diff options
Diffstat (limited to 'drivers/gles2/shaders')
-rw-r--r-- | drivers/gles2/shaders/SCsub | 22 | ||||
-rw-r--r-- | drivers/gles2/shaders/blend_shape.glsl | 197 | ||||
-rw-r--r-- | drivers/gles2/shaders/canvas.glsl | 141 | ||||
-rw-r--r-- | drivers/gles2/shaders/canvas_shadow.glsl | 49 | ||||
-rw-r--r-- | drivers/gles2/shaders/copy.glsl | 72 | ||||
-rw-r--r-- | drivers/gles2/shaders/cube_to_dp.glsl | 79 | ||||
-rw-r--r-- | drivers/gles2/shaders/cubemap_filter.glsl | 294 | ||||
-rw-r--r-- | drivers/gles2/shaders/effect_blur.glsl | 301 | ||||
-rw-r--r-- | drivers/gles2/shaders/exposure.glsl | 98 | ||||
-rw-r--r-- | drivers/gles2/shaders/particles.glsl | 260 | ||||
-rw-r--r-- | drivers/gles2/shaders/resolve.glsl | 44 | ||||
-rw-r--r-- | drivers/gles2/shaders/scene.glsl | 2113 | ||||
-rw-r--r-- | drivers/gles2/shaders/screen_space_reflection.glsl | 318 | ||||
-rw-r--r-- | drivers/gles2/shaders/ssao.glsl | 293 | ||||
-rw-r--r-- | drivers/gles2/shaders/ssao_blur.glsl | 124 | ||||
-rw-r--r-- | drivers/gles2/shaders/ssao_minify.glsl | 59 | ||||
-rw-r--r-- | drivers/gles2/shaders/subsurf_scattering.glsl | 192 | ||||
-rw-r--r-- | drivers/gles2/shaders/tonemap.glsl | 323 |
18 files changed, 4979 insertions, 0 deletions
diff --git a/drivers/gles2/shaders/SCsub b/drivers/gles2/shaders/SCsub new file mode 100644 index 0000000000..5de3e1ac90 --- /dev/null +++ b/drivers/gles2/shaders/SCsub @@ -0,0 +1,22 @@ +#!/usr/bin/env python + +Import('env') + +if 'GLES2_GLSL' in env['BUILDERS']: + env.GLES2_GLSL('copy.glsl'); +# env.GLES2_GLSL('resolve.glsl'); + env.GLES2_GLSL('canvas.glsl'); +# env.GLES2_GLSL('canvas_shadow.glsl'); + env.GLES2_GLSL('scene.glsl'); +# env.GLES2_GLSL('cubemap_filter.glsl'); +# env.GLES2_GLSL('cube_to_dp.glsl'); +# env.GLES2_GLSL('blend_shape.glsl'); +# env.GLES2_GLSL('screen_space_reflection.glsl'); +# env.GLES2_GLSL('effect_blur.glsl'); +# env.GLES2_GLSL('subsurf_scattering.glsl'); +# env.GLES2_GLSL('ssao.glsl'); +# env.GLES2_GLSL('ssao_minify.glsl'); +# env.GLES2_GLSL('ssao_blur.glsl'); +# env.GLES2_GLSL('exposure.glsl'); +# env.GLES2_GLSL('tonemap.glsl'); +# env.GLES2_GLSL('particles.glsl'); diff --git a/drivers/gles2/shaders/blend_shape.glsl b/drivers/gles2/shaders/blend_shape.glsl new file mode 100644 index 0000000000..4e0d066823 --- /dev/null +++ b/drivers/gles2/shaders/blend_shape.glsl @@ -0,0 +1,197 @@ +[vertex] + + +/* +from VisualServer: + +ARRAY_VERTEX=0, +ARRAY_NORMAL=1, +ARRAY_TANGENT=2, +ARRAY_COLOR=3, +ARRAY_TEX_UV=4, +ARRAY_TEX_UV2=5, +ARRAY_BONES=6, +ARRAY_WEIGHTS=7, +ARRAY_INDEX=8, +*/ + +#ifdef USE_2D_VERTEX +#define VFORMAT vec2 +#else +#define VFORMAT vec3 +#endif + +/* INPUT ATTRIBS */ + +layout(location=0) in highp VFORMAT vertex_attrib; +layout(location=1) in vec3 normal_attrib; + +#ifdef ENABLE_TANGENT +layout(location=2) in vec4 tangent_attrib; +#endif + +#ifdef ENABLE_COLOR +layout(location=3) in vec4 color_attrib; +#endif + +#ifdef ENABLE_UV +layout(location=4) in vec2 uv_attrib; +#endif + +#ifdef ENABLE_UV2 +layout(location=5) in vec2 uv2_attrib; +#endif + +#ifdef ENABLE_SKELETON +layout(location=6) in ivec4 bone_attrib; +layout(location=7) in vec4 weight_attrib; +#endif + +/* BLEND ATTRIBS */ + +#ifdef ENABLE_BLEND + +layout(location=8) in highp VFORMAT vertex_attrib_blend; +layout(location=9) in vec3 normal_attrib_blend; + +#ifdef ENABLE_TANGENT +layout(location=10) in vec4 tangent_attrib_blend; +#endif + +#ifdef ENABLE_COLOR +layout(location=11) in vec4 color_attrib_blend; +#endif + +#ifdef ENABLE_UV +layout(location=12) in vec2 uv_attrib_blend; +#endif + +#ifdef ENABLE_UV2 +layout(location=13) in vec2 uv2_attrib_blend; +#endif + +#ifdef ENABLE_SKELETON +layout(location=14) in ivec4 bone_attrib_blend; +layout(location=15) in vec4 weight_attrib_blend; +#endif + +#endif + +/* OUTPUTS */ + +out VFORMAT vertex_out; //tfb: + +#ifdef ENABLE_NORMAL +out vec3 normal_out; //tfb:ENABLE_NORMAL +#endif + +#ifdef ENABLE_TANGENT +out vec4 tangent_out; //tfb:ENABLE_TANGENT +#endif + +#ifdef ENABLE_COLOR +out vec4 color_out; //tfb:ENABLE_COLOR +#endif + +#ifdef ENABLE_UV +out vec2 uv_out; //tfb:ENABLE_UV +#endif + +#ifdef ENABLE_UV2 +out vec2 uv2_out; //tfb:ENABLE_UV2 +#endif + +#ifdef ENABLE_SKELETON +out ivec4 bone_out; //tfb:ENABLE_SKELETON +out vec4 weight_out; //tfb:ENABLE_SKELETON +#endif + +uniform float blend_amount; + +void main() { + + +#ifdef ENABLE_BLEND + + vertex_out = vertex_attrib_blend + vertex_attrib * blend_amount; + +#ifdef ENABLE_NORMAL + normal_out = normal_attrib_blend + normal_attrib * blend_amount; +#endif + +#ifdef ENABLE_TANGENT + + tangent_out.xyz = tangent_attrib_blend.xyz + tangent_attrib.xyz * blend_amount; + tangent_out.w = tangent_attrib_blend.w; //just copy, no point in blending his +#endif + +#ifdef ENABLE_COLOR + + color_out = color_attrib_blend + color_attrib * blend_amount; +#endif + +#ifdef ENABLE_UV + + uv_out = uv_attrib_blend + uv_attrib * blend_amount; +#endif + +#ifdef ENABLE_UV2 + + uv2_out = uv2_attrib_blend + uv2_attrib * blend_amount; +#endif + + +#ifdef ENABLE_SKELETON + + bone_out = bone_attrib_blend; + weight_out = weight_attrib_blend + weight_attrib * blend_amount; +#endif + +#else //ENABLE_BLEND + + + vertex_out = vertex_attrib * blend_amount; + +#ifdef ENABLE_NORMAL + normal_out = normal_attrib * blend_amount; +#endif + +#ifdef ENABLE_TANGENT + + tangent_out.xyz = tangent_attrib.xyz * blend_amount; + tangent_out.w = tangent_attrib.w; //just copy, no point in blending his +#endif + +#ifdef ENABLE_COLOR + + color_out = color_attrib * blend_amount; +#endif + +#ifdef ENABLE_UV + + uv_out = uv_attrib * blend_amount; +#endif + +#ifdef ENABLE_UV2 + + uv2_out = uv2_attrib * blend_amount; +#endif + + +#ifdef ENABLE_SKELETON + + bone_out = bone_attrib; + weight_out = weight_attrib * blend_amount; +#endif + +#endif + gl_Position = vec4(0.0); +} + +[fragment] + + +void main() { + +} + diff --git a/drivers/gles2/shaders/canvas.glsl b/drivers/gles2/shaders/canvas.glsl new file mode 100644 index 0000000000..11c6ab9b76 --- /dev/null +++ b/drivers/gles2/shaders/canvas.glsl @@ -0,0 +1,141 @@ +[vertex] + +#ifdef USE_GLES_OVER_GL +#define mediump +#define highp +#else +precision mediump float; +precision mediump int; +#endif + +uniform highp mat4 projection_matrix; +uniform highp mat4 modelview_matrix; +uniform highp mat4 extra_matrix; +attribute highp vec2 vertex; // attrib:0 +attribute vec4 color_attrib; // attrib:3 +attribute vec2 uv_attrib; // attrib:4 + +varying vec2 uv_interp; +varying vec4 color_interp; + +uniform highp vec2 color_texpixel_size; + +#ifdef USE_TEXTURE_RECT + +uniform vec4 dst_rect; +uniform vec4 src_rect; + +#endif + +uniform bool blit_pass; + +VERTEX_SHADER_GLOBALS + +vec2 select(vec2 a, vec2 b, bvec2 c) { + vec2 ret; + + ret.x = c.x ? b.x : a.x; + ret.y = c.y ? b.y : a.y; + + return ret; +} + +void main() { + + vec4 color = color_attrib; + +#ifdef USE_TEXTURE_RECT + + if (dst_rect.z < 0.0) { // Transpose is encoded as negative dst_rect.z + uv_interp = src_rect.xy + abs(src_rect.zw) * vertex.yx; + } else { + uv_interp = src_rect.xy + abs(src_rect.zw) * vertex; + } + + vec4 outvec = vec4(0.0, 0.0, 0.0, 1.0); + + // This is what is done in the GLES 3 bindings and should + // take care of flipped rects. + // + // But it doesn't. + // I don't know why, will need to investigate further. + + outvec.xy = dst_rect.xy + abs(dst_rect.zw) * select(vertex, vec2(1.0, 1.0) - vertex, lessThan(src_rect.zw, vec2(0.0, 0.0))); + + // outvec.xy = dst_rect.xy + abs(dst_rect.zw) * vertex; +#else + vec4 outvec = vec4(vertex.xy, 0.0, 1.0); + +#ifdef USE_UV_ATTRIBUTE + uv_interp = uv_attrib; +#else + uv_interp = vertex.xy; +#endif + +#endif + +{ + vec2 src_vtx=outvec.xy; +VERTEX_SHADER_CODE + +} + + color_interp = color; + + gl_Position = projection_matrix * modelview_matrix * outvec; + +} + +[fragment] + +#ifdef USE_GLES_OVER_GL +#define mediump +#define highp +#else +precision mediump float; +precision mediump int; +#endif + +uniform sampler2D color_texture; // texunit:0 +uniform highp vec2 color_texpixel_size; +uniform mediump sampler2D normal_texture; // texunit:1 + +varying mediump vec2 uv_interp; +varying mediump vec4 color_interp; + +uniform bool blit_pass; + +uniform vec4 final_modulate; + +#ifdef SCREEN_TEXTURE_USED + +uniform sampler2D screen_texture; // texunit:2 + +#endif + +#ifdef SCREEN_UV_USED + +uniform vec2 screen_pixel_size; + +#endif + +FRAGMENT_SHADER_GLOBALS + + +void main() { + + vec4 color = color_interp; + + color *= texture2D(color_texture, uv_interp); +{ + +FRAGMENT_SHADER_CODE + + +} + + color *= final_modulate; + + gl_FragColor = color; + +} diff --git a/drivers/gles2/shaders/canvas_shadow.glsl b/drivers/gles2/shaders/canvas_shadow.glsl new file mode 100644 index 0000000000..c757990de0 --- /dev/null +++ b/drivers/gles2/shaders/canvas_shadow.glsl @@ -0,0 +1,49 @@ +[vertex] + + + +uniform highp mat4 projection_matrix; +uniform highp mat4 light_matrix; +uniform highp mat4 world_matrix; +uniform highp float distance_norm; + +layout(location=0) in highp vec3 vertex; + +out highp vec4 position_interp; + +void main() { + + gl_Position = projection_matrix * (light_matrix * (world_matrix * vec4(vertex,1.0))); + position_interp=gl_Position; +} + +[fragment] + +in highp vec4 position_interp; + +#ifdef USE_RGBA_SHADOWS + +layout(location=0) out lowp vec4 distance_buf; + +#else + +layout(location=0) out highp float distance_buf; + +#endif + +void main() { + + highp float depth = ((position_interp.z / position_interp.w) + 1.0) * 0.5 + 0.0;//bias; + +#ifdef USE_RGBA_SHADOWS + + highp vec4 comp = fract(depth * vec4(256.0 * 256.0 * 256.0, 256.0 * 256.0, 256.0, 1.0)); + comp -= comp.xxyz * vec4(0, 1.0 / 256.0, 1.0 / 256.0, 1.0 / 256.0); + distance_buf=comp; +#else + + distance_buf=depth; + +#endif +} + diff --git a/drivers/gles2/shaders/copy.glsl b/drivers/gles2/shaders/copy.glsl new file mode 100644 index 0000000000..a21da68525 --- /dev/null +++ b/drivers/gles2/shaders/copy.glsl @@ -0,0 +1,72 @@ +[vertex] + +#ifdef USE_GLES_OVER_GL +#define mediump +#define highp +#else +precision mediump float; +precision mediump int; +#endif + +attribute highp vec4 vertex_attrib; // attrib:0 +attribute vec2 uv_in; // attrib:4 +attribute vec2 uv2_in; // attrib:5 + +varying vec2 uv_interp; + +varying vec2 uv2_interp; + +#ifdef USE_COPY_SECTION +uniform vec4 copy_section; +#endif + +void main() { + + uv_interp = uv_in; + uv2_interp = uv2_in; + gl_Position = vertex_attrib; + +#ifdef USE_COPY_SECTION + uv_interp = copy_section.xy + uv_interp * copy_section.zw; + gl_Position.xy = (copy_section.xy + (gl_Position.xy * 0.5 + 0.5) * copy_section.zw) * 2.0 - 1.0; +#endif +} + +[fragment] + +#ifdef USE_GLES_OVER_GL +#define mediump +#define highp +#else +precision mediump float; +precision mediump int; +#endif + + +varying vec2 uv_interp; +uniform sampler2D source; // texunit:0 + +varying vec2 uv2_interp; + +#ifdef USE_CUSTOM_ALPHA +uniform float custom_alpha; +#endif + + +void main() { + + //vec4 color = color_interp; + vec4 color = texture2D( source, uv_interp ); + + +#ifdef USE_NO_ALPHA + color.a=1.0; +#endif + +#ifdef USE_CUSTOM_ALPHA + color.a=custom_alpha; +#endif + + + gl_FragColor = color; +} diff --git a/drivers/gles2/shaders/cube_to_dp.glsl b/drivers/gles2/shaders/cube_to_dp.glsl new file mode 100644 index 0000000000..5ffc78c0b9 --- /dev/null +++ b/drivers/gles2/shaders/cube_to_dp.glsl @@ -0,0 +1,79 @@ +[vertex] + + +layout(location=0) in highp vec4 vertex_attrib; +layout(location=4) in vec2 uv_in; + +out vec2 uv_interp; + +void main() { + + uv_interp = uv_in; + gl_Position = vertex_attrib; +} + +[fragment] + + +uniform highp samplerCube source_cube; //texunit:0 +in vec2 uv_interp; + +uniform bool z_flip; +uniform highp float z_far; +uniform highp float z_near; +uniform highp float bias; + +void main() { + + highp vec3 normal = vec3( uv_interp * 2.0 - 1.0, 0.0 ); +/* + if(z_flip) { + normal.z = 0.5 - 0.5*((normal.x * normal.x) + (normal.y * normal.y)); + } else { + normal.z = -0.5 + 0.5*((normal.x * normal.x) + (normal.y * normal.y)); + } +*/ + + //normal.z = sqrt(1.0-dot(normal.xy,normal.xy)); + //normal.xy*=1.0+normal.z; + + normal.z = 0.5 - 0.5*((normal.x * normal.x) + (normal.y * normal.y)); + normal = normalize(normal); + +/* + normal.z=0.5; + normal=normalize(normal); +*/ + if (!z_flip) { + normal.z=-normal.z; + } + + //normal = normalize(vec3( uv_interp * 2.0 - 1.0, 1.0 )); + float depth = texture(source_cube,normal).r; + + // absolute values for direction cosines, bigger value equals closer to basis axis + vec3 unorm = abs(normal); + + if ( (unorm.x >= unorm.y) && (unorm.x >= unorm.z) ) { + // x code + unorm = normal.x > 0.0 ? vec3( 1.0, 0.0, 0.0 ) : vec3( -1.0, 0.0, 0.0 ) ; + } else if ( (unorm.y > unorm.x) && (unorm.y >= unorm.z) ) { + // y code + unorm = normal.y > 0.0 ? vec3( 0.0, 1.0, 0.0 ) : vec3( 0.0, -1.0, 0.0 ) ; + } else if ( (unorm.z > unorm.x) && (unorm.z > unorm.y) ) { + // z code + unorm = normal.z > 0.0 ? vec3( 0.0, 0.0, 1.0 ) : vec3( 0.0, 0.0, -1.0 ) ; + } else { + // oh-no we messed up code + // has to be + unorm = vec3( 1.0, 0.0, 0.0 ); + } + + float depth_fix = 1.0 / dot(normal,unorm); + + + depth = 2.0 * depth - 1.0; + float linear_depth = 2.0 * z_near * z_far / (z_far + z_near - depth * (z_far - z_near)); + gl_FragDepth = (linear_depth*depth_fix+bias) / z_far; +} + diff --git a/drivers/gles2/shaders/cubemap_filter.glsl b/drivers/gles2/shaders/cubemap_filter.glsl new file mode 100644 index 0000000000..485fbb6ee0 --- /dev/null +++ b/drivers/gles2/shaders/cubemap_filter.glsl @@ -0,0 +1,294 @@ +[vertex] + + +layout(location=0) in highp vec2 vertex; + +layout(location=4) in highp vec2 uv; + +out highp vec2 uv_interp; + +void main() { + + uv_interp=uv; + gl_Position=vec4(vertex,0,1); +} + +[fragment] + + +precision highp float; +precision highp int; + +#ifdef USE_SOURCE_PANORAMA +uniform sampler2D source_panorama; //texunit:0 +#endif + +#ifdef USE_SOURCE_DUAL_PARABOLOID_ARRAY +uniform sampler2DArray source_dual_paraboloid_array; //texunit:0 +uniform int source_array_index; +#endif + +#if !defined(USE_SOURCE_DUAL_PARABOLOID_ARRAY) && !defined(USE_SOURCE_PANORAMA) +uniform samplerCube source_cube; //texunit:0 +#endif + +uniform int face_id; +uniform float roughness; +in highp vec2 uv_interp; + + +layout(location = 0) out vec4 frag_color; + + +#define M_PI 3.14159265359 + + +vec3 texelCoordToVec(vec2 uv, int faceID) +{ + mat3 faceUvVectors[6]; +/* + // -x + faceUvVectors[1][0] = vec3(0.0, 0.0, 1.0); // u -> +z + faceUvVectors[1][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[1][2] = vec3(-1.0, 0.0, 0.0); // -x face + + // +x + faceUvVectors[0][0] = vec3(0.0, 0.0, -1.0); // u -> -z + faceUvVectors[0][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[0][2] = vec3(1.0, 0.0, 0.0); // +x face + + // -y + faceUvVectors[3][0] = vec3(1.0, 0.0, 0.0); // u -> +x + faceUvVectors[3][1] = vec3(0.0, 0.0, -1.0); // v -> -z + faceUvVectors[3][2] = vec3(0.0, -1.0, 0.0); // -y face + + // +y + faceUvVectors[2][0] = vec3(1.0, 0.0, 0.0); // u -> +x + faceUvVectors[2][1] = vec3(0.0, 0.0, 1.0); // v -> +z + faceUvVectors[2][2] = vec3(0.0, 1.0, 0.0); // +y face + + // -z + faceUvVectors[5][0] = vec3(-1.0, 0.0, 0.0); // u -> -x + faceUvVectors[5][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[5][2] = vec3(0.0, 0.0, -1.0); // -z face + + // +z + faceUvVectors[4][0] = vec3(1.0, 0.0, 0.0); // u -> +x + faceUvVectors[4][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[4][2] = vec3(0.0, 0.0, 1.0); // +z face +*/ + + // -x + faceUvVectors[0][0] = vec3(0.0, 0.0, 1.0); // u -> +z + faceUvVectors[0][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[0][2] = vec3(-1.0, 0.0, 0.0); // -x face + + // +x + faceUvVectors[1][0] = vec3(0.0, 0.0, -1.0); // u -> -z + faceUvVectors[1][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[1][2] = vec3(1.0, 0.0, 0.0); // +x face + + // -y + faceUvVectors[2][0] = vec3(1.0, 0.0, 0.0); // u -> +x + faceUvVectors[2][1] = vec3(0.0, 0.0, -1.0); // v -> -z + faceUvVectors[2][2] = vec3(0.0, -1.0, 0.0); // -y face + + // +y + faceUvVectors[3][0] = vec3(1.0, 0.0, 0.0); // u -> +x + faceUvVectors[3][1] = vec3(0.0, 0.0, 1.0); // v -> +z + faceUvVectors[3][2] = vec3(0.0, 1.0, 0.0); // +y face + + // -z + faceUvVectors[4][0] = vec3(-1.0, 0.0, 0.0); // u -> -x + faceUvVectors[4][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[4][2] = vec3(0.0, 0.0, -1.0); // -z face + + // +z + faceUvVectors[5][0] = vec3(1.0, 0.0, 0.0); // u -> +x + faceUvVectors[5][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[5][2] = vec3(0.0, 0.0, 1.0); // +z face + + // out = u * s_faceUv[0] + v * s_faceUv[1] + s_faceUv[2]. + vec3 result = (faceUvVectors[faceID][0] * uv.x) + (faceUvVectors[faceID][1] * uv.y) + faceUvVectors[faceID][2]; + return normalize(result); +} + +vec3 ImportanceSampleGGX(vec2 Xi, float Roughness, vec3 N) +{ + float a = Roughness * Roughness; // DISNEY'S ROUGHNESS [see Burley'12 siggraph] + + // Compute distribution direction + float Phi = 2.0 * M_PI * Xi.x; + float CosTheta = sqrt((1.0 - Xi.y) / (1.0 + (a*a - 1.0) * Xi.y)); + float SinTheta = sqrt(1.0 - CosTheta * CosTheta); + + // Convert to spherical direction + vec3 H; + H.x = SinTheta * cos(Phi); + H.y = SinTheta * sin(Phi); + H.z = CosTheta; + + vec3 UpVector = abs(N.z) < 0.999 ? vec3(0.0, 0.0, 1.0) : vec3(1.0, 0.0, 0.0); + vec3 TangentX = normalize(cross(UpVector, N)); + vec3 TangentY = cross(N, TangentX); + + // Tangent to world space + return TangentX * H.x + TangentY * H.y + N * H.z; +} + +// http://graphicrants.blogspot.com.au/2013/08/specular-brdf-reference.html +float GGX(float NdotV, float a) +{ + float k = a / 2.0; + return NdotV / (NdotV * (1.0 - k) + k); +} + +// http://graphicrants.blogspot.com.au/2013/08/specular-brdf-reference.html +float G_Smith(float a, float nDotV, float nDotL) +{ + return GGX(nDotL, a * a) * GGX(nDotV, a * a); +} + +float radicalInverse_VdC(uint bits) { + bits = (bits << 16u) | (bits >> 16u); + bits = ((bits & 0x55555555u) << 1u) | ((bits & 0xAAAAAAAAu) >> 1u); + bits = ((bits & 0x33333333u) << 2u) | ((bits & 0xCCCCCCCCu) >> 2u); + bits = ((bits & 0x0F0F0F0Fu) << 4u) | ((bits & 0xF0F0F0F0u) >> 4u); + bits = ((bits & 0x00FF00FFu) << 8u) | ((bits & 0xFF00FF00u) >> 8u); + return float(bits) * 2.3283064365386963e-10; // / 0x100000000 +} + +vec2 Hammersley(uint i, uint N) { + return vec2(float(i)/float(N), radicalInverse_VdC(i)); +} + + + +#ifdef LOW_QUALITY + +#define SAMPLE_COUNT 64u + +#else + +#define SAMPLE_COUNT 512u + +#endif + +uniform bool z_flip; + +#ifdef USE_SOURCE_PANORAMA + +vec4 texturePanorama(vec3 normal,sampler2D pano ) { + + vec2 st = vec2( + atan(normal.x, normal.z), + acos(normal.y) + ); + + if(st.x < 0.0) + st.x += M_PI*2.0; + + st/=vec2(M_PI*2.0,M_PI); + + return textureLod(pano,st,0.0); + +} + +#endif + +#ifdef USE_SOURCE_DUAL_PARABOLOID_ARRAY + + +vec4 textureDualParaboloidArray(vec3 normal) { + + vec3 norm = normalize(normal); + norm.xy/=1.0+abs(norm.z); + norm.xy=norm.xy * vec2(0.5,0.25) + vec2(0.5,0.25); + if (norm.z<0.0) { + norm.y=0.5-norm.y+0.5; + } + return textureLod(source_dual_paraboloid_array, vec3(norm.xy, float(source_array_index) ), 0.0); + +} + +#endif + +void main() { + +#ifdef USE_DUAL_PARABOLOID + + vec3 N = vec3( uv_interp * 2.0 - 1.0, 0.0 ); + N.z = 0.5 - 0.5*((N.x * N.x) + (N.y * N.y)); + N = normalize(N); + + if (z_flip) { + N.y=-N.y; //y is flipped to improve blending between both sides + N.z=-N.z; + } + + +#else + vec2 uv = (uv_interp * 2.0) - 1.0; + vec3 N = texelCoordToVec(uv, face_id); +#endif + //vec4 color = color_interp; + +#ifdef USE_DIRECT_WRITE + +#ifdef USE_SOURCE_PANORAMA + + frag_color=vec4(texturePanorama(N,source_panorama).rgb,1.0); +#endif + +#ifdef USE_SOURCE_DUAL_PARABOLOID_ARRAY + + frag_color=vec4(textureDualParaboloidArray(N).rgb,1.0); +#endif + +#if !defined(USE_SOURCE_DUAL_PARABOLOID_ARRAY) && !defined(USE_SOURCE_PANORAMA) + + N.y=-N.y; + frag_color=vec4(texture(N,source_cube).rgb,1.0); +#endif + + + + +#else + + vec4 sum = vec4(0.0, 0.0, 0.0, 0.0); + + for(uint sampleNum = 0u; sampleNum < SAMPLE_COUNT; sampleNum++) { + vec2 xi = Hammersley(sampleNum, SAMPLE_COUNT); + + vec3 H = ImportanceSampleGGX( xi, roughness, N ); + vec3 V = N; + vec3 L = normalize(2.0 * dot( V, H ) * H - V); + + float ndotl = clamp(dot(N, L),0.0,1.0); + + if (ndotl>0.0) { +#ifdef USE_SOURCE_PANORAMA + sum.rgb += texturePanorama(H,source_panorama).rgb *ndotl; +#endif + +#ifdef USE_SOURCE_DUAL_PARABOLOID_ARRAY + + sum.rgb += textureDualParaboloidArray(H).rgb *ndotl; +#endif + +#if !defined(USE_SOURCE_DUAL_PARABOLOID_ARRAY) && !defined(USE_SOURCE_PANORAMA) + H.y=-H.y; + sum.rgb += textureLod(source_cube, H, 0.0).rgb *ndotl; +#endif + sum.a += ndotl; + } + } + sum /= sum.a; + + frag_color = vec4(sum.rgb, 1.0); + +#endif + +} + diff --git a/drivers/gles2/shaders/effect_blur.glsl b/drivers/gles2/shaders/effect_blur.glsl new file mode 100644 index 0000000000..b5f98a1244 --- /dev/null +++ b/drivers/gles2/shaders/effect_blur.glsl @@ -0,0 +1,301 @@ +[vertex] + + +layout(location=0) in highp vec4 vertex_attrib; +layout(location=4) in vec2 uv_in; + +out vec2 uv_interp; + +#ifdef USE_BLUR_SECTION + +uniform vec4 blur_section; + +#endif + +void main() { + + uv_interp = uv_in; + gl_Position = vertex_attrib; +#ifdef USE_BLUR_SECTION + + uv_interp = blur_section.xy + uv_interp * blur_section.zw; + gl_Position.xy = (blur_section.xy + (gl_Position.xy * 0.5 + 0.5) * blur_section.zw) * 2.0 - 1.0; +#endif +} + +[fragment] + +#if !defined(GLES_OVER_GL) +precision mediump float; +#endif + +in vec2 uv_interp; +uniform sampler2D source_color; //texunit:0 + +#ifdef SSAO_MERGE +uniform sampler2D source_ssao; //texunit:1 +#endif + +uniform float lod; +uniform vec2 pixel_size; + + +layout(location = 0) out vec4 frag_color; + +#ifdef SSAO_MERGE + +uniform vec4 ssao_color; + +#endif + +#if defined (GLOW_GAUSSIAN_HORIZONTAL) || defined(GLOW_GAUSSIAN_VERTICAL) + +uniform float glow_strength; + +#endif + +#if defined(DOF_FAR_BLUR) || defined (DOF_NEAR_BLUR) + +#ifdef DOF_QUALITY_LOW +const int dof_kernel_size=5; +const int dof_kernel_from=2; +const float dof_kernel[5] = float[] (0.153388,0.221461,0.250301,0.221461,0.153388); +#endif + +#ifdef DOF_QUALITY_MEDIUM +const int dof_kernel_size=11; +const int dof_kernel_from=5; +const float dof_kernel[11] = float[] (0.055037,0.072806,0.090506,0.105726,0.116061,0.119726,0.116061,0.105726,0.090506,0.072806,0.055037); + +#endif + +#ifdef DOF_QUALITY_HIGH +const int dof_kernel_size=21; +const int dof_kernel_from=10; +const float dof_kernel[21] = float[] (0.028174,0.032676,0.037311,0.041944,0.046421,0.050582,0.054261,0.057307,0.059587,0.060998,0.061476,0.060998,0.059587,0.057307,0.054261,0.050582,0.046421,0.041944,0.037311,0.032676,0.028174); +#endif + +uniform sampler2D dof_source_depth; //texunit:1 +uniform float dof_begin; +uniform float dof_end; +uniform vec2 dof_dir; +uniform float dof_radius; + +#ifdef DOF_NEAR_BLUR_MERGE + +uniform sampler2D source_dof_original; //texunit:2 +#endif + +#endif + + +#ifdef GLOW_FIRST_PASS + +uniform float exposure; +uniform float white; + +#ifdef GLOW_USE_AUTO_EXPOSURE + +uniform highp sampler2D source_auto_exposure; //texunit:1 +uniform highp float auto_exposure_grey; + +#endif + +uniform float glow_bloom; +uniform float glow_hdr_threshold; +uniform float glow_hdr_scale; + +#endif + +uniform float camera_z_far; +uniform float camera_z_near; + +void main() { + + + +#ifdef GAUSSIAN_HORIZONTAL + vec2 pix_size = pixel_size; + pix_size*=0.5; //reading from larger buffer, so use more samples + vec4 color =textureLod( source_color, uv_interp+vec2( 0.0, 0.0)*pix_size,lod )*0.214607; + color+=textureLod( source_color, uv_interp+vec2( 1.0, 0.0)*pix_size,lod )*0.189879; + color+=textureLod( source_color, uv_interp+vec2( 2.0, 0.0)*pix_size,lod )*0.157305; + color+=textureLod( source_color, uv_interp+vec2( 3.0, 0.0)*pix_size,lod )*0.071303; + color+=textureLod( source_color, uv_interp+vec2(-1.0, 0.0)*pix_size,lod )*0.189879; + color+=textureLod( source_color, uv_interp+vec2(-2.0, 0.0)*pix_size,lod )*0.157305; + color+=textureLod( source_color, uv_interp+vec2(-3.0, 0.0)*pix_size,lod )*0.071303; + frag_color = color; +#endif + +#ifdef GAUSSIAN_VERTICAL + vec4 color =textureLod( source_color, uv_interp+vec2( 0.0, 0.0)*pixel_size,lod )*0.38774; + color+=textureLod( source_color, uv_interp+vec2( 0.0, 1.0)*pixel_size,lod )*0.24477; + color+=textureLod( source_color, uv_interp+vec2( 0.0, 2.0)*pixel_size,lod )*0.06136; + color+=textureLod( source_color, uv_interp+vec2( 0.0,-1.0)*pixel_size,lod )*0.24477; + color+=textureLod( source_color, uv_interp+vec2( 0.0,-2.0)*pixel_size,lod )*0.06136; + frag_color = color; +#endif + +//glow uses larger sigma for a more rounded blur effect + +#ifdef GLOW_GAUSSIAN_HORIZONTAL + vec2 pix_size = pixel_size; + pix_size*=0.5; //reading from larger buffer, so use more samples + vec4 color =textureLod( source_color, uv_interp+vec2( 0.0, 0.0)*pix_size,lod )*0.174938; + color+=textureLod( source_color, uv_interp+vec2( 1.0, 0.0)*pix_size,lod )*0.165569; + color+=textureLod( source_color, uv_interp+vec2( 2.0, 0.0)*pix_size,lod )*0.140367; + color+=textureLod( source_color, uv_interp+vec2( 3.0, 0.0)*pix_size,lod )*0.106595; + color+=textureLod( source_color, uv_interp+vec2(-1.0, 0.0)*pix_size,lod )*0.165569; + color+=textureLod( source_color, uv_interp+vec2(-2.0, 0.0)*pix_size,lod )*0.140367; + color+=textureLod( source_color, uv_interp+vec2(-3.0, 0.0)*pix_size,lod )*0.106595; + color*=glow_strength; + frag_color = color; +#endif + +#ifdef GLOW_GAUSSIAN_VERTICAL + vec4 color =textureLod( source_color, uv_interp+vec2(0.0, 0.0)*pixel_size,lod )*0.288713; + color+=textureLod( source_color, uv_interp+vec2(0.0, 1.0)*pixel_size,lod )*0.233062; + color+=textureLod( source_color, uv_interp+vec2(0.0, 2.0)*pixel_size,lod )*0.122581; + color+=textureLod( source_color, uv_interp+vec2(0.0,-1.0)*pixel_size,lod )*0.233062; + color+=textureLod( source_color, uv_interp+vec2(0.0,-2.0)*pixel_size,lod )*0.122581; + color*=glow_strength; + frag_color = color; +#endif + +#ifdef DOF_FAR_BLUR + + vec4 color_accum = vec4(0.0); + + float depth = textureLod( dof_source_depth, uv_interp, 0.0).r; + depth = depth * 2.0 - 1.0; +#ifdef USE_ORTHOGONAL_PROJECTION + depth = ((depth + (camera_z_far + camera_z_near)/(camera_z_far - camera_z_near)) * (camera_z_far - camera_z_near))/2.0; +#else + depth = 2.0 * camera_z_near * camera_z_far / (camera_z_far + camera_z_near - depth * (camera_z_far - camera_z_near)); +#endif + + float amount = smoothstep(dof_begin,dof_end,depth); + float k_accum=0.0; + + for(int i=0;i<dof_kernel_size;i++) { + + int int_ofs = i-dof_kernel_from; + vec2 tap_uv = uv_interp + dof_dir * float(int_ofs) * amount * dof_radius; + + float tap_k = dof_kernel[i]; + + float tap_depth = texture( dof_source_depth, tap_uv, 0.0).r; + tap_depth = tap_depth * 2.0 - 1.0; +#ifdef USE_ORTHOGONAL_PROJECTION + tap_depth = ((tap_depth + (camera_z_far + camera_z_near)/(camera_z_far - camera_z_near)) * (camera_z_far - camera_z_near))/2.0; +#else + tap_depth = 2.0 * camera_z_near * camera_z_far / (camera_z_far + camera_z_near - tap_depth * (camera_z_far - camera_z_near)); +#endif + float tap_amount = mix(smoothstep(dof_begin,dof_end,tap_depth),1.0,int_ofs==0); + tap_amount*=tap_amount*tap_amount; //prevent undesired glow effect + + vec4 tap_color = textureLod( source_color, tap_uv, 0.0) * tap_k; + + k_accum+=tap_k*tap_amount; + color_accum+=tap_color*tap_amount; + + + } + + if (k_accum>0.0) { + color_accum/=k_accum; + } + + frag_color = color_accum;///k_accum; + +#endif + +#ifdef DOF_NEAR_BLUR + + vec4 color_accum = vec4(0.0); + + float max_accum=0; + + for(int i=0;i<dof_kernel_size;i++) { + + int int_ofs = i-dof_kernel_from; + vec2 tap_uv = uv_interp + dof_dir * float(int_ofs) * dof_radius; + float ofs_influence = max(0.0,1.0-float(abs(int_ofs))/float(dof_kernel_from)); + + float tap_k = dof_kernel[i]; + + vec4 tap_color = textureLod( source_color, tap_uv, 0.0); + + float tap_depth = texture( dof_source_depth, tap_uv, 0.0).r; + tap_depth = tap_depth * 2.0 - 1.0; +#ifdef USE_ORTHOGONAL_PROJECTION + tap_depth = ((tap_depth + (camera_z_far + camera_z_near)/(camera_z_far - camera_z_near)) * (camera_z_far - camera_z_near))/2.0; +#else + tap_depth = 2.0 * camera_z_near * camera_z_far / (camera_z_far + camera_z_near - tap_depth * (camera_z_far - camera_z_near)); +#endif + float tap_amount = 1.0-smoothstep(dof_end,dof_begin,tap_depth); + tap_amount*=tap_amount*tap_amount; //prevent undesired glow effect + +#ifdef DOF_NEAR_FIRST_TAP + + tap_color.a= 1.0-smoothstep(dof_end,dof_begin,tap_depth); + +#endif + + max_accum=max(max_accum,tap_amount*ofs_influence); + + color_accum+=tap_color*tap_k; + + } + + color_accum.a=max(color_accum.a,sqrt(max_accum)); + + +#ifdef DOF_NEAR_BLUR_MERGE + + vec4 original = textureLod( source_dof_original, uv_interp, 0.0); + color_accum = mix(original,color_accum,color_accum.a); + +#endif + +#ifndef DOF_NEAR_FIRST_TAP + //color_accum=vec4(vec3(color_accum.a),1.0); +#endif + frag_color = color_accum; + +#endif + + + +#ifdef GLOW_FIRST_PASS + +#ifdef GLOW_USE_AUTO_EXPOSURE + + frag_color/=texelFetch(source_auto_exposure,ivec2(0,0),0).r/auto_exposure_grey; +#endif + frag_color*=exposure; + + float luminance = max(frag_color.r,max(frag_color.g,frag_color.b)); + float feedback = max( smoothstep(glow_hdr_threshold,glow_hdr_threshold+glow_hdr_scale,luminance), glow_bloom ); + + frag_color *= feedback; + +#endif + + +#ifdef SIMPLE_COPY + vec4 color =textureLod( source_color, uv_interp,0.0); + frag_color = color; +#endif + +#ifdef SSAO_MERGE + + vec4 color =textureLod( source_color, uv_interp,0.0); + float ssao =textureLod( source_ssao, uv_interp,0.0).r; + + frag_color = vec4( mix(color.rgb,color.rgb*mix(ssao_color.rgb,vec3(1.0),ssao),color.a), 1.0 ); + +#endif + + +} diff --git a/drivers/gles2/shaders/exposure.glsl b/drivers/gles2/shaders/exposure.glsl new file mode 100644 index 0000000000..001b90a0f1 --- /dev/null +++ b/drivers/gles2/shaders/exposure.glsl @@ -0,0 +1,98 @@ +[vertex] + + +layout(location=0) in highp vec4 vertex_attrib; + + +void main() { + + gl_Position = vertex_attrib; + +} + +[fragment] + + +uniform highp sampler2D source_exposure; //texunit:0 + +#ifdef EXPOSURE_BEGIN + +uniform highp ivec2 source_render_size; +uniform highp ivec2 target_size; + +#endif + +#ifdef EXPOSURE_END + +uniform highp sampler2D prev_exposure; //texunit:1 +uniform highp float exposure_adjust; +uniform highp float min_luminance; +uniform highp float max_luminance; + +#endif + +layout(location = 0) out highp float exposure; + + + +void main() { + + + +#ifdef EXPOSURE_BEGIN + + + ivec2 src_pos = ivec2(gl_FragCoord.xy)*source_render_size/target_size; + +#if 1 + //more precise and expensive, but less jittery + ivec2 next_pos = ivec2(gl_FragCoord.xy+ivec2(1))*source_render_size/target_size; + next_pos = max(next_pos,src_pos+ivec2(1)); //so it at least reads one pixel + highp vec3 source_color=vec3(0.0); + for(int i=src_pos.x;i<next_pos.x;i++) { + for(int j=src_pos.y;j<next_pos.y;j++) { + source_color += texelFetch(source_exposure,ivec2(i,j),0).rgb; + } + } + + source_color/=float( (next_pos.x-src_pos.x)*(next_pos.y-src_pos.y) ); +#else + highp vec3 source_color = texelFetch(source_exposure,src_pos,0).rgb; + +#endif + + exposure = max(source_color.r,max(source_color.g,source_color.b)); + +#else + + ivec2 coord = ivec2(gl_FragCoord.xy); + exposure = texelFetch(source_exposure,coord*3+ivec2(0,0),0).r; + exposure += texelFetch(source_exposure,coord*3+ivec2(1,0),0).r; + exposure += texelFetch(source_exposure,coord*3+ivec2(2,0),0).r; + exposure += texelFetch(source_exposure,coord*3+ivec2(0,1),0).r; + exposure += texelFetch(source_exposure,coord*3+ivec2(1,1),0).r; + exposure += texelFetch(source_exposure,coord*3+ivec2(2,1),0).r; + exposure += texelFetch(source_exposure,coord*3+ivec2(0,2),0).r; + exposure += texelFetch(source_exposure,coord*3+ivec2(1,2),0).r; + exposure += texelFetch(source_exposure,coord*3+ivec2(2,2),0).r; + exposure *= (1.0/9.0); + +#ifdef EXPOSURE_END + +#ifdef EXPOSURE_FORCE_SET + //will stay as is +#else + highp float prev_lum = texelFetch(prev_exposure,ivec2(0,0),0).r; //1 pixel previous exposure + exposure = clamp( prev_lum + (exposure-prev_lum)*exposure_adjust,min_luminance,max_luminance); + +#endif //EXPOSURE_FORCE_SET + + +#endif //EXPOSURE_END + +#endif //EXPOSURE_BEGIN + + +} + + diff --git a/drivers/gles2/shaders/particles.glsl b/drivers/gles2/shaders/particles.glsl new file mode 100644 index 0000000000..a62c124dfe --- /dev/null +++ b/drivers/gles2/shaders/particles.glsl @@ -0,0 +1,260 @@ +[vertex] + + + +layout(location=0) in highp vec4 color; +layout(location=1) in highp vec4 velocity_active; +layout(location=2) in highp vec4 custom; +layout(location=3) in highp vec4 xform_1; +layout(location=4) in highp vec4 xform_2; +layout(location=5) in highp vec4 xform_3; + + +struct Attractor { + + vec3 pos; + vec3 dir; + float radius; + float eat_radius; + float strength; + float attenuation; +}; + +#define MAX_ATTRACTORS 64 + +uniform bool emitting; +uniform float system_phase; +uniform float prev_system_phase; +uniform int