From 5cf8391c5a281e6d6886aa240bfe2cba018fe042 Mon Sep 17 00:00:00 2001
From: aliaspider <aliaspider@gmail.com>
Date: Sun, 11 Dec 2016 15:52:32 +0100
Subject: [PATCH] replace the glsl snow shader with a more lightweight version.
 (gpu usage is approximately 25% compared to the original.)

---
 .../gl_shaders/pipeline_snow.glsl.frag.h      | 118 +++++++++---------
 menu/drivers_display/menu_display_gl.c        |   7 ++
 2 files changed, 68 insertions(+), 57 deletions(-)

diff --git a/gfx/drivers/gl_shaders/pipeline_snow.glsl.frag.h b/gfx/drivers/gl_shaders/pipeline_snow.glsl.frag.h
index e9f47c4554..2f4d96f0ac 100644
--- a/gfx/drivers/gl_shaders/pipeline_snow.glsl.frag.h
+++ b/gfx/drivers/gl_shaders/pipeline_snow.glsl.frag.h
@@ -1,65 +1,69 @@
 #include "shaders_common.h"
 
-static const char *stock_fragment_xmb_snow = GLSL(
-	uniform float time;
-	vec2 res = vec2(1920*3, 1080*3);
+static const char* stock_fragment_xmb_snow = GLSL(
+   uniform float time;
+   uniform vec2 OutputSize;
 
-	// High quality = larger patterns
-	// quality * 84 is the number of seconds in a loop
-	float quality = 8.0;
+   float baseScale = 3.5;  //  [1.0  .. 10.0]
+   float density   = 0.7;  //  [0.01 ..  1.0]
+   float speed     = 0.25; //  [0.1  ..  1.0]
 
-	float Cellular2D(vec2 P) {
-		//  https://github.com/BrianSharpe/Wombat/blob/master/Cellular2D.glsl
-		vec2 Pi = floor(P);
-		vec2 Pf = P - Pi;
-		vec4 Pt = vec4( Pi.xy, Pi.xy + 1.0 );
-		Pt = Pt - floor(Pt * ( 1.0 / 71.0 )) * 71.0;
-		Pt += vec2( 26.0, 161.0 ).xyxy;
-		Pt *= Pt;
-		Pt = Pt.xzxz * Pt.yyww;
-		vec4 hash_x = fract( Pt * ( 1.0 / 951.135664 ) );
-		vec4 hash_y = fract( Pt * ( 1.0 / 642.949883 ) );
-		hash_x = hash_x * 2.0 - 1.0;
-		hash_y = hash_y * 2.0 - 1.0;
-		const float JITTER_WINDOW = 0.25;
-		hash_x = ( ( hash_x * hash_x * hash_x ) - sign( hash_x ) ) * JITTER_WINDOW + vec4( 0.0, 1.0, 0.0, 1.0 );
-		hash_y = ( ( hash_y * hash_y * hash_y ) - sign( hash_y ) ) * JITTER_WINDOW + vec4( 0.0, 0.0, 1.0, 1.0 );
-		vec4 dx = Pf.xxxx - hash_x;
-		vec4 dy = Pf.yyyy - hash_y;
-		vec4 d = dx * dx + dy * dy;
-		d.xy = min(d.xy, d.zw);
-		return min(d.x, d.y) * ( 1.0 / 1.125 );
-	}
+   float rand(vec2 co)
+   {
+      return fract(sin(dot(co.xy, vec2(12.9898, 78.233))) * 43758.5453);
+   }
 
-	float snow(vec2 pos, float time, float scale) {
-		// add wobble
-		pos.x += cos(pos.y*4.0 + time*3.14159*2.0 + 1.0/scale)/(8.0/scale);
-		// add gravity
-		pos += time*scale*vec2(-0.5, 1.0);
-		return max(
-			1.0 - Cellular2D(mod(pos/scale, scale*quality)*16.0)*1024.0,
-			0.0
-		) * (scale*0.5 + 0.5);
-	}
+   float dist_func(vec2 distv)
+   {
+      float dist = sqrt((distv.x * distv.x) + (distv.y * distv.y)) * (40.0 / baseScale);
+      dist = clamp(dist, 0.0, 1.0);
+      return cos(dist * (3.14159265358 * 0.5)) * 0.5;
+   }
 
