mirror of
https://github.com/CTCaer/RetroArch.git
synced 2024-12-23 03:10:14 +00:00
347 lines
7.3 KiB
C
347 lines
7.3 KiB
C
/* RetroArch - A frontend for libretro.
|
|
* Copyright (C) 2014-2015 - Jean-André Santoni
|
|
* Copyright (C) 2011-2015 - Daniel De Matteis
|
|
*
|
|
* RetroArch is free software: you can redistribute it and/or modify it under the terms
|
|
* of the GNU General Public License as published by the Free Software Found-
|
|
* ation, either version 3 of the License, or (at your option) any later version.
|
|
*
|
|
* RetroArch is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY;
|
|
* without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
|
|
* PURPOSE. See the GNU General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU General Public License along with RetroArch.
|
|
* If not, see <http://www.gnu.org/licenses/>.
|
|
*/
|
|
|
|
#include "menu_animation.h"
|
|
#include <math.h>
|
|
|
|
static tween_t* tweens = NULL;
|
|
static int numtweens = 0;
|
|
|
|
static void tween_free(tween_t *tw)
|
|
{
|
|
if (tw)
|
|
free(tw);
|
|
tw = NULL;
|
|
}
|
|
|
|
void add_tween(float duration, float target_value, float* subject,
|
|
easingFunc easing, tweenCallback callback)
|
|
{
|
|
tween_t *tween = NULL;
|
|
tween_t *temp_tweens = (tween_t*)
|
|
realloc(tweens, (numtweens + 1) * sizeof(tween_t));
|
|
|
|
if (temp_tweens)
|
|
{
|
|
tweens = temp_tweens;
|
|
}
|
|
else /* Realloc failed. */
|
|
{
|
|
tween_free(tweens);
|
|
return;
|
|
}
|
|
|
|
numtweens++;
|
|
tween = (tween_t*)&tweens[numtweens-1];
|
|
|
|
if (!tween)
|
|
return;
|
|
|
|
tween->alive = 1;
|
|
tween->duration = duration;
|
|
tween->running_since = 0;
|
|
tween->initial_value = *subject;
|
|
tween->target_value = target_value;
|
|
tween->subject = subject;
|
|
tween->easing = easing;
|
|
tween->callback = callback;
|
|
}
|
|
|
|
static void update_tween(tween_t *tween, float dt, int *active_tweens)
|
|
{
|
|
if (!tween)
|
|
return;
|
|
|
|
if (tween->running_since >= tween->duration)
|
|
return;
|
|
|
|
tween->running_since += dt;
|
|
|
|
if (tween->easing)
|
|
*tween->subject = tween->easing(
|
|
tween->running_since,
|
|
tween->initial_value,
|
|
tween->target_value - tween->initial_value,
|
|
tween->duration);
|
|
|
|
if (tween->running_since >= tween->duration)
|
|
{
|
|
*tween->subject = tween->target_value;
|
|
|
|
if (tween->callback)
|
|
tween->callback();
|
|
}
|
|
|
|
if (tween->running_since < tween->duration)
|
|
*active_tweens += 1;
|
|
}
|
|
|
|
void update_tweens(float dt)
|
|
{
|
|
int i;
|
|
int active_tweens = 0;
|
|
|
|
for(i = 0; i < numtweens; i++)
|
|
update_tween(&tweens[i], dt, &active_tweens);
|
|
|
|
if (!active_tweens)
|
|
numtweens = 0;
|
|
}
|
|
|
|
// linear
|
|
|
|