total_particles; +uniform float explosiveness; +uniform float randomness; +uniform float time; +uniform float delta; + +uniform int attractor_count; +uniform Attractor attractors[MAX_ATTRACTORS]; +uniform bool clear; +uniform uint cycle; +uniform float lifetime; +uniform mat4 emission_transform; +uniform uint random_seed; + + +out highp vec4 out_color; //tfb: +out highp vec4 out_velocity_active; //tfb: +out highp vec4 out_custom; //tfb: +out highp vec4 out_xform_1; //tfb: +out highp vec4 out_xform_2; //tfb: +out highp vec4 out_xform_3; //tfb: + + +#if defined(USE_MATERIAL) + +layout(std140) uniform UniformData { //ubo:0 + +MATERIAL_UNIFORMS + +}; + +#endif + + +VERTEX_SHADER_GLOBALS + +uint hash(uint x) { + + x = ((x >> uint(16)) ^ x) * uint(0x45d9f3b); + x = ((x >> uint(16)) ^ x) * uint(0x45d9f3b); + x = (x >> uint(16)) ^ x; + return x; +} + + +void main() { + +#ifdef PARTICLES_COPY + + out_color=color; + out_velocity_active=velocity_active; + out_custom = custom; + out_xform_1 = xform_1; + out_xform_2 = xform_2; + out_xform_3 = xform_3; + +#else + + bool apply_forces=true; + bool apply_velocity=true; + float local_delta=delta; + + float mass = 1.0; + + float restart_phase = float(gl_VertexID)/float(total_particles); + + if (randomness>0.0) { + uint seed = cycle; + if (restart_phase >= system_phase) { + seed-=uint(1); + } + seed*=uint(total_particles); + seed+=uint(gl_VertexID); + float random = float(hash(seed) % uint(65536)) / 65536.0; + restart_phase+=randomness * random * 1.0 / float(total_particles); + } + + restart_phase*= (1.0-explosiveness); + bool restart=false; + bool shader_active = velocity_active.a > 0.5; + + if (system_phase > prev_system_phase) { + // restart_phase >= prev_system_phase is used so particles emit in the first frame they are processed + + if (restart_phase >= prev_system_phase && restart_phase < system_phase ) { + restart=true; +#ifdef USE_FRACTIONAL_DELTA + local_delta = (system_phase - restart_phase) * lifetime; +#endif + } + + } else { + if (restart_phase >= prev_system_phase) { + restart=true; +#ifdef USE_FRACTIONAL_DELTA + local_delta = (1.0 - restart_phase + system_phase) * lifetime; +#endif + } else if (restart_phase < system_phase ) { + restart=true; +#ifdef USE_FRACTIONAL_DELTA + local_delta = (system_phase - restart_phase) * lifetime; +#endif + } + } + + uint current_cycle = cycle; + + if (system_phase < restart_phase) { + current_cycle-=uint(1); + } + + uint particle_number = current_cycle * uint(total_particles) + uint(gl_VertexID); + int index = int(gl_VertexID); + + if (restart) { + shader_active=emitting; + } + + mat4 xform; + +#if defined(ENABLE_KEEP_DATA) + if (clear) { +#else + if (clear || restart) { +#endif + out_color=vec4(1.0); + out_velocity_active=vec4(0.0); + out_custom=vec4(0.0); + if (!restart) + shader_active=false; + + xform = mat4( + vec4(1.0,0.0,0.0,0.0), + vec4(0.0,1.0,0.0,0.0), + vec4(0.0,0.0,1.0,0.0), + vec4(0.0,0.0,0.0,1.0) + ); + } else { + out_color=color; + out_velocity_active=velocity_active; + out_custom=custom; + xform = transpose(mat4(xform_1,xform_2,xform_3,vec4(vec3(0.0),1.0))); + } + + if (shader_active) { + //execute shader + + { +VERTEX_SHADER_CODE + } + +#if !defined(DISABLE_FORCE) + + if (false) { + + vec3 force = vec3(0.0); + for(int i=0;i<attractor_count;i++) { + + vec3 rel_vec = xform[3].xyz - attractors[i].pos; + float dist = length(rel_vec); + if (attractors[i].radius < dist) + continue; + if (attractors[i].eat_radius>0.0 && attractors[i].eat_radius > dist) { + out_velocity_active.a=0.0; + } + + rel_vec = normalize(rel_vec); + + float attenuation = pow(dist / attractors[i].radius,attractors[i].attenuation); + + if (attractors[i].dir==vec3(0.0)) { + //towards center + force+=attractors[i].strength * rel_vec * attenuation * mass; + } else { + force+=attractors[i].strength * attractors[i].dir * attenuation *mass; + + } + } + + out_velocity_active.xyz += force * local_delta; + } +#endif + +#if !defined(DISABLE_VELOCITY) + + if (true) { + + xform[3].xyz += out_velocity_active.xyz * local_delta; + } +#endif + } else { + xform=mat4(0.0); + } + + xform = transpose(xform); + + out_velocity_active.a = mix(0.0,1.0,shader_active); + + out_xform_1 = xform[0]; + out_xform_2 = xform[1]; + out_xform_3 = xform[2]; + +#endif //PARTICLES_COPY + +} + +[fragment] + +//any code here is never executed, stuff is filled just so it works + + +#if defined(USE_MATERIAL) + +layout(std140) uniform UniformData { + +MATERIAL_UNIFORMS + +}; + +#endif + +FRAGMENT_SHADER_GLOBALS + +void main() { + + { +LIGHT_SHADER_CODE + } + + { +FRAGMENT_SHADER_CODE + } +} diff --git a/drivers/gles2/shaders/resolve.glsl b/drivers/gles2/shaders/resolve.glsl new file mode 100644 index 0000000000..0b50a9c57b --- /dev/null +++ b/drivers/gles2/shaders/resolve.glsl @@ -0,0 +1,44 @@ +[vertex] + + +layout(location=0) in highp vec4 vertex_attrib; +layout(location=4) in vec2 uv_in; + +out vec2 uv_interp; + + +void main() { + + uv_interp = uv_in; + gl_Position = vertex_attrib; +} + +[fragment] + +#if !defined(GLES_OVER_GL) +precision mediump float; +#endif + +in vec2 uv_interp; +uniform sampler2D source_specular; //texunit:0 +uniform sampler2D source_ssr; //texunit:1 + +uniform vec2 pixel_size; + +in vec2 uv2_interp; + +layout(location = 0) out vec4 frag_color; + +void main() { + + vec4 specular = texture( source_specular, uv_interp ); + +#ifdef USE_SSR + + vec4 ssr = textureLod(source_ssr,uv_interp,0.0); + specular.rgb = mix(specular.rgb,ssr.rgb*specular.a,ssr.a); +#endif + + frag_color = vec4(specular.rgb,1.0); +} + diff --git a/drivers/gles2/shaders/scene.glsl b/drivers/gles2/shaders/scene.glsl new file mode 100644 index 0000000000..79b989be4a --- /dev/null +++ b/drivers/gles2/shaders/scene.glsl @@ -0,0 +1,2113 @@ +[vertex] + +#define M_PI 3.14159265359 + +/* +from VisualServer: + +ARRAY_VERTEX=0, +ARRAY_NORMAL=1, +ARRAY_TANGENT=2, +ARRAY_COLOR=3, +ARRAY_TEX_UV=4, +ARRAY_TEX_UV2=5, +ARRAY_BONES=6, +ARRAY_WEIGHTS=7, +ARRAY_INDEX=8, +*/ + +//hack to use uv if no uv present so it works with lightmap + + +/* INPUT ATTRIBS */ + +layout(location=0) in highp vec4 vertex_attrib; +layout(location=1) in vec3 normal_attrib; +#if defined(ENABLE_TANGENT_INTERP) || defined(ENABLE_NORMALMAP) || defined(LIGHT_USE_ANISOTROPY) +layout(location=2) in vec4 tangent_attrib; +#endif + +#if defined(ENABLE_COLOR_INTERP) +layout(location=3) in vec4 color_attrib; +#endif + +#if defined(ENABLE_UV_INTERP) +layout(location=4) in vec2 uv_attrib; +#endif + +#if defined(ENABLE_UV2_INTERP) +layout(location=5) in vec2 uv2_attrib; +#endif + +uniform float normal_mult; + +#ifdef USE_SKELETON +layout(location=6) in ivec4 bone_indices; // attrib:6 +layout(location=7) in vec4 bone_weights; // attrib:7 +#endif + +#ifdef USE_INSTANCING + +layout(location=8) in highp vec4 instance_xform0; +layout(location=9) in highp vec4 instance_xform1; +layout(location=10) in highp vec4 instance_xform2; +layout(location=11) in lowp vec4 instance_color; + +#if defined(ENABLE_INSTANCE_CUSTOM) +layout(location=12) in highp vec4 instance_custom_data; +#endif + +#endif + +layout(std140) uniform SceneData { //ubo:0 + + highp mat4 projection_matrix; + highp mat4 inv_projection_matrix; + highp mat4 camera_inverse_matrix; + highp mat4 camera_matrix; + + mediump vec4 ambient_light_color; + mediump vec4 bg_color; + + mediump vec4 fog_color_enabled; + mediump vec4 fog_sun_color_amount; + + mediump float ambient_energy; + mediump float bg_energy; + + mediump float z_offset; + mediump float z_slope_scale; + highp float shadow_dual_paraboloid_render_zfar; + highp float shadow_dual_paraboloid_render_side; + + highp vec2 viewport_size; + highp vec2 screen_pixel_size; + highp vec2 shadow_atlas_pixel_size; + highp vec2 directional_shadow_pixel_size; + + highp float time; + highp float z_far; + mediump float reflection_multiplier; + mediump float subsurface_scatter_width; + mediump float ambient_occlusion_affect_light; + + bool fog_depth_enabled; + highp float fog_depth_begin; + highp float fog_depth_curve; + bool fog_transmit_enabled; + highp float fog_transmit_curve; + bool fog_height_enabled; + highp float fog_height_min; + highp float fog_height_max; + highp float fog_height_curve; + +}; + +uniform highp mat4 world_transform; + + +#ifdef USE_LIGHT_DIRECTIONAL + +layout(std140) uniform DirectionalLightData { //ubo:3 + + highp vec4 light_pos_inv_radius; + mediump vec4 light_direction_attenuation; + mediump vec4 light_color_energy; + mediump vec4 light_params; //cone attenuation, angle, specular, shadow enabled, + mediump vec4 light_clamp; + mediump vec4 shadow_color_contact; + highp mat4 shadow_matrix1; + highp mat4 shadow_matrix2; + highp mat4 shadow_matrix3; + highp mat4 shadow_matrix4; + mediump vec4 shadow_split_offsets; +}; + +#endif + +#ifdef USE_VERTEX_LIGHTING +//omni and spot + +struct LightData { + + highp vec4 light_pos_inv_radius; + mediump vec4 light_direction_attenuation; + mediump vec4 light_color_energy; + mediump vec4 light_params; //cone attenuation, angle, specular, shadow enabled, + mediump vec4 light_clamp; + mediump vec4 shadow_color_contact; + highp mat4 shadow_matrix; + +}; + + +layout(std140) uniform OmniLightData { //ubo:4 + + LightData omni_lights[MAX_LIGHT_DATA_STRUCTS]; +}; + +layout(std140) uniform SpotLightData { //ubo:5 + + LightData spot_lights[MAX_LIGHT_DATA_STRUCTS]; +}; + +#ifdef USE_FORWARD_LIGHTING + + +uniform int omni_light_indices[MAX_FORWARD_LIGHTS]; +uniform int omni_light_count; + +uniform int spot_light_indices[MAX_FORWARD_LIGHTS]; +uniform int spot_light_count; + +#endif + +out vec4 diffuse_light_interp; +out vec4 specular_light_interp; + +void light_compute(vec3 N, vec3 L,vec3 V, vec3 light_color, float roughness, inout vec3 diffuse, inout vec3 specular) { + + float dotNL = max(dot(N,L), 0.0 ); + diffuse += dotNL * light_color / M_PI; + + if (roughness > 0.0) { + + vec3 H = normalize(V + L); + float dotNH = max(dot(N,H), 0.0 ); + float intensity = pow( dotNH, (1.0-roughness) * 256.0); + specular += light_color * intensity; + + } +} + +void light_process_omni(int idx, vec3 vertex, vec3 eye_vec,vec3 normal, float roughness,inout vec3 diffuse, inout vec3 specular) { + + vec3 light_rel_vec = omni_lights[idx].light_pos_inv_radius.xyz-vertex; + float light_length = length( light_rel_vec ); + float normalized_distance = light_length*omni_lights[idx].light_pos_inv_radius.w; + vec3 light_attenuation = vec3(pow( max(1.0 - normalized_distance, 0.0), omni_lights[idx].light_direction_attenuation.w )); + + light_compute(normal,normalize(light_rel_vec),eye_vec,omni_lights[idx].light_color_energy.rgb * light_attenuation,roughness,diffuse,specular); + +} + +void light_process_spot(int idx, vec3 vertex, vec3 eye_vec, vec3 normal, float roughness, inout vec3 diffuse, inout vec3 specular) { + + vec3 light_rel_vec = spot_lights[idx].light_pos_inv_radius.xyz-vertex; + float light_length = length( light_rel_vec ); + float normalized_distance = light_length*spot_lights[idx].light_pos_inv_radius.w; + vec3 light_attenuation = vec3(pow( max(1.0 - normalized_distance, 0.001), spot_lights[idx].light_direction_attenuation.w )); + vec3 spot_dir = spot_lights[idx].light_direction_attenuation.xyz; + float spot_cutoff=spot_lights[idx].light_params.y; + float scos = max(dot(-normalize(light_rel_vec), spot_dir),spot_cutoff); + float spot_rim = (1.0 - scos) / (1.0 - spot_cutoff); + light_attenuation *= 1.0 - pow( max(spot_rim,0.001), spot_lights[idx].light_params.x); + + + light_compute(normal,normalize(light_rel_vec),eye_vec,spot_lights[idx].light_color_energy.rgb*light_attenuation,roughness,diffuse,specular); +} + + +#endif + +/* Varyings */ + +out highp vec3 vertex_interp; +out vec3 normal_interp; + +#if defined(ENABLE_COLOR_INTERP) +out vec4 color_interp; +#endif + +#if defined(ENABLE_UV_INTERP) +out vec2 uv_interp; +#endif + +#if defined(ENABLE_UV2_INTERP) +out vec2 uv2_interp; +#endif + + +#if defined(ENABLE_TANGENT_INTERP) || defined(ENABLE_NORMALMAP) || defined(LIGHT_USE_ANISOTROPY) +out vec3 tangent_interp; +out vec3 binormal_interp; +#endif + + + + + +#if defined(USE_MATERIAL) + +layout(std140) uniform UniformData { //ubo:1 + +MATERIAL_UNIFORMS + +}; + +#endif + +VERTEX_SHADER_GLOBALS + +#ifdef RENDER_DEPTH_DUAL_PARABOLOID + +out highp float dp_clip; + +#endif + +#define SKELETON_TEXTURE_WIDTH 256 + +#ifdef USE_SKELETON +uniform highp sampler2D skeleton_texture; //texunit:-1 +#endif + +out highp vec4 position_interp; + +// FIXME: This triggers a Mesa bug that breaks rendering, so disabled for now. +// See GH-13450 and https://bugs.freedesktop.org/show_bug.cgi?id=100316 +//invariant gl_Position; + +void main() { + + highp vec4 vertex = vertex_attrib; // vec4(vertex_attrib.xyz * data_attrib.x,1.0); + + mat4 world_matrix = world_transform; + + +#ifdef USE_INSTANCING + + { + highp mat4 m=mat4(instance_xform0,instance_xform1,instance_xform2,vec4(0.0,0.0,0.0,1.0)); + world_matrix = world_matrix * transpose(m); + } +#endif + + vec3 normal = normal_attrib * normal_mult; + + +#if defined(ENABLE_TANGENT_INTERP) || defined(ENABLE_NORMALMAP) || defined(LIGHT_USE_ANISOTROPY) + vec3 tangent = tangent_attrib.xyz; + tangent*=normal_mult; + float binormalf = tangent_attrib.a; +#endif + +#if defined(ENABLE_COLOR_INTERP) + color_interp = color_attrib; +#if defined(USE_INSTANCING) + color_interp *= instance_color; +#endif + +#endif + +#ifdef USE_SKELETON + { + //skeleton transform + ivec2 tex_ofs = ivec2( bone_indices.x%256, (bone_indices.x/256)*3 ); + highp mat3x4 m = mat3x4( + texelFetch(skeleton_texture,tex_ofs,0), + texelFetch(skeleton_texture,tex_ofs+ivec2(0,1),0), + texelFetch(skeleton_texture,tex_ofs+ivec2(0,2),0) + ) * bone_weights.x; + + tex_ofs = ivec2( bone_indices.y%256, (bone_indices.y/256)*3 ); + + m+= mat3x4( + texelFetch(skeleton_texture,tex_ofs,0), + texelFetch(skeleton_texture,tex_ofs+ivec2(0,1),0), + texelFetch(skeleton_texture,tex_ofs+ivec2(0,2),0) + ) * bone_weights.y; + + tex_ofs = ivec2( bone_indices.z%256, (bone_indices.z/256)*3 ); + + m+= mat3x4( + texelFetch(skeleton_texture,tex_ofs,0), + texelFetch(skeleton_texture,tex_ofs+ivec2(0,1),0), + texelFetch(skeleton_texture,tex_ofs+ivec2(0,2),0) + ) * bone_weights.z; + + + tex_ofs = ivec2( bone_indices.w%256, (bone_indices.w/256)*3 ); + + m+= mat3x4( + texelFetch(skeleton_texture,tex_ofs,0), + texelFetch(skeleton_texture,tex_ofs+ivec2(0,1),0), + texelFetch(skeleton_texture,tex_ofs+ivec2(0,2),0) + ) * bone_weights.w; + + + vertex.xyz = vertex * m; + + normal = vec4(normal,0.0) * m; +#if defined(ENABLE_TANGENT_INTERP) || defined(ENABLE_NORMALMAP) || defined(LIGHT_USE_ANISOTROPY) + tangent.xyz = vec4(tangent.xyz,0.0) * m; +#endif + } +#endif + + +#if defined(ENABLE_TANGENT_INTERP) || defined(ENABLE_NORMALMAP) || defined(LIGHT_USE_ANISOTROPY) + + vec3 binormal = normalize( cross(normal,tangent) * binormalf ); +#endif + +#if defined(ENABLE_UV_INTERP) + uv_interp = uv_attrib; +#endif + +#if defined(ENABLE_UV2_INTERP) + uv2_interp = uv2_attrib; +#endif + +#if defined(USE_INSTANCING) && defined(ENABLE_INSTANCE_CUSTOM) + vec4 instance_custom = instance_custom_data; +#else + vec4 instance_custom = vec4(0.0); +#endif + + highp mat4 modelview = camera_inverse_matrix * world_matrix; + highp mat4 local_projection = projection_matrix; + +//using world coordinates +#if !defined(SKIP_TRANSFORM_USED) && defined(VERTEX_WORLD_COORDS_USED) + + vertex = world_matrix * vertex; + normal = normalize((world_matrix * vec4(normal,0.0)).xyz); + +#if defined(ENABLE_TANGENT_INTERP) || defined(ENABLE_NORMALMAP) || defined(LIGHT_USE_ANISOTROPY) + + tangent = normalize((world_matrix * vec4(tangent,0.0)).xyz); + binormal = normalize((world_matrix * vec4(binormal,0.0)).xyz); +#endif +#endif + + float roughness=0.0; + +//defines that make writing custom shaders easier +#define projection_matrix local_projection +#define world_transform world_matrix +{ + +VERTEX_SHADER_CODE + +} + + + +//using local coordinates (default) +#if !defined(SKIP_TRANSFORM_USED) && !defined(VERTEX_WORLD_COORDS_USED) + + vertex = modelview * vertex; + normal = normalize((modelview * vec4(normal,0.0)).xyz); + +#if defined(ENABLE_TANGENT_INTERP) || defined(ENABLE_NORMALMAP) || defined(LIGHT_USE_ANISOTROPY) + + tangent = normalize((modelview * vec4(tangent,0.0)).xyz); + binormal = normalize((modelview * vec4(binormal,0.0)).xyz); +#endif +#endif + +//using world coordinates +#if !defined(SKIP_TRANSFORM_USED) && defined(VERTEX_WORLD_COORDS_USED) + + vertex = camera_inverse_matrix * vertex; + normal = normalize((camera_inverse_matrix * vec4(normal,0.0)).xyz); + +#if