-	void main( void ) {
-		float winscale = max(res.x, res.y);
-		float tim = mod(time/8.0, 84.0*quality);
-		vec2 pos = gl_FragCoord.xy/winscale;
-		float a = 0.0;
-		// Each of these is a layer of snow
-		// Remove some for better performance
-		// Changing the scale (3rd value) will mess with the looping
-		a += snow(pos, tim, 1.0);
-		a += snow(pos, tim, 0.7);
-		a += snow(pos, tim, 0.6);
-		a += snow(pos, tim, 0.5);
-		a += snow(pos, tim, 0.4);
-		a += snow(pos, tim, 0.3);
-		a += snow(pos, tim, 0.25);
-		a += snow(pos, tim, 0.125);
-		a = a * min(pos.y*4.0, 1.0);
-		gl_FragColor = vec4(1.0, 1.0, 1.0, a);
-	}
+   float random_dots(vec2 co)
+   {
+      float part = 1.0 / 20.0;
+      vec2 cd = floor(co / part);
+      float p = rand(cd);
+
+      if (p > 0.005 * (density * 40.0))
+         return 0.0;
+
+      vec2 dpos = (vec2(fract(p * 2.0) , p) + vec2(2.0, 2.0)) * 0.25;
+
+      vec2 cellpos = fract(co / part);
+      vec2 distv = (cellpos - dpos);
+
+      return dist_func(distv);
+   }
+
+   float snow(vec2 pos, float time, float scale)
+   {
+      // add wobble
+      pos.x += cos(pos.y * 1.2 + time * 3.14159 * 2.0 + 1.0 / scale) / (8.0 / scale) * 4.0;
+      // add gravity
+      pos += time * scale * vec2(-0.5, 1.0) * 4.0;
+      return random_dots(pos / scale) * (scale * 0.5 + 0.5);
+   }
+
+   void main(void)
+   {
+      float tim = time * 0.4 * speed;
+      vec2 pos = gl_FragCoord.xy / OutputSize.xx;
+      float a = 0.0;
+      // Each of these is a layer of snow
+      // Remove some for better performance
+      // Changing the scale (3rd value) will mess with the looping
+      a += snow(pos, tim, 1.0);
+      a += snow(pos, tim, 0.7);
+      a += snow(pos, tim, 0.6);
+      a += snow(pos, tim, 0.5);
+      a += snow(pos, tim, 0.4);
+      a += snow(pos, tim, 0.3);
+      a += snow(pos, tim, 0.25);
+      a += snow(pos, tim, 0.125);
+      a = a * min(pos.y * 4.0, 1.0);
+      gl_FragColor = vec4(1.0, 1.0, 1.0, a);
+   }
 
 );
diff --git a/menu/drivers_display/menu_display_gl.c b/menu/drivers_display/menu_display_gl.c
index 6048a1085c..f1d056862d 100644
--- a/menu/drivers_display/menu_display_gl.c
+++ b/menu/drivers_display/menu_display_gl.c
@@ -193,6 +193,13 @@ static void menu_display_gl_draw_pipeline(void *data)
             uniform_param.lookup.ident      = "time";
             uniform_param.result.f.v0       = t;
 
+            video_shader_driver_set_parameter(uniform_param);            
+
+            uniform_param.type              = UNIFORM_2F;
+            uniform_param.lookup.ident      = "OutputSize";
+            uniform_param.result.f.v0       = draw->width;
+            uniform_param.result.f.v1       = draw->height;
+
             video_shader_driver_set_parameter(uniform_param);
          }
          break;