float linear(float t, float b, float c, float d)
|
|
{
|
|
return c * t / d + b;
|
|
}
|
|
|
|
// quad
|
|
|
|
float inQuad(float t, float b, float c, float d)
|
|
{
|
|
return c * pow(t / d, 2) + b;
|
|
}
|
|
|
|
float outQuad(float t, float b, float c, float d)
|
|
{
|
|
t = t / d;
|
|
return -c * t * (t - 2) + b;
|
|
}
|
|
|
|
float inOutQuad(float t, float b, float c, float d)
|
|
{
|
|
t = t / d * 2;
|
|
if (t < 1)
|
|
return c / 2 * pow(t, 2) + b;
|
|
return -c / 2 * ((t - 1) * (t - 3) - 1) + b;
|
|
}
|
|
|
|
float outInQuad(float t, float b, float c, float d)
|
|
{
|
|
if (t < d / 2)
|
|
return outQuad(t * 2, b, c / 2, d);
|
|
return inQuad((t * 2) - d, b + c / 2, c / 2, d);
|
|
}
|
|
|
|
// cubic
|
|
|
|
float inCubic(float t, float b, float c, float d)
|
|
{
|
|
return c * pow(t / d, 3) + b;
|
|
}
|
|
|
|
float outCubic(float t, float b, float c, float d)
|
|
{
|
|
return c * (pow(t / d - 1, 3) + 1) + b;
|
|
}
|
|
|
|
float inOutCubic(float t, float b, float c, float d)
|
|
{
|
|
t = t / d * 2;
|
|
if (t < 1)
|
|
return c / 2 * t * t * t + b;
|
|
t = t - 2;
|
|
return c / 2 * (t * t * t + 2) + b;
|
|
}
|
|
|
|
float outInCubic(float t, float b, float c, float d)
|
|
{
|
|
if (t < d / 2)
|
|
return outCubic(t * 2, b, c / 2, d);
|
|
return inCubic((t * 2) - d, b + c / 2, c / 2, d);
|
|
}
|
|
|
|
// quart
|
|
|
|
float inQuart(float t, float b, float c, float d)
|
|
{
|
|
return c * pow(t / d, 4) + b;
|
|
}
|
|
|
|
float outQuart(float t, float b, float c, float d)
|
|
{
|
|
return -c * (pow(t / d - 1, 4) - 1) + b;
|
|
}
|
|
|
|
float inOutQuart(float t, float b, float c, float d)
|
|
{
|
|
t = t / d * 2;
|
|
if (t < 1)
|
|
return c / 2 * pow(t, 4) + b;
|
|
return -c / 2 * (pow(t - 2, 4) - 2) + b;
|
|
}
|
|
|
|
float outInQuart(float t, float b, float c, float d)
|
|
{
|
|
if (t < d / 2)
|
|
return outQuart(t * 2, b, c / 2, d);
|
|
return inQuart((t * 2) - d, b + c / 2, c / 2, d);
|
|
}
|
|
|
|
// quint
|
|
|
|
float inQuint(float t, float b, float c, float d)
|
|
{
|
|
return c * pow(t / d, 5) + b;
|
|
}
|
|
|
|
float outQuint(float t, float b, float c, float d)
|
|
{
|
|
return c * (pow(t / d - 1, 5) + 1) + b;
|
|
}
|
|
|
|
float inOutQuint(float t, float b, float c, float d)
|
|
{
|
|
t = t / d * 2;
|
|
if (t < 1)
|
|
return c / 2 * pow(t, 5) + b;
|
|
return c / 2 * (pow(t - 2, 5) + 2) + b;
|
|
}
|
|
|
|
float outInQuint(float t, float b, float c, float d)
|
|
{
|
|
if (t < d / 2)
|
|
return outQuint(t * 2, b, c / 2, d);
|
|
return inQuint((t * 2) - d, b + c / 2, c / 2, d);
|
|
}
|
|
|
|
// sine
|
|
|
|
float inSine(float t, float b, float c, float d)
|
|
{
|
|
return -c * cos(t / d * (M_PI / 2)) + c + b;
|
|
}
|
|
|
|
float outSine(float t, float b, float c, float d)
|
|
{
|
|
return c * sin(t / d * (M_PI / 2)) + b;