defined(ENABLE_TANGENT_INTERP) || defined(ENABLE_NORMALMAP) || defined(LIGHT_USE_ANISOTROPY) + + tangent = normalize((camera_inverse_matrix * vec4(tangent,0.0)).xyz); + binormal = normalize((camera_inverse_matrix * vec4(binormal,0.0)).xyz); +#endif +#endif + + vertex_interp = vertex.xyz; + normal_interp = normal; + + +#if defined(ENABLE_TANGENT_INTERP) || defined(ENABLE_NORMALMAP) || defined(LIGHT_USE_ANISOTROPY) + tangent_interp = tangent; + binormal_interp = binormal; +#endif + + +#ifdef RENDER_DEPTH + + +#ifdef RENDER_DEPTH_DUAL_PARABOLOID + + vertex_interp.z*= shadow_dual_paraboloid_render_side; + normal_interp.z*= shadow_dual_paraboloid_render_side; + + dp_clip=vertex_interp.z; //this attempts to avoid noise caused by objects sent to the other parabolloid side due to bias + + //for dual paraboloid shadow mapping, this is the fastest but least correct way, as it curves straight edges + + highp vec3 vtx = vertex_interp+normalize(vertex_interp)*z_offset; + highp float distance = length(vtx); + vtx = normalize(vtx); + vtx.xy/=1.0-vtx.z; + vtx.z=(distance/shadow_dual_paraboloid_render_zfar); + vtx.z=vtx.z * 2.0 - 1.0; + + vertex.xyz=vtx; + vertex.w=1.0; + + +#else + + float z_ofs = z_offset; + z_ofs += (1.0-abs(normal_interp.z))*z_slope_scale; + vertex_interp.z-=z_ofs; + +#endif //RENDER_DEPTH_DUAL_PARABOLOID + +#endif //RENDER_DEPTH + + gl_Position = projection_matrix * vec4(vertex_interp,1.0); + + position_interp=gl_Position; + +#ifdef USE_VERTEX_LIGHTING + + diffuse_light_interp=vec4(0.0); + specular_light_interp=vec4(0.0); + +#ifdef USE_FORWARD_LIGHTING + + for(int i=0;i<omni_light_count;i++) { + light_process_omni(omni_light_indices[i],vertex_interp,-normalize( vertex_interp ),normal_interp,roughness,diffuse_light_interp.rgb,specular_light_interp.rgb); + } + + for(int i=0;i<spot_light_count;i++) { + light_process_spot(spot_light_indices[i],vertex_interp,-normalize( vertex_interp ),normal_interp,roughness,diffuse_light_interp.rgb,specular_light_interp.rgb); + } +#endif + +#ifdef USE_LIGHT_DIRECTIONAL + + vec3 directional_diffuse = vec3(0.0); + vec3 directional_specular = vec3(0.0); + light_compute(normal_interp,-light_direction_attenuation.xyz,-normalize( vertex_interp ),light_color_energy.rgb,roughness,directional_diffuse,directional_specular); + + float diff_avg = dot(diffuse_light_interp.rgb,vec3(0.33333)); + float diff_dir_avg = dot(directional_diffuse,vec3(0.33333)); + if (diff_avg>0.0) { + diffuse_light_interp.a=diff_dir_avg/(diff_avg+diff_dir_avg); + } else { + diffuse_light_interp.a=1.0; + } + + diffuse_light_interp.rgb+=directional_diffuse; + + float spec_avg = dot(specular_light_interp.rgb,vec3(0.33333)); + float spec_dir_avg = dot(directional_specular,vec3(0.33333)); + if (spec_avg>0.0) { + specular_light_interp.a=spec_dir_avg/(spec_avg+spec_dir_avg); + } else { + specular_light_interp.a=1.0; + } + + specular_light_interp.rgb+=directional_specular; + +#endif //USE_LIGHT_DIRECTIONAL + + +#endif // USE_VERTEX_LIGHTING + +} + + +[fragment] + +/* texture unit usage, N is max_texture_unity-N + +1-skeleton +2-radiance +3-reflection_atlas +4-directional_shadow +5-shadow_atlas +6-decal_atlas +7-screen +8-depth +9-probe1 +10-probe2 + +*/ + +uniform highp mat4 world_transform; + +#define M_PI 3.14159265359 + +/* Varyings */ + +#if defined(ENABLE_COLOR_INTERP) +in vec4 color_interp; +#endif + +#if defined(ENABLE_UV_INTERP) +in vec2 uv_interp; +#endif + +#if defined(ENABLE_UV2_INTERP) +in vec2 uv2_interp; +#endif + +#if defined(ENABLE_TANGENT_INTERP) || defined(ENABLE_NORMALMAP) || defined(LIGHT_USE_ANISOTROPY) +in vec3 tangent_interp; +in vec3 binormal_interp; +#endif + +in highp vec3 vertex_interp; +in vec3 normal_interp; + + +/* PBR CHANNELS */ + +//used on forward mainly +uniform bool no_ambient_light; + + + +#ifdef USE_RADIANCE_MAP + + + +layout(std140) uniform Radiance { //ubo:2 + + mat4 radiance_inverse_xform; + float radiance_ambient_contribution; + +}; + +#define RADIANCE_MAX_LOD 5.0 + +#ifdef USE_RADIANCE_MAP_ARRAY + +uniform sampler2DArray radiance_map; //texunit:-2 + +vec3 textureDualParaboloid(sampler2DArray p_tex, vec3 p_vec,float p_roughness) { + + vec3 norm = normalize(p_vec); + norm.xy/=1.0+abs(norm.z); + norm.xy=norm.xy * vec2(0.5,0.25) + vec2(0.5,0.25); + + // we need to lie the derivatives (normg) and assume that DP side is always the same + // to get proper texture filtering + vec2 normg=norm.xy; + if (norm.z>0.0) { + norm.y=0.5-norm.y+0.5; + } + + // thanks to OpenGL spec using floor(layer + 0.5) for texture arrays, + // it's easy to have precision errors using fract() to interpolate layers + // as such, using fixed point to ensure it works. + + float index = p_roughness * RADIANCE_MAX_LOD; + int indexi = int(index * 256.0); + vec3 base = textureGrad(p_tex, vec3(norm.xy, float(indexi/256)),dFdx(normg),dFdy(normg)).xyz; + vec3 next = textureGrad(p_tex, vec3(norm.xy, float(indexi/256+1)),dFdx(normg),dFdy(normg)).xyz; + return mix(base,next,float(indexi%256)/256.0); +} + +#else + +uniform sampler2D radiance_map; //texunit:-2 + +vec3 textureDualParaboloid(sampler2D p_tex, vec3 p_vec,float p_roughness) { + + vec3 norm = normalize(p_vec); + norm.xy/=1.0+abs(norm.z); + norm.xy=norm.xy * vec2(0.5,0.25) + vec2(0.5,0.25); + if (norm.z>0.0) { + norm.y=0.5-norm.y+0.5; + } + return textureLod(p_tex, norm.xy, p_roughness * RADIANCE_MAX_LOD).xyz; +} + +#endif + +#endif + +/* Material Uniforms */ + + + +#if defined(USE_MATERIAL) + +layout(std140) uniform UniformData { + +MATERIAL_UNIFORMS + +}; + +#endif + +FRAGMENT_SHADER_GLOBALS + +layout(std140) uniform SceneData { + + highp mat4 projection_matrix; + highp mat4 inv_projection_matrix; + highp mat4 camera_inverse_matrix; + highp mat4 camera_matrix; + + mediump vec4 ambient_light_color; + mediump vec4 bg_color; + + mediump vec4 fog_color_enabled; + mediump vec4 fog_sun_color_amount; + + mediump float ambient_energy; + mediump float bg_energy; + + mediump float z_offset; + mediump float z_slope_scale; + highp float shadow_dual_paraboloid_render_zfar; + highp float shadow_dual_paraboloid_render_side; + + highp vec2 viewport_size; + highp vec2 screen_pixel_size; + highp vec2 shadow_atlas_pixel_size; + highp vec2 directional_shadow_pixel_size; + + highp float time; + highp float z_far; + mediump float reflection_multiplier; + mediump float subsurface_scatter_width; + mediump float ambient_occlusion_affect_light; + + bool fog_depth_enabled; + highp float fog_depth_begin; + highp float fog_depth_curve; + bool fog_transmit_enabled; + highp float fog_transmit_curve; + bool fog_height_enabled; + highp float fog_height_min; + highp float fog_height_max; + highp float fog_height_curve; +}; + +//directional light data + +#ifdef USE_LIGHT_DIRECTIONAL + +layout(std140) uniform DirectionalLightData { + + highp vec4 light_pos_inv_radius; + mediump vec4 light_direction_attenuation; + mediump vec4 light_color_energy; + mediump vec4 light_params; //cone attenuation, angle, specular, shadow enabled, + mediump vec4 light_clamp; + mediump vec4 shadow_color_contact; + highp mat4 shadow_matrix1; + highp mat4 shadow_matrix2; + highp mat4 shadow_matrix3; + highp mat4 shadow_matrix4; + mediump vec4 shadow_split_offsets; +}; + + +uniform highp sampler2DShadow directional_shadow; //texunit:-4 + +#endif + +#ifdef USE_VERTEX_LIGHTING +in vec4 diffuse_light_interp; +in vec4 specular_light_interp; +#endif +//omni and spot + +struct LightData { + + highp vec4 light_pos_inv_radius; + mediump vec4 light_direction_attenuation; + mediump vec4 light_color_energy; + mediump vec4 light_params; //cone attenuation, angle, specular, shadow enabled, + mediump vec4 light_clamp; + mediump vec4 shadow_color_contact; + highp mat4 shadow_matrix; + +}; + + +layout(std140) uniform OmniLightData { //ubo:4 + + LightData omni_lights[MAX_LIGHT_DATA_STRUCTS]; +}; + +layout(std140) uniform SpotLightData { //ubo:5 + + LightData spot_lights[MAX_LIGHT_DATA_STRUCTS]; +}; + + +uniform highp sampler2DShadow shadow_atlas; //texunit:-5 + + +struct ReflectionData { + + mediump vec4 box_extents; + mediump vec4 box_offset; + mediump vec4 params; // intensity, 0, interior , boxproject + mediump vec4 ambient; //ambient color, energy + mediump vec4 atlas_clamp; + highp mat4 local_matrix; //up to here for spot and omni, rest is for directional + //notes: for ambientblend, use distance to edge to blend between already existing global environment +}; + +layout(std140) uniform ReflectionProbeData { //ubo:6 + + ReflectionData reflections[MAX_REFLECTION_DATA_STRUCTS]; +}; +uniform mediump sampler2D reflection_atlas; //texunit:-3 + + +#ifdef USE_FORWARD_LIGHTING + +uniform int omni_light_indices[MAX_FORWARD_LIGHTS]; +uniform int omni_light_count; + +uniform int spot_light_indices[MAX_FORWARD_LIGHTS]; +uniform int spot_light_count; + +uniform int reflection_indices[MAX_FORWARD_LIGHTS]; +uniform int reflection_count; + +#endif + + +#if defined(SCREEN_TEXTURE_USED) + +uniform highp sampler2D screen_texture; //texunit:-7 + +#endif + +#ifdef USE_MULTIPLE_RENDER_TARGETS + +layout(location=0) out vec4 diffuse_buffer; +layout(location=1) out vec4 specular_buffer; +layout(location=2) out vec4 normal_mr_buffer; +#if defined(ENABLE_SSS) +layout(location=3) out float sss_buffer; +#endif + +#else + +layout(location=0) out vec4 frag_color; + +#endif + +in highp vec4 position_interp; +uniform highp sampler2D depth_buffer; //texunit:-8 + +#ifdef USE_CONTACT_SHADOWS + +float contact_shadow_compute(vec3 pos, vec3 dir, float max_distance) { + + if (abs(dir.z)>0.99) + return 1.0; + + vec3 endpoint = pos+dir*max_distance; + vec4 source = position_interp; + vec4 dest = projection_matrix * vec4(endpoint, 1.0); + + vec2 from_screen = (source.xy / source.w) * 0.5 + 0.5; + vec2 to_screen = (dest.xy / dest.w) * 0.5 + 0.5; + + vec2 screen_rel = to_screen - from_screen; + + if (length(screen_rel)<0.00001) + return 1.0; //too small, don't do anything + + /*float pixel_size; //approximate pixel size + + if (screen_rel.x > screen_rel.y) { + + pixel_size = abs((pos.x-endpoint.x)/(screen_rel.x/screen_pixel_size.x)); + } else { + pixel_size = abs((pos.y-endpoint.y)/(screen_rel.y/screen_pixel_size.y)); + + }*/ + vec4 bias = projection_matrix * vec4(pos+vec3(0.0,0.0,0.04), 1.0); //todo un-harcode the 0.04 + + + + vec2 pixel_incr = normalize(screen_rel)*screen_pixel_size; + + + float steps = length(screen_rel) / length(pixel_incr); + steps = min(2000.0,steps); //put a limit to avoid freezing in some strange situation + //steps=10.0; + + vec4 incr = (dest - source)/steps; + float ratio=0.0; + float ratio_incr = 1.0/steps; + + while(steps>0.0) { + source += incr*2.0; + bias+=incr*2.0; + + vec3 uv_depth = (source.xyz / source.w) * 0.5 + 0.5; + float depth = texture(depth_buffer,uv_depth.xy).r; + + if (depth < uv_depth.z) { + if (depth > (bias.z/bias.w) * 0.5 + 0.5) { + return min(pow(ratio,4.0),1.0); + } else { + return 1.0; + } + } + + + ratio+=ratio_incr; + steps-=1.0; + } + + return 1.0; +} + +#endif + + +// This returns the G_GGX function divided by 2 cos_theta_m, where in practice cos_theta_m is either N.L or N.V. +// We're dividing this factor off because the overall term we'll end up looks like +// (see, for example, the first unnumbered equation in B. Burley, "Physically Based Shading at Disney", SIGGRAPH 2012): +// +// F(L.V) D(N.H) G(N.L) G(N.V) / (4 N.L N.V) +// +// We're basically regouping this as +// +// F(L.V) D(N.H) [G(N.L)/(2 N.L)] [G(N.V) / (2 N.V)] +// +// and thus, this function implements the [G(N.m)/(2 N.m)] part with m = L or V. +// +// The contents of the D and G (G1) functions (GGX) are taken from +// E. Heitz, "Understanding the Masking-Shadowing Function in Microfacet-Based BRDFs", J. Comp. Graph. Tech. 3 (2) (2014). +// Eqns 71-72 and 85-86 (see also Eqns 43 and 80). + +float G_GGX_2cos(float cos_theta_m, float alpha) { + // Schlick's approximation + // C. Schlick, "An Inexpensive BRDF Model for Physically-based Rendering", Computer Graphics Forum. 13 (3): 233 (1994) + // Eq. (19), although see Heitz (2014) the about the problems with his derivation. + // It nevertheless approximates GGX well with k = alpha/2. + float k = 0.5*alpha; + return 0.5 / (cos_theta_m * (1.0 - k) + k); + + // float cos2 = cos_theta_m*cos_theta_m; + // float sin2 = (1.0-cos2); + // return 1.0 /( cos_theta_m + sqrt(cos2 + alpha*alpha*sin2) ); +} + +float D_GGX(float cos_theta_m, float alpha) { + float alpha2 = alpha*alpha; + float d = 1.0 + (alpha2-1.0)*cos_theta_m*cos_theta_m; + return alpha2/(M_PI * d * d); +} + +float G_GGX_anisotropic_2cos(float cos_theta_m, float alpha_x, float alpha_y, float cos_phi, float sin_phi) { + float cos2 = cos_theta_m * cos_theta_m; + float sin2 = (1.0-cos2); + float s_x = alpha_x * cos_phi; + float s_y = alpha_y * sin_phi; + return 1.0 / (cos_theta_m + sqrt(cos2 + (s_x*s_x + s_y*s_y)*sin2 )); +} + +float D_GGX_anisotropic(float cos_theta_m, float alpha_x, float alpha_y, float cos_phi, float sin_phi) { + float cos2 = cos_theta_m * cos_theta_m; + float sin2 = (1.0-cos2); + float r_x = cos_phi/alpha_x; + float r_y = sin_phi/alpha_y; + float d = cos2 + sin2*(r_x * r_x + r_y * r_y); + return 1.0 / (M_PI * alpha_x * alpha_y * d * d ); +} + + +float SchlickFresnel(float u) +{ + float m = 1.0-u; + float m2 = m*m; + return m2*m2*m; // pow(m,5) +} + +float GTR1(float NdotH, float a) +{ + if (a >= 1.0) return 1.0/M_PI; + float a2 = a*a; + float t = 1.0 + (a2-1.0)*NdotH*NdotH; + return (a2-1.0) / (M_PI*log(a2)*t); +} + +vec3 metallic_to_specular_color(float metallic, float specular, vec3 albedo) { + float dielectric = (0.034 * 2.0) * specular; + // energy conservation + return mix(vec3(dielectric), albedo, metallic); // TODO: reference? +} + +void light_compute(vec3 N, vec3 L, vec3 V, vec3 B, vec3 T, vec3 light_color, vec3 attenuation, vec3 diffuse_color, vec3 transmission, float specular_blob_intensity, float roughness, float metallic, float rim, float rim_tint, float clearcoat, float clearcoat_gloss, float anisotropy, inout vec3 diffuse_light, inout vec3 specular_light) { + +#if defined(USE_LIGHT_SHADER_CODE) +//light is written by the light shader + + vec3 normal = N; + vec3 albedo = diffuse_color; + vec3 light = L; + vec3 view = V; + +LIGHT_SHADER_CODE + + +#else + float NdotL = dot(N,L); + float cNdotL = max(NdotL, 0.0); // clamped NdotL + float NdotV = dot(N, V); + float cNdotV = max(NdotV, 0.0); + + if (metallic < 1.0) { +#if defined(DIFFUSE_OREN_NAYAR) + vec3 diffuse_brdf_NL; +#else + float diffuse_brdf_NL; // BRDF times N.L for calculating diffuse radiance +#endif + + +#if defined(DIFFUSE_LAMBERT_WRAP) + //energy conserving lambert wrap shader + diffuse_brdf_NL = max(0.0,(NdotL + roughness) / ((1.0 + roughness) * (1.0 + roughness))); + +#elif defined(DIFFUSE_OREN_NAYAR) + + { + // see http://mimosa-pudica.net/improved-oren-nayar.html + float LdotV = dot(L, V); + + + float s = LdotV - NdotL * NdotV; + float t = mix(1.0, max(NdotL, NdotV), step(0.0, s)); + + float sigma2 = roughness * roughness; // TODO: this needs checking + vec3 A = 1.0 + sigma2 * (- 0.5 / (sigma2 + 0.33) + 0.17*diffuse_color / (sigma2 + 0.13) ); + float B = 0.45 * sigma2 / (sigma2 + 0.09); + + diffuse_brdf_NL = cNdotL * (A + vec3(B) * s / t) * (1.0 / M_PI); + } + +#elif defined(DIFFUSE_TOON) + + diffuse_brdf_NL = smoothstep(-roughness,max(roughness,0.01),NdotL); + +#elif defined(DIFFUSE_BURLEY) + + { + + + vec3 H = normalize(V + L); + float cLdotH = max(0.0,dot(L, H)); + + float FD90 = 0.5 + 2.0 * cLdotH * cLdotH * roughness; + float FdV = 1.0 + (FD90 - 1.0) * SchlickFresnel(cNdotV); + float FdL = 1.0 + (FD90 - 1.0) * SchlickFresnel(cNdotL); + diffuse_brdf_NL = (1.0 / M_PI) * FdV * FdL * cNdotL; + /* + float energyBias = mix(roughness, 0.0, 0.5); + float energyFactor = mix(roughness, 1.0, 1.0 / 1.51); + float fd90 = energyBias + 2.0 * VoH * VoH * roughness; + float f0 = 1.0; + float lightScatter = f0 + (fd90 - f0) * pow(1.0 - cNdotL, 5.0); + float viewScatter = f0 + (fd90 - f0) * pow(1.0 - cNdotV, 5.0); + + diffuse_brdf_NL = lightScatter * viewScatter * energyFactor;*/ + } +#else + //lambert + diffuse_brdf_NL = cNdotL * (1.0 / M_PI); +#endif + +#if defined(TRANSMISSION_USED) + diffuse_light += light_color * diffuse_color * mix(vec3(diffuse_brdf_NL), vec3(M_PI), transmission) * attenuation; +#else + diffuse_light += light_color * diffuse_color * diffuse_brdf_NL * attenuation; +#endif + + + +#if defined(LIGHT_USE_RIM) + float rim_light = pow(1.0-cNdotV, (1.0-roughness)*16.0); + diffuse_light += rim_light * rim * mix(vec3(1.0),diffuse_color,rim_tint) * light_color; +#endif + } + + + if (roughness > 0.0) { // FIXME: