2, b, c / 2, d) : in((t * 2) - d, b + c / 2, c / 2, d);
|
||
|
}
|
||
|
}; // namespace elastic
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
// cubic
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
namespace cubic {
|
||
|
static real_t in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
t /= d;
|
||
|
return c * t * t * t + b;
|
||
|
}
|
||
|
|
||
|
static real_t out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
t = t / d - 1;
|
||
|
return c * (t * t * t + 1) + b;
|
||
|
}
|
||
|
|
||
|
static real_t in_out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
t /= d / 2;
|
||
|
if (t < 1) return c / 2 * t * t * t + b;
|
||
|
t -= 2;
|
||
|
return c / 2 * (t * t * t + 2) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out_in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return (t < d / 2) ? out(t * 2, b, c / 2, d) : in((t * 2) - d, b + c / 2, c / 2, d);
|
||
|
}
|
||
|
}; // namespace cubic
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
// circ
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
namespace circ {
|
||
|
static real_t in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
t /= d;
|
||
|
return -c * (sqrt(1 - t * t) - 1) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
t = t / d - 1;
|
||
|
return c * sqrt(1 - t * t) + b;
|
||
|
}
|
||
|
|
||
|
static real_t in_out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
t /= d / 2;
|
||
|
if (t < 1) {
|
||
|
return -c / 2 * (sqrt(1 - t * t) - 1) + b;
|
||
|
}
|
||
|
t -= 2;
|
||
|
return c / 2 * (sqrt(1 - t * t) + 1) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out_in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return (t < d / 2) ? out(t * 2, b, c / 2, d) : in((t * 2) - d, b + c / 2, c / 2, d);
|
||
|
}
|
||
|
}; // namespace circ
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
// bounce
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
namespace bounce {
|
||
|
static real_t out(real_t t, real_t b, real_t c, real_t d);
|
||
|
|
||
|
static real_t in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return c - out(d - t, 0, c, d) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
if ((t /= d) < (1 / 2.75f)) {
|
||
|
return c * (7.5625f * t * t) + b;
|
||
|
} else if (t < (2 / 2.75f)) {
|
||
|
float postFix = t -= (1.5f / 2.75f);
|
||
|
return c * (7.5625f * (postFix)*t + .75f) + b;
|
||
|
} else if (t < (2.5 / 2.75)) {
|
||
|
float postFix = t -= (2.25f / 2.75f);
|
||
|
return c * (7.5625f * (postFix)*t + .9375f) + b;
|
||
|
} else {
|
||
|
float postFix = t -= (2.625f / 2.75f);
|
||
|
return c * (7.5625f * (postFix)*t + .984375f) + b;
|
||
|
}
|
||
|
}
|
||
|
|
||
|
static real_t in_out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return (t < d / 2) ? in(t * 2, b, c / 2, d) : out((t * 2) - d, b + c / 2, c / 2, d);
|
||
|
}
|
||
|
|
||
|
static real_t out_in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return (t < d / 2) ? out(t * 2, b, c / 2, d) : in((t * 2) - d, b + c / 2, c / 2, d);
|
||
|
}
|
||
|
}; // namespace bounce
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
// back
|
||
|
///////////////////////////////////////////////////////////////////////////
|
||
|
namespace back {
|
||
|
static real_t in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
float s = 1.70158f;
|
||
|
float postFix = t /= d;
|
||
|
return c * (postFix)*t * ((s + 1) * t - s) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
float s = 1.70158f;
|
||
|
t = t / d - 1;
|
||
|
return c * (t * t * ((s + 1) * t + s) + 1) + b;
|
||
|
}
|
||
|
|
||
|
static real_t in_out(real_t t, real_t b, real_t c, real_t d) {
|
||
|
float s = 1.70158f * 1.525f;
|
||
|
t /= d / 2;
|
||
|
if (t < 1) return c / 2 * (t * t * ((s + 1) * t - s)) + b;
|
||
|
t -= 2;
|
||
|
return c / 2 * (t * t * ((s + 1) * t + s) + 2) + b;
|
||
|
}
|
||
|
|
||
|
static real_t out_in(real_t t, real_t b, real_t c, real_t d) {
|
||
|
return (t < d / 2) ? out(t * 2, b, c / 2, d) : in((t * 2) - d, b + c / 2, c / 2, d);
|
||
|
}
|
||
|
}; // namespace back
|
||
|
|
||
|
Tween::interpolater Tween::interpolaters[Tween::TRANS_COUNT][Tween::EASE_COUNT] = {
|
||
|
{ &linear::in, &linear::out, &linear::in_out, &linear::out_in },
|
||
|
{ &sine::in, &sine::out, &sine::in_out, &sine::out_in },
|
||
|
{ &quint::in, &quint::out, &quint::in_out, &quint::out_in },
|
||
|
{ &quart::in, &quart::out, &quart::in_out, &quart::out_in },
|
||
|
{ &quad::in, &quad::out, &quad::in_out, &quad::out_in },
|
||
|
{ &expo::in, &expo::out, &expo::in_out, &expo::out_in },
|
||
|
{ &elastic::in, &elastic::out, &elastic::in_out, &elastic::out_in },
|
||
|
{ &cubic::in, &cubic::out, &cubic::in_out, &cubic::out_in },
|
||
|
{ &circ::in, &circ::out, &circ::in_out, &circ::out_in },
|
||
|
{ &bounce::in, &bounce::out, &bounce::in_out, &bounce::out_in },
|
||
|
{ &back::in, &back::out, &back::in_out, &back::out_in },
|
||
|
};
|
||
|
|
||
|
real_t Tween::_run_equation(TransitionType p_trans_type, EaseType p_ease_type, real_t t, real_t b, real_t c, real_t d) {
|
||
|
if (d == 0) {
|
||
|
// Special case to avoid dividing by 0 in equations.
|
||
|
return b + c;
|
||
|
}
|
||
|
|
||
|
interpolater cb = interpolaters[p_trans_type][p_ease_type];
|
||
|
ERR_FAIL_COND_V(cb == NULL, b);
|
||
|
return cb(t, b, c, d);
|
||
|
}
|