|
|
}
|
|
|
|
float inOutSine(float t, float b, float c, float d)
|
|
{
|
|
return -c / 2 * (cos(M_PI * t / d) - 1) + b;
|
|
}
|
|
|
|
float outInSine(float t, float b, float c, float d)
|
|
{
|
|
if (t < d / 2)
|
|
return outSine(t * 2, b, c / 2, d);
|
|
return inSine((t * 2) -d, b + c / 2, c / 2, d);
|
|
}
|
|
|
|
// expo
|
|
|
|
float inExpo(float t, float b, float c, float d)
|
|
{
|
|
if (t == 0)
|
|
return b;
|
|
return c * powf(2, 10 * (t / d - 1)) + b - c * 0.001;
|
|
}
|
|
|
|
float outExpo(float t, float b, float c, float d)
|
|
{
|
|
if (t == d)
|
|
return b + c;
|
|
return c * 1.001 * (-powf(2, -10 * t / d) + 1) + b;
|
|
}
|
|
|
|
float inOutExpo(float t, float b, float c, float d)
|
|
{
|
|
if (t == 0)
|
|
return b;
|
|
if (t == d)
|
|
return b + c;
|
|
t = t / d * 2;
|
|
if (t < 1)
|
|
return c / 2 * powf(2, 10 * (t - 1)) + b - c * 0.0005;
|
|
return c / 2 * 1.0005 * (-powf(2, -10 * (t - 1)) + 2) + b;
|
|
}
|
|
|
|
float outInExpo(float t, float b, float c, float d)
|
|
{
|
|
if (t < d / 2)
|
|
return outExpo(t * 2, b, c / 2, d);
|
|
return inExpo((t * 2) - d, b + c / 2, c / 2, d);
|
|
}
|
|
|
|
// circ
|
|
|
|
float inCirc(float t, float b, float c, float d)
|
|
{
|
|
return(-c * (sqrt(1 - powf(t / d, 2)) - 1) + b);
|
|
}
|
|
|
|
float outCirc(float t, float b, float c, float d)
|
|
{
|
|
return(c * sqrt(1 - powf(t / d - 1, 2)) + b);
|
|
}
|
|
|
|
float inOutCirc(float t, float b, float c, float d)
|
|
{
|
|
t = t / d * 2;
|
|
if (t < 1)
|
|
return -c / 2 * (sqrt(1 - t * t) - 1) + b;
|
|
t = t - 2;
|
|
return c / 2 * (sqrt(1 - t * t) + 1) + b;
|
|
}
|
|
|
|
float outInCirc(float t, float b, float c, float d)
|
|
{
|
|
if (t < d / 2)
|
|
return outCirc(t * 2, b, c / 2, d);
|
|
return inCirc((t * 2) - d, b + c / 2, c / 2, d);
|
|
}
|
|
|
|
// bounce
|
|
|
|
float outBounce(float t, float b, float c, float d)
|
|
{
|
|
t = t / d;
|
|
if (t < 1 / 2.75)
|
|
return c * (7.5625 * t * t) + b;
|
|
if (t < 2 / 2.75)
|
|
{
|
|
t = t - (1.5 / 2.75);
|
|
return c * (7.5625 * t * t + 0.75) + b;
|
|
}
|
|
else if (t < 2.5 / 2.75)
|
|
{
|
|
t = t - (2.25 / 2.75);
|
|
return c * (7.5625 * t * t + 0.9375) + b;
|
|
}
|
|
t = t - (2.625 / 2.75);
|
|
return c * (7.5625 * t * t + 0.984375) + b;
|
|
}
|
|
|
|
float inBounce(float t, float b, float c, float d)
|
|
{
|
|
return c - outBounce(d - t, 0, c, d) + b;
|
|
}
|
|
|
|
float inOutBounce(float t, float b, float c, float d)
|
|
{
|
|
if (t < d / 2)
|
|
return inBounce(t * 2, 0, c, d) * 0.5 + b;
|
|
return outBounce(t * 2 - d, 0, c, d) * 0.5 + c * .5 + b;
|
|
}
|
|
|
|
float outInBounce(float t, float b, float c, float d)
|
|
{
|
|
if (t < d / 2)
|
|
return outBounce(t * 2, b, c / 2, d);
|
|
return inBounce((t * 2) - d, b + c / 2, c / 2, d);
|
|
}
|