roughness == 0 should not disable specular light entirely + + + // D + +#if defined(SPECULAR_BLINN) + + vec3 H = normalize(V + L); + float cNdotH = max(dot(N,H), 0.0 ); + float intensity = pow( cNdotH, (1.0-roughness) * 256.0); + specular_light += light_color * intensity * specular_blob_intensity * attenuation; + +#elif defined(SPECULAR_PHONG) + + vec3 R = normalize(-reflect(L,N)); + float cRdotV = max(0.0,dot(R,V)); + float intensity = pow( cRdotV, (1.0-roughness) * 256.0); + specular_light += light_color * intensity * specular_blob_intensity * attenuation; + +#elif defined(SPECULAR_TOON) + + vec3 R = normalize(-reflect(L,N)); + float RdotV = dot(R,V); + float mid = 1.0-roughness; + mid*=mid; + float intensity = smoothstep(mid-roughness*0.5, mid+roughness*0.5, RdotV) * mid; + diffuse_light += light_color * intensity * specular_blob_intensity * attenuation; // write to diffuse_light, as in toon shading you generally want no reflection + +#elif defined(SPECULAR_DISABLED) + //none.. + +#elif defined(SPECULAR_SCHLICK_GGX) + // shlick+ggx as default + + vec3 H = normalize(V + L); + + float cNdotH = max(dot(N,H), 0.0); + float cLdotH = max(dot(L,H), 0.0); + +# if defined(LIGHT_USE_ANISOTROPY) + + float aspect = sqrt(1.0-anisotropy*0.9); + float rx = roughness/aspect; + float ry = roughness*aspect; + float ax = rx*rx; + float ay = ry*ry; + float XdotH = dot( T, H ); + float YdotH = dot( B, H ); + float D = D_GGX_anisotropic(cNdotH, ax, ay, XdotH, YdotH); + float G = G_GGX_anisotropic_2cos(cNdotL, ax, ay, XdotH, YdotH) * G_GGX_anisotropic_2cos(cNdotV, ax, ay, XdotH, YdotH); + +# else + float alpha = roughness * roughness; + float D = D_GGX(cNdotH, alpha); + float G = G_GGX_2cos(cNdotL, alpha) * G_GGX_2cos(cNdotV, alpha); +# endif + // F + float F0 = 1.0; // FIXME + float cLdotH5 = SchlickFresnel(cLdotH); + float F = mix(cLdotH5, 1.0, F0); + + float specular_brdf_NL = cNdotL * D * F * G; + + specular_light += specular_brdf_NL * light_color * specular_blob_intensity * attenuation; +#endif + +#if defined(LIGHT_USE_CLEARCOAT) + if (clearcoat_gloss > 0.0) { +# if !defined(SPECULAR_SCHLICK_GGX) && !defined(SPECULAR_BLINN) + vec3 H = normalize(V + L); +# endif +# if !defined(SPECULAR_SCHLICK_GGX) + float cNdotH = max(dot(N,H), 0.0); + float cLdotH = max(dot(L,H), 0.0); + float cLdotH5 = SchlickFresnel(cLdotH); +#endif + float Dr = GTR1(cNdotH, mix(.1, .001, clearcoat_gloss)); + float Fr = mix(.04, 1.0, cLdotH5); + float Gr = G_GGX_2cos(cNdotL, .25) * G_GGX_2cos(cNdotV, .25); + + + float specular_brdf_NL = 0.25 * clearcoat * Gr * Fr * Dr * cNdotL; + + specular_light += specular_brdf_NL * light_color * specular_blob_intensity * attenuation; + } +#endif + } + + +#endif //defined(USE_LIGHT_SHADER_CODE) +} + + +float sample_shadow(highp sampler2DShadow shadow, vec2 shadow_pixel_size, vec2 pos, float depth, vec4 clamp_rect) { + +#ifdef SHADOW_MODE_PCF_13 + + float avg=textureProj(shadow,vec4(pos,depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(shadow_pixel_size.x,0.0),depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(-shadow_pixel_size.x,0.0),depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(0.0,shadow_pixel_size.y),depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(0.0,-shadow_pixel_size.y),depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(shadow_pixel_size.x,shadow_pixel_size.y),depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(-shadow_pixel_size.x,shadow_pixel_size.y),depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(shadow_pixel_size.x,-shadow_pixel_size.y),depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(-shadow_pixel_size.x,-shadow_pixel_size.y),depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(shadow_pixel_size.x*2.0,0.0),depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(-shadow_pixel_size.x*2.0,0.0),depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(0.0,shadow_pixel_size.y*2.0),depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(0.0,-shadow_pixel_size.y*2.0),depth,1.0)); + return avg*(1.0/13.0); + +#elif defined(SHADOW_MODE_PCF_5) + + float avg=textureProj(shadow,vec4(pos,depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(shadow_pixel_size.x,0.0),depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(-shadow_pixel_size.x,0.0),depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(0.0,shadow_pixel_size.y),depth,1.0)); + avg+=textureProj(shadow,vec4(pos+vec2(0.0,-shadow_pixel_size.y),depth,1.0)); + return avg*(1.0/5.0); + +#else + + return textureProj(shadow,vec4(pos,depth,1.0)); + +#endif + +} + +#ifdef RENDER_DEPTH_DUAL_PARABOLOID + +in highp float dp_clip; + +#endif + + + +#if 0 +//need to save texture depth for this + +vec3 light_transmittance(float translucency,vec3 light_vec, vec3 normal, vec3 pos, float distance) { + + float scale = 8.25 * (1.0 - translucency) / subsurface_scatter_width; + float d = scale * distance; + + /** + * Armed with the thickness, we can now calculate the color by means of the + * precalculated transmittance profile. + * (It can be precomputed into a texture, for maximum performance): + */ + float dd = -d * d; + vec3 profile = vec3(0.233, 0.455, 0.649) * exp(dd / 0.0064) + + vec3(0.1, 0.336, 0.344) * exp(dd / 0.0484) + + vec3(0.118, 0.198, 0.0) * exp(dd / 0.187) + + vec3(0.113, 0.007, 0.007) * exp(dd / 0.567) + + vec3(0.358, 0.004, 0.0) * exp(dd / 1.99) + + vec3(0.078, 0.0, 0.0) * exp(dd / 7.41); + + /** + * Using the profile, we finally approximate the transmitted lighting from + * the back of the object: + */ + return profile * clamp(0.3 + dot(light_vec, normal),0.0,1.0); +} +#endif + +void light_process_omni(int idx, vec3 vertex, vec3 eye_vec,vec3 normal,vec3 binormal, vec3 tangent, vec3 albedo, vec3 transmission, float roughness, float metallic, float rim, float rim_tint, float clearcoat, float clearcoat_gloss, float anisotropy, float p_blob_intensity, inout vec3 diffuse_light, inout vec3 specular_light) { + + vec3 light_rel_vec = omni_lights[idx].light_pos_inv_radius.xyz-vertex; + float light_length = length( light_rel_vec ); + float normalized_distance = light_length*omni_lights[idx].light_pos_inv_radius.w; + vec3 light_attenuation = vec3(pow( max(1.0 - normalized_distance, 0.0), omni_lights[idx].light_direction_attenuation.w )); + + if (omni_lights[idx].light_params.w>0.5) { + //there is a shadowmap + + highp vec3 splane=(omni_lights[idx].shadow_matrix * vec4(vertex,1.0)).xyz; + float shadow_len=length(splane); + splane=normalize(splane); + vec4 clamp_rect=omni_lights[idx].light_clamp; + + if (splane.z>=0.0) { + + splane.z+=1.0; + + clamp_rect.y+=clamp_rect.w; + + } else { + + splane.z=1.0 - splane.z; + + /* + if (clamp_rect.z<clamp_rect.w) { + clamp_rect.x+=clamp_rect.z; + } else { + clamp_rect.y+=clamp_rect.w; + } + */ + + } + + splane.xy/=splane.z; + splane.xy=splane.xy * 0.5 + 0.5; + splane.z = shadow_len * omni_lights[idx].light_pos_inv_radius.w; + + splane.xy = clamp_rect.xy+splane.xy*clamp_rect.zw; + float shadow = sample_shadow(shadow_atlas,shadow_atlas_pixel_size,splane.xy,splane.z,clamp_rect); + +#ifdef USE_CONTACT_SHADOWS + + if (shadow>0.01 && omni_lights[idx].shadow_color_contact.a>0.0) { + + float contact_shadow = contact_shadow_compute(vertex,normalize(light_rel_vec),min(light_length,omni_lights[idx].shadow_color_contact.a)); + shadow=min(shadow,contact_shadow); + + } +#endif + light_attenuation*=mix(omni_lights[idx].shadow_color_contact.rgb,vec3(1.0),shadow); + } + + light_compute(normal,normalize(light_rel_vec),eye_vec,binormal,tangent,omni_lights[idx].light_color_energy.rgb,light_attenuation,albedo,transmission,omni_lights[idx].light_params.z*p_blob_intensity,roughness,metallic,rim,rim_tint,clearcoat,clearcoat_gloss,anisotropy,diffuse_light,specular_light); + +} + +void light_process_spot(int idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 binormal, vec3 tangent,vec3 albedo, vec3 transmission,float roughness, float metallic, float rim, float rim_tint, float clearcoat, float clearcoat_gloss,float anisotropy,float p_blob_intensity, inout vec3 diffuse_light, inout vec3 specular_light) { + + vec3 light_rel_vec = spot_lights[idx].light_pos_inv_radius.xyz-vertex; + float light_length = length( light_rel_vec ); + float normalized_distance = light_length*spot_lights[idx].light_pos_inv_radius.w; + vec3 light_attenuation = vec3(pow( max(1.0 - normalized_distance, 0.001), spot_lights[idx].light_direction_attenuation.w )); + vec3 spot_dir = spot_lights[idx].light_direction_attenuation.xyz; + float spot_cutoff=spot_lights[idx].light_params.y; + float scos = max(dot(-normalize(light_rel_vec), spot_dir),spot_cutoff); + float spot_rim = (1.0 - scos) / (1.0 - spot_cutoff); + light_attenuation *= 1.0 - pow( max(spot_rim,0.001), spot_lights[idx].light_params.x); + + if (spot_lights[idx].light_params.w>0.5) { + //there is a shadowmap + highp vec4 splane=(spot_lights[idx].shadow_matrix * vec4(vertex,1.0)); + splane.xyz/=splane.w; + + float shadow = sample_shadow(shadow_atlas,shadow_atlas_pixel_size,splane.xy,splane.z,spot_lights[idx].light_clamp); + +#ifdef USE_CONTACT_SHADOWS + if (shadow>0.01 && spot_lights[idx].shadow_color_contact.a>0.0) { + + float contact_shadow = contact_shadow_compute(vertex,normalize(light_rel_vec),min(light_length,spot_lights[idx].shadow_color_contact.a)); + shadow=min(shadow,contact_shadow); + + } +#endif + light_attenuation*=mix(spot_lights[idx].shadow_color_contact.rgb,vec3(1.0),shadow); + } + + light_compute(normal,normalize(light_rel_vec),eye_vec,binormal,tangent,spot_lights[idx].light_color_energy.rgb,light_attenuation,albedo,transmission,spot_lights[idx].light_params.z*p_blob_intensity,roughness,metallic,rim,rim_tint,clearcoat,clearcoat_gloss,anisotropy,diffuse_light,specular_light); + +} + +void reflection_process(int idx, vec3 vertex, vec3 normal,vec3 binormal, vec3 tangent,float roughness,float anisotropy,vec3 ambient,vec3 skybox, inout highp vec4 reflection_accum,inout highp vec4 ambient_accum) { + + vec3 ref_vec = normalize(reflect(vertex,normal)); + vec3 local_pos = (reflections[idx].local_matrix * vec4(vertex,1.0)).xyz; + vec3 box_extents = reflections[idx].box_extents.xyz; + + if (any(greaterThan(abs(local_pos),box_extents))) { //out of the reflection box + return; + } + + vec3 inner_pos = abs(local_pos / box_extents); + float blend = max(inner_pos.x,max(inner_pos.y,inner_pos.z)); + //make blend more rounded + blend=mix(length(inner_pos),blend,blend); + blend*=blend; + blend=1.001-blend; + + if (reflections[idx].params.x>0.0){// compute reflection + + vec3 local_ref_vec = (reflections[idx].local_matrix * vec4(ref_vec,0.0)).xyz; + + if (reflections[idx].params.w > 0.5) { //box project + + vec3 nrdir = normalize(local_ref_vec); + vec3 rbmax = (box_extents - local_pos)/nrdir; + vec3 rbmin = (-box_extents - local_pos)/nrdir; + + + vec3 rbminmax = mix(rbmin,rbmax,greaterThan(nrdir,vec3(0.0,0.0,0.0))); + + float fa = min(min(rbminmax.x, rbminmax.y), rbminmax.z); + vec3 posonbox = local_pos + nrdir * fa; + local_ref_vec = posonbox - reflections[idx].box_offset.xyz; + } + + + vec4 clamp_rect=reflections[idx].atlas_clamp; + vec3 norm = normalize(local_ref_vec); + norm.xy/=1.0+abs(norm.z); + norm.xy=norm.xy * vec2(0.5,0.25) + vec2(0.5,0.25); + if (norm.z>0.0) { + norm.y=0.5-norm.y+0.5; + } + + vec2 atlas_uv = norm.xy * clamp_rect.zw + clamp_rect.xy; + atlas_uv = clamp(atlas_uv,clamp_rect.xy,clamp_rect.xy+clamp_rect.zw); + + highp vec4 reflection; + reflection.rgb = textureLod(reflection_atlas,atlas_uv,roughness*5.0).rgb; + + if (reflections[idx].params.z < 0.5) { + reflection.rgb = mix(skybox,reflection.rgb,blend); + } + reflection.rgb*=reflections[idx].params.x; + reflection.a = blend; + reflection.rgb*=reflection.a; + + reflection_accum+=reflection; + } + + if (reflections[idx].ambient.a>0.0) { //compute ambient using skybox + + + vec3 local_amb_vec = (reflections[idx].local_matrix * vec4(normal,0.0)).xyz; + + vec3 splane=normalize(local_amb_vec); + vec4 clamp_rect=reflections[idx].atlas_clamp; + + splane.z*=-1.0; + if (splane.z>=0.0) { + splane.z+=1.0; + clamp_rect.y+=clamp_rect.w; + } else { + splane.z=1.0 - splane.z; + splane.y=-splane.y; + } + + splane.xy/=splane.z; + splane.xy=splane.xy * 0.5 + 0.5; + + splane.xy = splane.xy * clamp_rect.zw + clamp_rect.xy; + splane.xy = clamp(splane.xy,clamp_rect.xy,clamp_rect.xy+clamp_rect.zw); + + highp vec4 ambient_out; + ambient_out.a=blend; + ambient_out.rgb = textureLod(reflection_atlas,splane.xy,5.0).rgb; + ambient_out.rgb=mix(reflections[idx].ambient.rgb,ambient_out.rgb,reflections[idx].ambient.a); + if (reflections[idx].params.z < 0.5) { + ambient_out.rgb = mix(ambient,ambient_out.rgb,blend); + } + + ambient_out.rgb *= ambient_out.a; + ambient_accum+=ambient_out; + } else { + + highp vec4 ambient_out; + ambient_out.a=blend; + ambient_out.rgb=reflections[idx].ambient.rgb; + if (reflections[idx].params.z < 0.5) { + ambient_out.rgb = mix(ambient,ambient_out.rgb,blend); + } + ambient_out.rgb *= ambient_out.a; + ambient_accum+=ambient_out; + + } +} + +#ifdef USE_GI_PROBES + +uniform mediump sampler3D gi_probe1; //texunit:-9 +uniform highp mat4 gi_probe_xform1; +uniform highp vec3 gi_probe_bounds1; +uniform highp vec3 gi_probe_cell_size1; +uniform highp float gi_probe_multiplier1; +uniform highp float gi_probe_bias1; +uniform highp float gi_probe_normal_bias1; +uniform bool gi_probe_blend_ambient1; + +uniform mediump sampler3D gi_probe2; //texunit:-10 +uniform highp mat4 gi_probe_xform2; +uniform highp vec3 gi_probe_bounds2; +uniform highp vec3 gi_probe_cell_size2; +uniform highp float gi_probe_multiplier2; +uniform highp float gi_probe_bias2; +uniform highp float gi_probe_normal_bias2; +uniform bool gi_probe2_enabled; +uniform bool gi_probe_blend_ambient2; + +vec3 voxel_cone_trace(mediump sampler3D probe, vec3 cell_size, vec3 pos, vec3 ambient, bool blend_ambient, vec3 direction, float tan_half_angle, float max_distance, float p_bias) { + + float dist = p_bias;//1.0; //dot(direction,mix(vec3(-1.0),vec3(1.0),greaterThan(direction,vec3(0.0))))*2.0; + float alpha=0.0; + vec3 color = vec3(0.0); + + while(dist < max_distance && alpha < 0.95) { + float diameter = max(1.0, 2.0 * tan_half_angle * dist); + vec4 scolor = textureLod(probe, (pos + dist * direction) * cell_size, log2(diameter) ); + float a = (1.0 - alpha); + color += scolor.rgb * a; + alpha += a * scolor.a; + dist += diameter * 0.5; + } + + if (blend_ambient) { + color.rgb = mix(ambient,color.rgb,min(1.0,alpha/0.95)); + } + + return color; +} + +void gi_probe_compute(mediump sampler3D probe, mat4 probe_xform, vec3 bounds,vec3 cell_size,vec3 pos, vec3 ambient, vec3 environment, bool blend_ambient,float multiplier, mat3 normal_mtx,vec3 ref_vec, float roughness,float p_bias,float p_normal_bias, inout vec4 out_spec, inout vec4 out_diff) { + + + + vec3 probe_pos = (probe_xform * vec4(pos,1.0)).xyz; + vec3 ref_pos = (probe_xform * vec4(pos+ref_vec,1.0)).xyz; + ref_vec = normalize(ref_pos - probe_pos); + + probe_pos+=(probe_xform * vec4(normal_mtx[2],0.0)).xyz*p_normal_bias; + +/* out_diff.rgb = voxel_cone_trace(probe,cell_size,probe_pos,normalize((probe_xform * vec4(ref_vec,0.0)).xyz),0.0 ,100.0); + out_diff.a = 1.0; + return;*/ + //out_diff = vec4(textureLod(probe,probe_pos*cell_size,3.0).rgb,1.0); + //return; + + //this causes corrupted pixels, i have no idea why.. + if (any(bvec2(any(lessThan(probe_pos,vec3(0.0))),any(greaterThan(probe_pos,bounds))))) { + return; + } + + //vec3 blendv = probe_pos/bounds * 2.0 - 1.0; + //float blend = 1.001-max(blendv.x,max(blendv.y,blendv.z)); + float blend=1.0; + + float max_distance = length(bounds); + + //radiance +#ifdef VCT_QUALITY_HIGH + +#define MAX_CONE_DIRS 6 + vec3 cone_dirs[MAX_CONE_DIRS] = vec3[] ( + vec3(0, 0, 1), + vec3(0.866025, 0, 0.5), + vec3(0.267617, 0.823639, 0.5), + vec3(-0.700629, 0.509037, 0.5), + vec3(-0.700629, -0.509037, 0.5), + vec3(0.267617, -0.823639, 0.5) + ); + + float cone_weights[MAX_CONE_DIRS] = float[](0.25, 0.15, 0.15, 0.15, 0.15, 0.15); + float cone_angle_tan = 0.577; + float min_ref_tan = 0.0; +#else + +#define MAX_CONE_DIRS 4 + + vec3 cone_dirs[MAX_CONE_DIRS] = vec3[] ( + vec3(0.707107, 0, 0.707107), + vec3(0, 0.707107, 0.707107), + vec3(-0.707107, 0, 0.707107), + vec3(0, -0.707107, 0.707107) + ); + + float cone_weights[MAX_CONE_DIRS] = float[](0.25, 0.25, 0.25, 0.25); + float cone_angle_tan = 0.98269; + max_distance*=0.5; + float min_ref_tan = 0.2; + +#endif + vec3 light=vec3(0.0); + for(int i=0;i<MAX_CONE_DIRS;i++) { + + vec3 dir = normalize( (probe_xform * vec4(pos + normal_mtx * cone_dirs[i],1.0)).xyz - probe_pos); + light+=cone_weights[i] * voxel_cone_trace(probe,cell_size,probe_pos,ambient,blend_ambient,dir,cone_angle_tan,max_distance,p_bias); + + } + + light*=multiplier; + + out_diff += vec4(light*blend,blend); + + //irradiance + + vec3 irr_light = voxel_cone_trace(probe,cell_size,probe_pos,environment,blend_ambient,ref_vec,max(min_ref_tan,tan(roughness * 0.5 * M_PI)) ,max_distance,p_bias); + + irr_light *= multiplier; + //irr_light=vec3(0.0); + + out_spec += vec4(irr_light*blend,blend); + +} + + +void gi_probes_compute(vec3 pos, vec3 normal, float roughness, inout vec3 out_specular, inout vec3 out_ambient) { + + roughness = roughness * roughness; + + vec3 ref_vec = normalize(reflect(normalize(pos),normal)); + + //find arbitrary tangent and bitangent, then build a matrix + vec3 v0 = abs(normal.z) < 0.999 ? vec3(0, 0, 1) : vec3(0, 1, 0); + vec3 tangent = normalize(cross(v0, normal)); + vec3 bitangent = normalize(cross(tangent, normal)); + mat3 normal_mat = mat3(tangent,bitangent,normal); + + vec4 diff_accum = vec4(0.0); + vec4 spec_accum = vec4(0.0); + + vec3 ambient = out_ambient; + out_ambient = vec3(0.0); + + vec3 environment = out_specular; + + out_specular = vec3(0.0); + + gi_probe_compute(gi_probe1,gi_probe_xform1,gi_probe_bounds1,gi_probe_cell_size1,pos,ambient,environment,gi_probe_blend_ambient1,gi_probe_multiplier1,normal_mat,ref_vec,roughness,gi_probe_bias1,gi_probe_normal_bias1,spec_accum,diff_accum); + + if (gi_probe2_enabled) { + + gi_probe_compute(gi_probe2,gi_probe_xform2,gi_probe_bounds2,gi_probe_cell_size2,pos,ambient,environment,gi_probe_blend_ambient2,gi_probe_multiplier2,normal_mat,ref_vec,roughness,gi_probe_bias2,gi_probe_normal_bias2,spec_accum,diff_accum); + } + + if (diff_accum.a>0.0) { + diff_accum.rgb/=diff_accum.a; + } + + if (spec_accum.a>0.0) { + spec_accum.rgb/=spec_accum.a; + } + + out_specular+=spec_accum.rgb; + out_ambient+=diff_accum.rgb; + +} + +#endif + + + +void main() { + +#ifdef RENDER_DEPTH_DUAL_PARABOLOID + + if (dp_clip>0.0) + discard; +#endif + + //lay out everything, whathever is unused is optimized away anyway + highp vec3 vertex = vertex_interp; + vec3 albedo = vec3(0.8,0.8,0.8); + vec3 transmission = vec3(0.0); + float metallic = 0.0; + float specular = 0.5; + vec3 emission = vec3(0.0,0.0,0.0); + float roughness = 1.0; + float rim = 0.0; + float rim_tint = 0.0; + float clearcoat=0.0; + float clearcoat_gloss=0.0; + float anisotropy = 1.0; + vec2 anisotropy_flow = vec2(1.0,0.0); + +#if defined(ENABLE_AO) + float ao=1.0; + float ao_light_affect=0.0; +#endif + + float alpha = 1.0; + +#ifdef METERIAL_DOUBLESIDED + float side=float(gl_FrontFacing)*2.0-1.0; +#else + float side=1.0; +#endif + + +#if defined(ALPHA_SCISSOR_USED) + float alpha_scissor = 0.5; +#endif + +#if defined(ENABLE_TANGENT_INTERP) || defined(ENABLE_NORMALMAP) || defined(LIGHT_USE_ANISOTROPY) + vec3 binormal = normalize(binormal_interp)*side; + vec3 tangent = normalize(tangent_interp)*side; +#else + vec3 binormal = vec3(0.0); + vec3 tangent = vec3(0.0); +#endif + vec3 normal = normalize(normal_interp)*side; + +#if defined(ENABLE_UV_INTERP) + vec2 uv = uv_interp; +#endif + +#if defined(ENABLE_UV2_INTERP) + vec2 uv2 = uv2_interp; +#endif + +#if defined(ENABLE_COLOR_INTERP) + vec4 color = color_interp; +#endif + +#if defined(ENABLE_NORMALMAP) + + vec3 normalmap = vec3(0.0); +#endif + + float normaldepth=1.0; + +#if defined(SCREEN_UV_USED) + vec2 screen_uv = gl_FragCoord.xy*screen_pixel_size; +#endif + +#if defined (ENABLE_SSS) + float sss_strength=0.0; +#endif + +{ + + +FRAGMENT_SHADER_CODE + +} + + +#if defined(ALPHA_SCISSOR_USED) + if (alpha<alpha_scissor) { + discard; + } +#endif + +#ifdef USE_OPAQUE_PREPASS + + if (alpha<0.99) { + discard; + } +#endif + +#if defined(ENABLE_NORMALMAP) + + normalmap.xy=normalmap.xy*2.0-1.0; + normalmap.z=sqrt(1.0-dot(normalmap.xy,normalmap.xy)); //always ignore Z, as it can be RG packed, Z may be pos/neg, etc. + + normal = normalize( mix(normal_interp,tangent * normalmap.x + binormal * normalmap.y + normal * normalmap.z,normaldepth) ) * side; + +#endif + +#if defined(LIGHT_USE_ANISOTROPY) + + if (anisotropy>0.01) { + //rotation matrix + mat3 rot = mat3( tangent, binormal, normal ); + //make local to space + tangent = normalize(rot * vec3(anisotropy_flow.x,anisotropy_flow.y,0.0)); + binormal = normalize(rot * vec3(-anisotropy_flow.y,anisotropy_flow.x,0.0)); + } + +#endif + +#ifdef ENABLE_CLIP_ALPHA + if (albedo.a<0.99) { + //used for doublepass and shadowmapping + discard; + } +#endif + +/////////////////////// LIGHTING ////////////////////////////// + + //apply energy conservation + +#ifdef USE_VERTEX_LIGHTING + + vec3 specular_light = specular_light_interp.rgb; + vec3 diffuse_light = diffuse_light_interp.rgb; +#else + + vec3 specular_light = vec3(0.0,0.0,0.0); + vec3 diffuse_light = vec3(0.0,0.0,0.0); + +#endif + + vec3 ambient_light; + vec3 env_reflection_light = vec3(0.0,0.0,0.0); + + vec3 eye_vec = -normalize( vertex_interp ); + + + +#ifdef USE_RADIANCE_MAP + + if (no_ambient_light) { + ambient_light=vec3(0.0,0.0,0.0); + } else { + { + + { //read radiance from dual paraboloid + + vec3 ref_vec = reflect(-eye_vec,normal); //2.0 * ndotv * normal - view; // reflect(v, n); + ref_vec=normalize((radiance_inverse_xform * vec4(ref_vec,0.0)).xyz); + vec3 radiance = textureDualParaboloid(radiance_map,ref_vec,roughness) * bg_energy; + env_reflection_light = radiance; + + } + //no longer a cubemap + //vec3 radiance = textureLod(radiance_cube, r, lod).xyz * ( brdf.x + brdf.y); + + } + + { + + vec3 ambient_dir=normalize((radiance_inverse_xform * vec4(normal,0.0)).xyz); + vec3 env_ambient=textureDualParaboloid(radiance_map,ambient_dir,1.0) * bg_energy; + + ambient_light=mix(ambient_light_color.rgb,env_ambient,radiance_ambient_contribution); + //ambient_light=vec3(0.0,0.0,0.0); + } + } + +#else + + if (no_ambient_light){ + ambient_light=vec3(0.0,0.0,0.0); + } else { + ambient_light=ambient_light_color.rgb; + } +#endif + + ambient_light*=ambient_energy; + + float specular_blob_intensity=1.0; +#if defined(SPECULAR_TOON) + specular_blob_intensity*=specular * 2.0; +#endif + +#if defined(USE_LIGHT_DIRECTIONAL) + + vec3 light_attenuation=vec3(1.0); + + float depth_z = -vertex.z; +#ifdef LIGHT_DIRECTIONAL_SHADOW + +#ifdef LIGHT_USE_PSSM4 + if (depth_z < shadow_split_offsets.w) { +#elif defined(LIGHT_USE_PSSM2) + if (depth_z < shadow_split_offsets.y) { +#else + if (depth_z < shadow_split_offsets.x) { +#endif //LIGHT_USE_PSSM4 + + vec3 pssm_coord; + float pssm_fade=0.0; + +#ifdef LIGHT_USE_PSSM_BLEND + float pssm_blend; + vec3 pssm_coord2; + bool use_blend=true; +#endif + + +#ifdef LIGHT_USE_PSSM4 + + + if (depth_z < shadow_split_offsets.y) { + + if (depth_z < shadow_split_offsets.x) { + + highp vec4 splane=(shadow_matrix1 * vec4(vertex,1.0)); + pssm_coord=splane.xyz/splane.w; + + +#if defined(LIGHT_USE_PSSM_BLEND) + + splane=(shadow_matrix2 * vec4(vertex,1.0)); + pssm_coord2=splane.xyz/splane.w; + pssm_blend=smoothstep(0.0,shadow_split_offsets.x,depth_z); +#endif + + } else { + + highp vec4 splane=(shadow_matrix2 * vec4(vertex,1.0)); + pssm_coord=splane.xyz/splane.w; + +#if defined(LIGHT_USE_PSSM_BLEND) + splane=(shadow_matrix3 * vec4(vertex,1.0)); + pssm_coord2=splane.xyz/splane.w; + pssm_blend=smoothstep(shadow_split_offsets.x,shadow_split_offsets.y,depth_z); +#endif + + } + } else { + + + if (depth_z < shadow_split_offsets.z) { + + highp vec4 splane=(shadow_matrix3 * vec4(vertex,1.0)); + pssm_coord=splane.xyz/splane.w; + +#if defined(LIGHT_USE_PSSM_BLEND) + splane=(shadow_matrix4 * vec4(vertex,1.0)); + pssm_coord2=splane.xyz/splane.w; + pssm_blend=smoothstep(shadow_split_offsets.y,shadow_split_offsets.z,depth_z); +#endif + + } else { + + highp vec4 splane=(shadow_matrix4 * vec4(vertex,1.0)); + pssm_coord=splane.xyz/splane.w; + pssm_fade = smoothstep(shadow_split_offsets.z,shadow_split_offsets.w,depth_z); + +#if defined(LIGHT_USE_PSSM_BLEND) + use_blend=false; + +#endif + + } + } + + + +#endif //LIGHT_USE_PSSM4 + +#ifdef LIGHT_USE_PSSM2 + + if (depth_z < shadow_split_offsets.x) { + + highp vec4 splane=(shadow_matrix1 * vec4(vertex,1.0)); + pssm_coord=splane.xyz/splane.w; + + +#if defined(LIGHT_USE_PSSM_BLEND) + + splane=(shadow_matrix2 * vec4(vertex,1.0)); + pssm_coord2=splane.xyz/splane.w; + pssm_blend=smoothstep(0.0,shadow_split_offsets.x,depth_z); +#endif + + } else { + highp vec4 splane=(shadow_matrix2 * vec4(vertex,1.0)); + pssm_coord=splane.xyz/splane.w; + pssm_fade = smoothstep(shadow_split_offsets.x,shadow_split_offsets.y,depth_z); +#if defined(LIGHT_USE_PSSM_BLEND) + use_blend=false; + +#endif + + } + +#endif //LIGHT_USE_PSSM2 + +#if !defined(LIGHT_USE_PSSM4) && !defined(LIGHT_USE_PSSM2) + { //regular orthogonal + highp vec4 splane=(shadow_matrix1 * vec4(vertex,1.0)); + pssm_coord=splane.xyz/splane.w; + } +#endif + + + //one one sample + + float shadow = sample_shadow(directional_shadow,directional_shadow_pixel_size,pssm_coord.xy,pssm_coord.z,light_clamp); + +#if defined(LIGHT_USE_PSSM_BLEND) + + if (use_blend) { + shadow=mix(shadow, sample_shadow(directional_shadow,directional_shadow_pixel_size,pssm_coord2.xy,pssm_coord2.z,light_clamp),pssm_blend); + } +#endif + +#ifdef USE_CONTACT_SHADOWS + if (shadow>0.01 && shadow_color_contact.a>0.0) { + + float contact_shadow = contact_shadow_compute(vertex,-light_direction_attenuation.xyz,shadow_color_contact.a); + shadow=min(shadow,contact_shadow); + + } +#endif + light_attenuation=mix(mix(shadow_color_contact.rgb,vec3(1.0),shadow),vec3(1.0),pssm_fade); + + + } + + +#endif //LIGHT_DIRECTIONAL_SHADOW + +#ifdef USE_VERTEX_LIGHTING + diffuse_light*=mix(vec3(1.0),light_attenuation,diffuse_light_interp.a); + specular_light*=mix(vec3(1.0),light_attenuation,specular_light_interp.a); + +#else + light_compute(normal,-light_direction_attenuation.xyz,eye_vec,binormal,tangent,light_color_energy.rgb,light_attenuation,albedo,transmission,light_params.z*specular_blob_intensity,roughness,metallic,rim,rim_tint,clearcoat,clearcoat_gloss,anisotropy,diffuse_light,specular_light); +#endif + + +#endif //#USE_LIGHT_DIRECTIONAL + +#ifdef USE_GI_PROBES + gi_probes_compute(vertex,normal,roughness,env_reflection_light,ambient_light); + +#endif + +#ifdef USE_FORWARD_LIGHTING + + + highp vec4 reflection_accum = vec4(0.0,0.0,0.0,0.0); + highp vec4 ambient_accum = vec4(0.0,0.0,0.0,0.0); + for(int i=0;i<reflection_count;i++) { + reflection_process(reflection_indices[i],vertex,normal,binormal,tangent,roughness,anisotropy,ambient_light,env_reflection_light,reflection_accum,ambient_accum); + } + + if (reflection_accum.a>0.0) { + specular_light+=reflection_accum.rgb/reflection_accum.a; + } else { + specular_light+=env_reflection_light; + } + + if (ambient_accum.a>0.0) { + ambient_light+=ambient_accum.rgb/ambient_accum.a; + } + + + +#ifdef USE_VERTEX_LIGHTING + + diffuse_light*=albedo; +#else + + for(int i=0;i<omni_light_count;i++) { + light_process_omni(omni_light_indices[i],vertex,eye_vec,normal,binormal,tangent,albedo,transmission,roughness,metallic,rim,rim_tint,clearcoat,clearcoat_gloss,anisotropy,specular_blob_intensity,diffuse_light,specular_light); + } + + for(int i=0;i<spot_light_count;i++) { + light_process_spot(spot_light_indices[i],vertex,eye_vec,normal,binormal,tangent,albedo,transmission,roughness,metallic,rim,rim_tint,clearcoat,clearcoat_gloss,anisotropy,specular_blob_intensity,diffuse_light,specular_light); + } + +#endif //USE_VERTEX_LIGHTING + +#endif + + + + +#ifdef RENDER_DEPTH +//nothing happens, so a tree-ssa optimizer will result in no fragment shader :) +#else + + specular_light*=reflection_multiplier; + ambient_light*=albedo; //ambient must be multiplied by albedo at the end + +#if defined(ENABLE_AO) + ambient_light*=ao; + ao_light_affect = mix(1.0,ao,ao_light_affect); + specular_light*=ao_light_affect; + diffuse_light*=ao_light_affect; +#endif + + + + //energy conservation + diffuse_light *= 1.0-metallic; // TODO: avoid all diffuse and ambient light calculations when metallic == 1 up to this point + ambient_light *= 1.0-metallic; + + + { + +#if defined(DIFFUSE_TOON) + //simplify for toon, as + specular_light *= specular * metallic * albedo * 2.0; +#else + // Environment brdf approximation (Lazarov 2013) + // see https://www.unrealengine.com/en-US/blog/physically-based-shading-on-mobile + const vec4 c0 = vec4(-1.0, -0.0275, -0.572, 0.022); + const vec4 c1 = vec4( 1.0, 0.0425, 1.04, -0.04); + vec4 r = roughness * c0 + c1; + float ndotv = clamp(dot(normal,eye_vec),0.0,1.0); + float a004 = min( r.x * r.x, exp2( -9.28 * ndotv ) ) * r.x + r.y; + vec2 AB = vec2( -1.04, 1.04 ) * a004 + r.zw; + + vec3 specular_color = metallic_to_specular_color(metallic, specular, albedo); + specular_light *= AB.x * specular_color + AB.y; +#endif + + } + + if (fog_color_enabled.a > 0.5) { + + float fog_amount=0.0; + + + +#ifdef USE_LIGHT_DIRECTIONAL + + vec3 fog_color = mix( fog_color_enabled.rgb, fog_sun_color_amount.rgb,fog_sun_color_amount.a * pow(max( dot(normalize(vertex),-light_direction_attenuation.xyz), 0.0),8.0) ); +#else + + vec3 fog_color = fog_color_enabled.rgb; +#endif + + //apply fog + + if (fog_depth_enabled) { + + float fog_z = smoothstep(fog_depth_begin,z_far,length(vertex)); + + fog_amount = pow(fog_z,fog_depth_curve); + if (fog_transmit_enabled) { + vec3 total_light = emission + ambient_light + specular_light + diffuse_light; + float transmit = pow(fog_z,fog_transmit_curve); + fog_color = mix(max(total_light,fog_color),fog_color,transmit); + } + } + + if (fog_height_enabled) { + float y = (camera_matrix * vec4(vertex,1.0)).y; + fog_amount = max(fog_amount,pow(smoothstep(fog_height_min,fog_height_max,y),fog_height_curve)); + } + + float rev_amount = 1.0 - fog_amount; + + + emission = emission * rev_amount + fog_color * fog_amount; + ambient_light*=rev_amount; + specular_light*rev_amount; + diffuse_light*=rev_amount; + + } + +#ifdef USE_MULTIPLE_RENDER_TARGETS + + +#ifdef SHADELESS + diffuse_buffer=vec4(albedo.rgb,0.0); + specular_buffer=vec4(0.0); + +#else + +#if defined(ENABLE_AO) + + float ambient_scale=0.0; // AO is supplied by material +#else + //approximate ambient scale for SSAO, since we will lack full ambient + float max_emission=max(emission.r,max(emission.g,emission.b)); + float max_ambient=max(ambient_light.r,max(ambient_light.g,ambient_light.b)); + float max_diffuse=max(diffuse_light.r,max(diffuse_light.g,diffuse_light.b)); + float total_ambient = max_ambient+max_diffuse+max_emission; + float ambient_scale = (total_ambient>0.0) ? (max_ambient+ambient_occlusion_affect_light*max_diffuse)/total_ambient : 0.0; +#endif //ENABLE_AO + + diffuse_buffer=vec4(emission+diffuse_light+ambient_light,ambient_scale); + specular_buffer=vec4(specular_light,metallic); + +#endif //SHADELESS + + normal_mr_buffer=vec4(normalize(normal)*0.5+0.5,roughness); + +#if defined (ENABLE_SSS) + sss_buffer = sss_strength; +#endif + + +#else //USE_MULTIPLE_RENDER_TARGETS + + +#ifdef SHADELESS + frag_color=vec4(albedo,alpha); +#else + frag_color=vec4(emission+ambient_light+diffuse_light+specular_light,alpha); +#endif //SHADELESS + + +#endif //USE_MULTIPLE_RENDER_TARGETS + + + +#endif //RENDER_DEPTH + + +} diff --git a/drivers/gles2/shaders/screen_space_reflection.glsl b/drivers/gles2/shaders/screen_space_reflection.glsl new file mode 100644 index 0000000000..b2e6f7a736 --- /dev/null +++ b/drivers/gles2/shaders/screen_space_reflection.glsl @@ -0,0 +1,318 @@ +[vertex] + + +layout(location=0) in highp vec4 vertex_attrib; +layout(location=4) in vec2 uv_in; + +out vec2 uv_interp; +out vec2 pos_interp; + +void main() { + + uv_interp = uv_in; + gl_Position = vertex_attrib; + pos_interp.xy=gl_Position.xy; +} + +[fragment] + + +in vec2 uv_interp; +in vec2 pos_interp; + +uniform sampler2D source_diffuse; //texunit:0 +uniform sampler2D source_normal_roughness; //texunit:1 +uniform sampler2D source_depth; //texunit:2 + +uniform float camera_z_near; +uniform float camera_z_far; + +uniform vec2 viewport_size; +uniform vec2 pixel_size; + +uniform float filter_mipmap_levels; + +uniform mat4 inverse_projection; +uniform mat4 projection; + +uniform int num_steps; +uniform float depth_tolerance; +uniform float distance_fade; +uniform float curve_fade_in; + + +layout(location = 0) out vec4 frag_color; + + +vec2 view_to_screen(vec3 view_pos,out float w) { + vec4 projected = projection * vec4(view_pos, 1.0); + projected.xyz /= projected.w; + projected.xy = projected.xy * 0.5 + 0.5; + w=projected.w; + return projected.xy; +} + + + +#define M_PI 3.14159265359 + +void main() { + + + //// + + vec4 diffuse = texture( source_diffuse, uv_interp ); + vec4 normal_roughness = texture( source_normal_roughness, uv_interp); + + vec3 normal; + + normal = normal_roughness.xyz*2.0-1.0; + + float roughness = normal_roughness.w; + + float depth_tex = texture(source_depth,uv_interp).r; + + vec4 world_pos = inverse_projection * vec4( uv_interp*2.0-1.0, depth_tex*2.0-1.0, 1.0 ); + vec3 vertex = world_pos.xyz/world_pos.w; + + vec3 view_dir = normalize(vertex); + vec3 ray_dir = normalize(reflect(view_dir, normal)); + + if (dot(ray_dir,normal)<0.001) { + frag_color=vec4(0.0); + return; + } + //ray_dir = normalize(view_dir - normal * dot(normal,view_dir) * 2.0); + + //ray_dir = normalize(vec3(1,1,-1)); + + + //////////////// + + + //make ray length and clip it against the near plane (don't want to trace beyond visible) + float ray_len = (vertex.z + ray_dir.z * camera_z_far) > -camera_z_near ? (-camera_z_near - vertex.z) / ray_dir.z : camera_z_far; + vec3 ray_end = vertex + ray_dir*ray_len; + + float w_begin; + vec2 vp_line_begin = view_to_screen(vertex,w_begin); + float w_end; + vec2 vp_line_end = view_to_screen( ray_end, w_end); + vec2 vp_line_dir = vp_line_end-vp_line_begin; + + //we need to interpolate w along the ray, to generate perspective correct reflections + + w_begin = 1.0/w_begin; + w_end = 1.0/w_end; + + + float z_begin = vertex.z*w_begin; + float z_end = ray_end.z*w_end; + + vec2 line_begin = vp_line_begin/pixel_size; + vec2 line_dir = vp_line_dir/pixel_size; + float z_dir = z_end - z_begin; + float w_dir = w_end - w_begin; + + + // clip the line to the viewport edges + + float scale_max_x = min(1.0, 0.99 * (1.0 - vp_line_begin.x) / max(1e-5, vp_line_dir.x)); + float scale_max_y = min(1.0, 0.99 * (1.0 - vp_line_begin.y) / max(1e-5, vp_line_dir.y)); + float scale_min_x = min(1.0, 0.99 * vp_line_begin.x / max(1e-5, -vp_line_dir.x)); + float scale_min_y = min(1.0, 0.99 * vp_line_begin.y / max(1e-5, -vp_line_dir.y)); + float line_clip = min(scale_max_x, scale_max_y) * min(scale_min_x, scale_min_y); + line_dir *= line_clip; + z_dir *= line_clip; + w_dir *=line_clip; + + //clip z and w advance to line advance + vec2 line_advance = normalize(line_dir); //down to pixel + float step_size = length(line_advance)/length(line_dir); + float z_advance = z_dir*step_size; // adapt z advance to line advance + float w_advance = w_dir*step_size; // adapt w advance to line advance + + //make line advance faster if direction is closer to pixel edges (this avoids sampling the same pixel twice) + float advance_angle_adj = 1.0/max(abs(line_advance.x),abs(line_advance.y)); + line_advance*=advance_angle_adj; // adapt z advance to line advance + z_advance*=advance_angle_adj; + w_advance*=advance_angle_adj; + + vec2 pos = line_begin; + float z = z_begin; + float w = w_begin; + float z_from=z/w; + float z_to=z_from; + float depth; + vec2 prev_pos=pos; + + bool found=false; + + float steps_taken=0.0; + + for(int i=0;i<num_steps;i++) { + + pos+=line_advance; + z+=z_advance; + w+=w_advance; + + //convert to linear depth + + depth = texture(source_depth, pos*pixel_size).r * 2.0 - 1.0; +#ifdef USE_ORTHOGONAL_PROJECTION + depth = ((depth + (camera_z_far + camera_z_near)/(camera_z_far - camera_z_near)) * (camera_z_far - camera_z_near))/2.0; +#else + depth = 2.0 * camera_z_near * camera_z_far / (camera_z_far + camera_z_near - depth * (camera_z_far - camera_z_near)); +#endif + depth=-depth; + + z_from = z_to; + z_to = z/w; + + if (depth>z_to) { + //if depth was surpassed + if (depth<=max(z_to,z_from)+depth_tolerance) { + //check the depth tolerance + found=true; + } + break; + } + + steps_taken+=1.0; + prev_pos=pos; + } + + + + + if (found) { + + float margin_blend=1.0; + + + vec2 margin = vec2((viewport_size.x+viewport_size.y)*0.5*0.05); //make a uniform margin + if (any(bvec4(lessThan(pos,-margin),greaterThan(pos,viewport_size+margin)))) { + //clip outside screen + margin + frag_color=vec4(0.0); + return; + } + + { + //blend fading out towards external margin + vec2 margin_grad = mix(pos-viewport_size,-pos,lessThan(pos,vec2(0.0))); + margin_blend = 1.0-smoothstep(0.0,margin.x,max(margin_grad.x,margin_grad.y)); + //margin_blend=1.0; + + } + + vec2 final_pos; + float grad; + grad=steps_taken/float(num_steps); + float initial_fade = curve_fade_in==0.0 ? 1.0 : pow(clamp(grad,0.0,1.0),curve_fade_in); + float fade = pow(clamp(1.0-grad,0.0,1.0),distance_fade)*initial_fade; + final_pos=pos; + + + + + + + +#ifdef REFLECT_ROUGHNESS + + + vec4 final_color; + //if roughness is enabled, do screen space cone tracing + if (roughness > 0.001) { + /////////////////////////////////////////////////////////////////////////////////////// + //use a blurred version (in consecutive mipmaps) of the screen to simulate roughness + + float gloss = 1.0-roughness; + float cone_angle = roughness * M_PI * 0.5; + vec2 cone_dir = final_pos - line_begin; + float cone_len = length(cone_dir); + cone_dir = normalize(cone_dir); //will be used normalized from now on + float max_mipmap = filter_mipmap_levels - 1.0; + float gloss_mult=gloss; + + float rem_alpha=1.0; + final_color = vec4(0.0); + + for(int i=0;i<7;i++) { + + float op_len = 2.0 * tan(cone_angle) * cone_len; //opposite side of iso triangle + float radius; + { + //fit to sphere inside cone (sphere ends at end of cone), something like this: + // ___ + // \O/ + // V + // + // as it avoids bleeding from beyond the reflection as much as possible. As a plus + // it also makes the rough reflection more elongated. + float a = op_len; + float h = cone_len; + float a2 = a * a; + float fh2 = 4.0f * h * h; + radius = (a * (sqrt(a2 + fh2) - a)) / (4.0f * h); + } + + //find the place where screen must be sampled + vec2 sample_pos = ( line_begin + cone_dir * (cone_len - radius) ) * pixel_size; + //radius is in pixels, so it's natural that log2(radius) maps to the right mipmap for the amount of pixels + float mipmap = clamp( log2( radius ), 0.0, max_mipmap ); + + //mipmap = max(mipmap-1.0,0.0); + //do sampling + + vec4 sample_color; + { + sample_color = textureLod(source_diffuse,sample_pos,mipmap); + } + + //multiply by gloss + sample_color.rgb*=gloss_mult; + sample_color.a=gloss_mult; + + rem_alpha -= sample_color.a; + if(rem_alpha < 0.0) { + sample_color.rgb *= (1.0 - abs(rem_alpha)); + } + + final_color+=sample_color; + + if (final_color.a>=0.95) { + // This code of accumulating gloss and aborting on near one + // makes sense when you think of cone tracing. + // Think of it as if roughness was 0, then we could abort on the first + // iteration. For lesser roughness values, we need more iterations, but + // each needs to have less influence given the sphere is smaller + break; + } + + cone_len-=radius*2.0; //go to next (smaller) circle. + + gloss_mult*=gloss; + + + } + } else { + final_color = textureLod(source_diffuse,final_pos*pixel_size,0.0); + } + + frag_color = vec4(final_color.rgb,fade*margin_blend); + +#else + frag_color = vec4(textureLod(source_diffuse,final_pos*pixel_size,0.0).rgb,fade*margin_blend); +#endif + + + + } else { + frag_color = vec4(0.0,0.0,0.0,0.0); + } + + + +} + diff --git a/drivers/gles2/shaders/ssao.glsl b/drivers/gles2/shaders/ssao.glsl new file mode 100644 index 0000000000..219f0957e0 --- /dev/null +++ b/drivers/gles2/shaders/ssao.glsl @@ -0,0 +1,293 @@ +[vertex] + + +layout(location=0) in highp vec4 vertex_attrib; + +void main() { + + gl_Position = vertex_attrib; + gl_Position.z=1.0; +} + +[fragment] + +#define TWO_PI 6.283185307179586476925286766559 + +#ifdef SSAO_QUALITY_HIGH + +#define NUM_SAMPLES (80) + +#endif + +#ifdef SSAO_QUALITY_LOW + +#define NUM_SAMPLES (15) + +#endif + +#if !defined(SSAO_QUALITY_LOW) && !defined(SSAO_QUALITY_HIGH) + +#define NUM_SAMPLES (40) + +#endif + +// If using depth mip levels, the log of the maximum pixel offset before we need to switch to a lower +// miplevel to maintain reasonable spatial locality in the cache +// If this number is too small (< 3), too many taps will land in the same pixel, and we'll get bad variance that manifests as flashing. +// If it is too high (> 5), we'll get bad performance because we're not using the MIP levels effectively +#define LOG_MAX_OFFSET (3) + +// This must be less than or equal to the MAX_MIP_LEVEL defined in SSAO.cpp +#define MAX_MIP_LEVEL (4) + +// This is the number of turns around the circle that the spiral pattern makes. This should be prime to prevent +// taps from lining up. This particular choice was tuned for NUM_SAMPLES == 9 + +const int ROTATIONS[] = int[]( 1, 1, 2, 3, 2, 5, 2, 3, 2, +3, 3, 5, 5, 3, 4, 7, 5, 5, 7, +9, 8, 5, 5, 7, 7, 7, 8, 5, 8, +11, 12, 7, 10, 13, 8, 11, 8, 7, 14, +11, 11, 13, 12, 13, 19, 17, 13, 11, 18, +19, 11, 11, 14, 17, 21, 15, 16, 17, 18, +13, 17, 11, 17, 19, 18, 25, 18, 19, 19, +29, 21, 19, 27, 31, 29, 21, 18, 17, 29, +31, 31, 23, 18, 25, 26, 25, 23, 19, 34, +19, 27, 21, 25, 39, 29, 17, 21, 27 ); + +//#define NUM_SPIRAL_TURNS (7) +const int NUM_SPIRAL_TURNS = ROTATIONS[NUM_SAMPLES-1]; + +uniform sampler2D source_depth; //texunit:0 +uniform highp usampler2D source_depth_mipmaps; //texunit:1 +uniform sampler2D source_normal; //texunit:2 + +uniform ivec2 screen_size; +uniform float camera_z_far; +uniform float camera_z_near; + +uniform float intensity_div_r6; +uniform float radius; + +#ifdef ENABLE_RADIUS2 +uniform float intensity_div_r62; +uniform float radius2; +#endif + +uniform float bias; +uniform float proj_scale; + +layout(location = 0) out float visibility; + +uniform vec4 proj_info; + +vec3 reconstructCSPosition(vec2 S, float z) { +#ifdef USE_ORTHOGONAL_PROJECTION + return vec3((S.xy * proj_info.xy + proj_info.zw), z); +#else + return vec3((S.xy * proj_info.xy + proj_info.zw) * z, z); + +#endif +} + +vec3 getPosition(ivec2 ssP) { + vec3 P; + P.z = texelFetch(source_depth, ssP, 0).r; + + P.z = P.z * 2.0 - 1.0; +#ifdef USE_ORTHOGONAL_PROJECTION + P.z = ((P.z + (camera_z_far + camera_z_near)/(camera_z_far - camera_z_near)) * (camera_z_far - camera_z_near))/2.0; +#else + P.z = 2.0 * camera_z_near * camera_z_far / (camera_z_far + camera_z_near - P.z * (camera_z_far - camera_z_near)); +#endif + P.z = -P.z; + + // Offset to pixel center + P = reconstructCSPosition(vec2(ssP) + vec2(0.5), P.z); + return P; +} + +/** Reconstructs screen-space unit normal from screen-space position */ +vec3 reconstructCSFaceNormal(vec3 C) { + return normalize(cross(dFdy(C), dFdx(C))); +} + + + +/** Returns a unit vector and a screen-space radius for the tap on a unit disk (the caller should scale by the actual disk radius) */ +vec2 tapLocation(int sampleNumber, float spinAngle, out float ssR){ + // Radius relative to ssR + float alpha = (float(sampleNumber) + 0.5) * (1.0 / float(NUM_SAMPLES)); + float angle = alpha * (float(NUM_SPIRAL_TURNS) * 6.28) + spinAngle; + + ssR = alpha; + return vec2(cos(angle), sin(angle)); +} + + +/** Read the camera-space position of the point at screen-space pixel ssP + unitOffset * ssR. Assumes length(unitOffset) == 1 */ +vec3 getOffsetPosition(ivec2 ssC, vec2 unitOffset, float ssR) { + // Derivation: + // mipLevel = floor(log(ssR / MAX_OFFSET)); + int mipLevel = clamp(int(floor(log2(ssR))) - LOG_MAX_OFFSET, 0, MAX_MIP_LEVEL); + + ivec2 ssP = ivec2(ssR * unitOffset) + ssC; + + vec3 P; + + // We need to divide by 2^mipLevel to read the appropriately scaled coordinate from a MIP-map. + // Manually clamp to the texture size because texelFetch bypasses the texture unit + ivec2 mipP = clamp(ssP >> mipLevel, ivec2(0), (screen_size >> mipLevel) - ivec2(1)); + + + if (mipLevel < 1) { + //read from depth buffer + P.z = texelFetch(source_depth, mipP, 0).r; + P.z = P.z * 2.0 - 1.0; +#ifdef USE_ORTHOGONAL_PROJECTION + P.z = ((P.z + (camera_z_far + camera_z_near)/(camera_z_far - camera_z_near)) * (camera_z_far - camera_z_near))/2.0; +#else + P.z = 2.0 * camera_z_near * camera_z_far / (camera_z_far + camera_z_near - P.z * (camera_z_far - camera_z_near)); + +#endif + P.z = -P.z; + + } else { + //read from mipmaps + uint d = texelFetch(source_depth_mipmaps, mipP, mipLevel-1).r; + P.z = -(float(d)/65535.0)*camera_z_far; + } + + + // Offset to pixel center + P = reconstructCSPosition(vec2(ssP) + vec2(0.5), P.z); + + return P; +} + + + +/** Compute the occlusion due to sample with index \a i about the pixel at \a ssC that corresponds + to camera-space point \a C with unit normal \a n_C, using maximum screen-space sampling radius \a ssDiskRadius + + Note that units of H() in the HPG12 paper are meters, not + unitless. The whole falloff/sampling function is therefore + unitless. In this implementation, we factor out (9 / radius). + + Four versions of the falloff function are implemented below +*/ +float sampleAO(in ivec2 ssC, in vec3 C, in vec3 n_C, in float ssDiskRadius,in float p_radius, in int tapIndex, in float randomPatternRotationAngle) { + // Offset on the unit disk, spun for this pixel + float ssR; + vec2 unitOffset = tapLocation(tapIndex, randomPatternRotationAngle, ssR); + ssR *= ssDiskRadius; + + // The occluding point in camera space + vec3 Q = getOffsetPosition(ssC, unitOffset, ssR); + + vec3 v = Q - C; + + float vv = dot(v, v); + float vn = dot(v, n_C); + + const float epsilon = 0.01; + float radius2 = p_radius*p_radius; + + // A: From the HPG12 paper + // Note large epsilon to avoid overdarkening within cracks + //return float(vv < radius2) * max((vn - bias) / (epsilon + vv), 0.0) * radius2 * 0.6; + + // B: Smoother transition to zero (lowers contrast, smoothing out corners). [Recommended] + float f=max(radius2 - vv, 0.0); + return f * f * f * max((vn - bias) / (epsilon + vv), 0.0); + + // C: Medium contrast (which looks better at high radii), no division. Note that the + // contribution still falls off with radius^2, but we've adjusted the rate in a way that is + // more computationally efficient and happens to be aesthetically pleasing. + // return 4.0 * max(1.0 - vv * invRadius2, 0.0) * max(vn - bias, 0.0); + + // D: Low contrast, no division operation + // return 2.0 * float(vv < radius * radius) * max(vn - bias, 0.0); +} + + + +void main() { + + + // Pixel being shaded + ivec2 ssC = ivec2(gl_FragCoord.xy); + + // World space point being shaded + vec3 C = getPosition(ssC); + +/* if (C.z <= -camera_z_far*0.999) { + // We're on the skybox + visibility=1.0; + return; + }*/ + + //visibility=-C.z/camera_z_far; + //return; +#if 0 + vec3 n_C = texelFetch(source_normal,ssC,0).rgb * 2.0 - 1.0; +#else + vec3 n_C = reconstructCSFaceNormal(C); + n_C = -n_C; +#endif + + // Hash function used in the HPG12 AlchemyAO paper + float randomPatternRotationAngle = mod(float((3 * ssC.x ^ ssC.y + ssC.x * ssC.y) * 10), TWO_PI); + + // Reconstruct normals from positions. These will lead to 1-pixel black lines + // at depth discontinuities, however the blur will wipe those out so they are not visible + // in the final image. + + // Choose the screen-space sample radius + // proportional to the projected area of the sphere +#ifdef USE_ORTHOGONAL_PROJECTION + float ssDiskRadius = -proj_scale * radius; +#else + float ssDiskRadius = -proj_scale * radius / C.z; +#endif + float sum = 0.0; + for (int i = 0; i < NUM_SAMPLES; ++i) { + sum += sampleAO(ssC, C, n_C, ssDiskRadius, radius,i, randomPatternRotationAngle); + } + + float A = max(0.0, 1.0 - sum * intensity_div_r6 * (5.0 / float(NUM_SAMPLES))); + +#ifdef ENABLE_RADIUS2 + + //go again for radius2 + randomPatternRotationAngle = mod(float((5 * ssC.x ^ ssC.y + ssC.x * ssC.y) * 11), TWO_PI); + + // Reconstruct normals from positions. These will lead to 1-pixel black lines + // at depth discontinuities, however the blur will wipe those out so they are not visible + // in the final image. + + // Choose the screen-space sample radius + // proportional to the projected area of the sphere + ssDiskRadius = -proj_scale * radius2 / C.z; + + sum = 0.0; + for (int i = 0; i < NUM_SAMPLES; ++i) { + sum += sampleAO(ssC, C, n_C, ssDiskRadius,radius2, i, randomPatternRotationAngle); + } + + A= min(A,max(0.0, 1.0 - sum * intensity_div_r62 * (5.0 / float(NUM_SAMPLES)))); +#endif + // Bilateral box-filter over a quad for free, respecting depth edges + // (the difference that this makes is subtle) + if (abs(dFdx(C.z)) < 0.02) { + A -= dFdx(A) * (float(ssC.x & 1) - 0.5); + } + if (abs(dFdy(C.z)) < 0.02) { + A -= dFdy(A) * (float(ssC.y & 1) - 0.5); + } + + visibility = A; + +} + + + diff --git a/drivers/gles2/shaders/ssao_blur.glsl b/drivers/gles2/shaders/ssao_blur.glsl new file mode 100644 index 0000000000..472dc21acf --- /dev/null +++ b/drivers/gles2/shaders/ssao_blur.glsl @@ -0,0 +1,124 @@ +[vertex] + + +layout(location=0) in highp vec4 vertex_attrib; + + +void main() { + + gl_Position = vertex_attrib; + gl_Position.z=1.0; +} + +[fragment] + + +uniform sampler2D source_ssao; //texunit:0 +uniform sampler2D source_depth; //texunit:1 +uniform sampler2D source_normal; //texunit:3 + + +layout(location = 0) out float visibility; + + +////////////////////////////////////////////////////////////////////////////////////////////// +// Tunable Parameters: + +/** Increase to make depth edges crisper. Decrease to reduce flicker. */ +uniform float edge_sharpness; + +/** Step in 2-pixel intervals since we already blurred against neighbors in the + first AO pass. This constant can be increased while R decreases to improve + performance at the expense of some dithering artifacts. + + Morgan found that a scale of 3 left a 1-pixel checkerboard grid that was + unobjectionable after shading was applied but eliminated most temporal incoherence + from using small numbers of sample taps. + */ + +uniform int filter_scale; + +/** Filter radius in pixels. This will be multiplied by SCALE. */ +#define R (4) + + +////////////////////////////////////////////////////////////////////////////////////////////// + + +// Gaussian coefficients +const float gaussian[R + 1] = +// float[](0.356642, 0.239400, 0.072410, 0.009869); +// float[](0.398943, 0.241971, 0.053991, 0.004432, 0.000134); // stddev = 1.0 + float[](0.153170, 0.144893, 0.122649, 0.092902, 0.062970); // stddev = 2.0 +// float[](0.111220, 0.107798, 0.098151, 0.083953, 0.067458, 0.050920, 0.036108); // stddev = 3.0 + +/** (1, 0) or (0, 1)*/ +uniform ivec2 axis; + +uniform float camera_z_far; +uniform float camera_z_near; + +uniform ivec2 screen_size; + +void main() { + + ivec2 ssC = ivec2(gl_FragCoord.xy); + + float depth = texelFetch(source_depth, ssC, 0).r; + //vec3 normal = texelFetch(source_normal,ssC,0).rgb * 2.0 - 1.0; + + depth = depth * 2.0 - 1.0; + depth = 2.0 * camera_z_near * camera_z_far / (camera_z_far + camera_z_near - depth * (camera_z_far - camera_z_near)); + + float depth_divide = 1.0 / camera_z_far; + +// depth*=depth_divide; + + /* + if (depth > camera_z_far*0.999) { + discard;//skybox + } + */ + + float sum = texelFetch(source_ssao, ssC, 0).r; + + // Base weight for depth falloff. Increase this for more blurriness, + // decrease it for better edge discrimination + float BASE = gaussian[0]; + float totalWeight = BASE; + sum *= totalWeight; + + ivec2 clamp_limit = screen_size - ivec2(1); + + for (int r = -R; r <= R; ++r) { + // We already handled the zero case above. This loop should be unrolled and the static branch optimized out, + // so the IF statement has no runtime cost + if (r != 0) { + + ivec2 ppos = ssC + axis * (r * filter_scale); + float value = texelFetch(source_ssao, clamp(ppos,ivec2(0),clamp_limit), 0).r; + ivec2 rpos = clamp(ppos,ivec2(0),clamp_limit); + float temp_depth = texelFetch(source_depth, rpos, 0).r; + //vec3 temp_normal = texelFetch(source_normal, rpos, 0).rgb * 2.0 - 1.0; + + temp_depth = temp_depth * 2.0 - 1.0; + temp_depth = 2.0 * camera_z_near * camera_z_far / (camera_z_far + camera_z_near - temp_depth * (camera_z_far - camera_z_near)); +// temp_depth *= depth_divide; + + // spatial domain: offset gaussian tap + float weight = 0.3 + gaussian[abs(r)]; + //weight *= max(0.0,dot(temp_normal,normal)); + + // range domain (the "bilateral" weight). As depth difference increases, decrease weight. + weight *= max(0.0, 1.0 + - edge_sharpness * abs(temp_depth - depth) + ); + + sum += value * weight; + totalWeight += weight; + } + } + + const float epsilon = 0.0001; + visibility = sum / (totalWeight + epsilon); +} diff --git a/drivers/gles2/shaders/ssao_minify.glsl b/drivers/gles2/shaders/ssao_minify.glsl new file mode 100644 index 0000000000..647c762438 --- /dev/null +++ b/drivers/gles2/shaders/ssao_minify.glsl @@ -0,0 +1,59 @@ +[vertex] + + +layout(location=0) in highp vec4 vertex_attrib; + +void main() { + + gl_Position = vertex_attrib; +} + +[fragment] + + +#ifdef MINIFY_START + +#define SDEPTH_TYPE highp sampler2D +uniform float camera_z_far; +uniform float camera_z_near; + +#else + +#define SDEPTH_TYPE mediump usampler2D + +#endif + +uniform SDEPTH_TYPE source_depth; //texunit:0 + +uniform ivec2 from_size; +uniform int source_mipmap; + +layout(location = 0) out mediump uint depth; + +void main() { + + + ivec2 ssP = ivec2(gl_FragCoord.xy); + + // Rotated grid subsampling to avoid XY directional bias or Z precision bias while downsampling. + // On DX9, the bit-and can be implemented with floating-point modulo + +#ifdef MINIFY_START + float fdepth = texelFetch(source_depth, clamp(ssP * 2 + ivec2(ssP.y & 1, ssP.x & 1), ivec2(0), from_size - ivec2(1)), source_mipmap).r; + fdepth = fdepth * 2.0 - 1.0; +#ifdef USE_ORTHOGONAL_PROJECTION + fdepth = ((fdepth + (camera_z_far + camera_z_near)/(camera_z_far - camera_z_near)) * (camera_z_far - camera_z_near))/2.0; +#else + fdepth = 2.0 * camera_z_near * camera_z_far / (camera_z_far + camera_z_near - fdepth * (camera_z_far - camera_z_near)); +#endif + fdepth /= camera_z_far; + depth = uint(clamp(fdepth*65535.0,0.0,65535.0)); + +#else + depth = texelFetch(source_depth, clamp(ssP * 2 + ivec2(ssP.y & 1, ssP.x & 1), ivec2(0), from_size - ivec2(1)), source_mipmap).r; +#endif + + +} + + diff --git a/drivers/gles2/shaders/subsurf_scattering.glsl b/drivers/gles2/shaders/subsurf_scattering.glsl new file mode 100644 index 0000000000..fc66d66198 --- /dev/null +++ b/drivers/gles2/shaders/subsurf_scattering.glsl @@ -0,0 +1,192 @@ +[vertex] + + +layout(location=0) in highp vec4 vertex_attrib; +layout(location=4) in vec2 uv_in; + +out vec2 uv_interp; + + +void main() { + + uv_interp = uv_in; + gl_Position = vertex_attrib; +} + +[fragment] + +//#define QUALIFIER uniform // some guy on the interweb says it may be faster with this +#define QUALIFIER const + +#ifdef USE_25_SAMPLES + +const int kernel_size=25; +QUALIFIER vec2 kernel[25] = vec2[] ( + vec2(0.530605, 0.0), + vec2(0.000973794, -3.0), + vec2(0.00333804, -2.52083), + vec2(0.00500364, -2.08333), + vec2(0.00700976, -1.6875), + vec2(0.0094389, -1.33333), + vec2(0.0128496, -1.02083), + vec2(0.017924, -0.75), + vec2(0.0263642, -0.520833), + vec2(0.0410172, -0.333333), + vec2(0.0493588, -0.1875), + vec2(0.0402784, -0.0833333), + vec2(0.0211412, -0.0208333), + vec2(0.0211412, 0.0208333), + vec2(0.0402784, 0.0833333), + vec2(0.0493588, 0.1875), + vec2(0.0410172, 0.333333), + vec2(0.0263642, 0.520833), + vec2(0.017924, 0.75), + vec2(0.0128496, 1.02083), + vec2(0.0094389, 1.33333), + vec2(0.00700976, 1.6875), + vec2(0.00500364, 2.08333), + vec2(0.00333804, 2.52083), + vec2(0.000973794, 3.0) +); + +#endif //USE_25_SAMPLES + +#ifdef USE_17_SAMPLES + +const int kernel_size=17; + +QUALIFIER vec2 kernel[17] = vec2[]( + vec2(0.536343, 0.0), + vec2(0.00317394, -2.0), + vec2(0.0100386, -1.53125), + vec2(0.0144609, -1.125), + vec2(0.0216301, -0.78125), + vec2(0.0347317, -0.5), + vec2(0.0571056, -0.28125), + vec2(0.0582416, -0.125), + vec2(0.0324462, -0.03125), + vec2(0.0324462, 0.03125), + vec2(0.0582416, 0.125), + vec2(0.0571056, 0.28125), + vec2(0.0347317, 0.5), + vec2(0.0216301, 0.78125), + vec2(0.0144609, 1.125), + vec2(0.0100386, 1.53125), + vec2(0.00317394,2.0) +); + +#endif //USE_17_SAMPLES + + +#ifdef USE_11_SAMPLES + +const int kernel_size=11; + +QUALIFIER vec2 kernel[11] = vec2[]( + vec2(0.560479, 0.0), + vec2(0.00471691, -2.0), + vec2(0.0192831, -1.28), + vec2(0.03639, -0.72), + vec2(0.0821904, -0.32), + vec2(0.0771802, -0.08), + vec2(0.0771802, 0.08), + vec2(0.0821904, 0.32), + vec2(0.03639, 0.72), + vec2(0.0192831, 1.28), + vec2(0.00471691,2.0) +); + +#endif //USE_11_SAMPLES + + + +uniform float max_radius; +uniform float camera_z_far; +uniform float camera_z_near; +uniform float unit_size; +uniform vec2 dir; +in vec2 uv_interp; + +uniform sampler2D source_diffuse; //texunit:0 +uniform sampler2D source_sss; //texunit:1 +uniform sampler2D source_depth; //texunit:2 + +layout(location = 0) out vec4 frag_color; + +void main() { + + float strength = texture(source_sss,uv_interp).r; + strength*=strength; //stored as sqrt + + // Fetch color of current pixel: + vec4 base_color = texture(source_diffuse, uv_interp); + + + if (strength>0.0) { + + + // Fetch linear depth of current pixel: + float depth = texture(source_depth, uv_interp).r * 2.0 - 1.0; +#ifdef USE_ORTHOGONAL_PROJECTION + depth = ((depth + (camera_z_far + camera_z_near)/(camera_z_far - camera_z_near)) * (camera_z_far - camera_z_near))/2.0; + float scale = unit_size; //remember depth is negative by default in OpenGL +#else + depth = 2.0 * camera_z_near * camera_z_far / (camera_z_far + camera_z_near - depth * (camera_z_far - camera_z_near)); + float scale = unit_size / depth; //remember depth is negative by default in OpenGL +#endif + + + + // Calculate the final step to fetch the surrounding pixels: + vec2 step = max_radius * scale * dir; + step *= strength; // Modulate it using the alpha channel. + step *= 1.0 / 3.0; // Divide by 3 as the kernels range from -3 to 3. + + // Accumulate the center sample: + vec3 color_accum = base_color.rgb; + color_accum *= kernel[0].x; +#ifdef ENABLE_STRENGTH_WEIGHTING + float color_weight = kernel[0].x; +#endif + + // Accumulate the other samples: + for (int i = 1; i < kernel_size; i++) { + // Fetch color and depth for current sample: + vec2 offset = uv_interp + kernel[i].y * step; + vec3 color = texture(source_diffuse, offset).rgb; + +#ifdef ENABLE_FOLLOW_SURFACE + // If the difference in depth is huge, we lerp color back to "colorM": + float depth_cmp = texture(source_depth, offset).r *2.0 - 1.0; + +#ifdef USE_ORTHOGONAL_PROJECTION + depth_cmp = ((depth_cmp + (camera_z_far + camera_z_near)/(camera_z_far - camera_z_near)) * (camera_z_far - camera_z_near))/2.0; +#else + depth_cmp = 2.0 * camera_z_near * camera_z_far / (camera_z_far + camera_z_near - depth_cmp * (camera_z_far - camera_z_near)); +#endif + + float s = clamp(300.0f * scale * + max_radius * abs(depth - depth_cmp),0.0,1.0); + color = mix(color, base_color.rgb, s); +#endif + + // Accumulate: + color*=kernel[i].x; + +#ifdef ENABLE_STRENGTH_WEIGHTING + float color_s = texture(source_sss, offset).r; + color_weight+=color_s * kernel[i].x; + color*=color_s; +#endif + color_accum += color; + + } + +#ifdef ENABLE_STRENGTH_WEIGHTING + color_accum/=color_weight; +#endif + frag_color = vec4(color_accum,base_color.a); //keep alpha (used for SSAO) + } else { + frag_color = base_color; + } +} diff --git a/drivers/gles2/shaders/tonemap.glsl b/drivers/gles2/shaders/tonemap.glsl new file mode 100644 index 0000000000..2f671158b2 --- /dev/null +++ b/drivers/gles2/shaders/tonemap.glsl @@ -0,0 +1,323 @@ +[vertex] + + +layout(location=0) in highp vec4 vertex_attrib; +layout(location=4) in vec2 uv_in; + +out vec2 uv_interp; + +void main() { + + gl_Position = vertex_attrib; + uv_interp = uv_in; +#ifdef V_FLIP + uv_interp.y = 1.0-uv_interp.y; +#endif + +} + +[fragment] + +#if !defined(GLES_OVER_GL) +precision mediump float; +#endif + + +in vec2 uv_interp; + +uniform highp sampler2D source; //texunit:0 + +uniform float exposure; +uniform float white; + +#ifdef USE_AUTO_EXPOSURE + +uniform highp sampler2D source_auto_exposure; //texunit:1 +uniform highp float auto_exposure_grey; + +#endif + +#if defined(USE_GLOW_LEVEL1) || defined(USE_GLOW_LEVEL2) || defined(USE_GLOW_LEVEL3) || defined(USE_GLOW_LEVEL4) || defined(USE_GLOW_LEVEL5) || defined(USE_GLOW_LEVEL6) || defined(USE_GLOW_LEVEL7) + +uniform highp sampler2D source_glow; //texunit:2 +uniform highp float glow_intensity; + +#endif + +#ifdef USE_BCS + +uniform vec3 bcs; + +#endif + +#ifdef USE_COLOR_CORRECTION + +uniform sampler2D color_correction; //texunit:3 + +#endif + + +layout(location = 0) out vec4 frag_color; + +#ifdef USE_GLOW_FILTER_BICUBIC + +// w0, w1, w2, and w3 are the four cubic B-spline basis functions +float w0(float a) +{ + return (1.0/6.0)*(a*(a*(-a + 3.0) - 3.0) + 1.0); +} + +float w1(float a) +{ + return (1.0/6.0)*(a*a*(3.0*a - 6.0) + 4.0); +} + +float w2(float a) +{ + return (1.0/6.0)*(a*(a*(-3.0*a + 3.0) + 3.0) + 1.0); +} + +float w3(float a) +{ + return (1.0/6.0)*(a*a*a); +} + +// g0 and g1 are the two amplitude functions +float g0(float a) +{ + return w0(a) + w1(a); +} + +float g1(float a) +{ + return w2(a) + w3(a); +} + +// h0 and h1 are the two offset functions +float h0(float a) +{ + return -1.0 + w1(a) / (w0(a) + w1(a)); +} + +float h1(float a) +{ + return 1.0 + w3(a) / (w2(a) + w3(a)); +} + +uniform ivec2 glow_texture_size; + +vec4 texture2D_bicubic(sampler2D tex, vec2 uv,int p_lod) +{ + float lod=float(p_lod); + vec2 tex_size = vec2(glow_texture_size >> p_lod); + vec2 pixel_size =1.0/tex_size; + uv = uv*tex_size + 0.5; + vec2 iuv = floor( uv ); + vec2 fuv = fract( uv ); + + float g0x = g0(fuv.x); + float g1x = g1(fuv.x); + float h0x = h0(fuv.x); + float h1x = h1(fuv.x); + float h0y = h0(fuv.y); + float h1y = h1(fuv.y); + + vec2 p0 = (vec2(iuv.x + h0x, iuv.y + h0y) - 0.5) * pixel_size; + vec2 p1 = (vec2(iuv.x + h1x, iuv.y + h0y) - 0.5) * pixel_size; + vec2 p2 = (vec2(iuv.x + h0x, iuv.y + h1y) - 0.5) * pixel_size; + vec2 p3 = (vec2(iuv.x + h1x, iuv.y + h1y) - 0.5) * pixel_size; + + return g0(fuv.y) * (g0x * textureLod(tex, p0,lod) + + g1x * textureLod(tex, p1,lod)) + + g1(fuv.y) * (g0x * textureLod(tex, p2,lod) + + g1x * textureLod(tex, p3,lod)); +} + + + +#define GLOW_TEXTURE_SAMPLE(m_tex,m_uv,m_lod) texture2D_bicubic(m_tex,m_uv,m_lod) + +#else + +#define GLOW_TEXTURE_SAMPLE(m_tex,m_uv,m_lod) textureLod(m_tex,m_uv,float(m_lod)) + +#endif + + +vec3 tonemap_filmic(vec3 color,float white) { + + float A = 0.15; + float B = 0.50; + float C = 0.10; + float D = 0.20; + float E = 0.02; + float F = 0.30; + float W = 11.2; + + vec3 coltn = ((color*(A*color+C*B)+D*E)/(color*(A*color+B)+D*F))-E/F; + float whitetn = ((white*(A*white+C*B)+D*E)/(white*(A*white+B)+D*F))-E/F; + + return coltn/whitetn; + +} + +vec3 tonemap_aces(vec3 color) { + float a = 2.51f; + float b = 0.03f; + float c = 2.43f; + float d = 0.59f; + float e = 0.14f; + return color = clamp((color*(a*color+b))/(color*(c*color+d)+e),vec3(0.0),vec3(1.0)); +} + +vec3 tonemap_reindhart(vec3 color,float white) { + + return ( color * ( 1.0 + ( color / ( white) ) ) ) / ( 1.0 + color ); +} + +void main() { + + vec4 color = textureLod(source, uv_interp, 0.0); + +#ifdef USE_AUTO_EXPOSURE + + color/=texelFetch(source_auto_exposure,ivec2(0,0),0).r/auto_exposure_grey; +#endif + + color*=exposure; + +#if defined(USE_GLOW_LEVEL1) || defined(USE_GLOW_LEVEL2) || defined(USE_GLOW_LEVEL3) || defined(USE_GLOW_LEVEL4) || defined(USE_GLOW_LEVEL5) || defined(USE_GLOW_LEVEL6) || defined(USE_GLOW_LEVEL7) +#define USING_GLOW +#endif + +#if defined(USING_GLOW) + vec3 glow = vec3(0.0); + +#ifdef USE_GLOW_LEVEL1 + + glow+=GLOW_TEXTURE_SAMPLE(source_glow,uv_interp,1).rgb; +#endif + +#ifdef USE_GLOW_LEVEL2 + glow+=GLOW_TEXTURE_SAMPLE(source_glow,uv_interp,2).rgb; +#endif + +#ifdef USE_GLOW_LEVEL3 + glow+=GLOW_TEXTURE_SAMPLE(source_glow,uv_interp,3).rgb; +#endif + +#ifdef USE_GLOW_LEVEL4 + glow+=GLOW_TEXTURE_SAMPLE(source_glow,uv_interp,4).rgb; +#endif + +#ifdef USE_GLOW_LEVEL5 + glow+=GLOW_TEXTURE_SAMPLE(source_glow,uv_interp,5).rgb; +#endif + +#ifdef USE_GLOW_LEVEL6 + glow+=GLOW_TEXTURE_SAMPLE(source_glow,uv_interp,6).rgb; +#endif + +#ifdef USE_GLOW_LEVEL7 + glow+=GLOW_TEXTURE_SAMPLE(source_glow,uv_interp,7).rgb; +#endif + + + glow *= glow_intensity; + +#endif + + +#ifdef USE_REINDHART_TONEMAPPER + + color.rgb = tonemap_reindhart(color.rgb,white); + +# if defined(USING_GLOW) + glow = tonemap_reindhart(glow,white); +# endif + +#endif + +#ifdef USE_FILMIC_TONEMAPPER + + color.rgb = tonemap_filmic(color.rgb,white); + +# if defined(USING_GLOW) + glow = tonemap_filmic(glow,white); +# endif + +#endif + +#ifdef USE_ACES_TONEMAPPER + + color.rgb = tonemap_aces(color.rgb); + +# if defined(USING_GLOW) + glow = tonemap_aces(glow); +# endif + +#endif + + //regular Linear -> SRGB conversion + vec3 a = vec3(0.055); + color.rgb = mix( (vec3(1.0)+a)*pow(color.rgb,vec3(1.0/2.4))-a , 12.92*color.rgb , lessThan(color.rgb,vec3(0.0031308))); + +#if defined(USING_GLOW) + glow = mix( (vec3(1.0)+a)*pow(glow,vec3(1.0/2.4))-a , 12.92*glow , lessThan(glow,vec3(0.0031308))); +#endif + +//glow needs to be added in SRGB space (together with image space effects) + + color.rgb = clamp(color.rgb,0.0,1.0); + +#if defined(USING_GLOW) + glow = clamp(glow,0.0,1.0); +#endif + +#ifdef USE_GLOW_REPLACE + + color.rgb = glow; + +#endif + +#ifdef USE_GLOW_SCREEN + + color.rgb = max((color.rgb + glow) - (color.rgb * glow), vec3(0.0)); + +#endif + +#ifdef USE_GLOW_SOFTLIGHT + + { + + glow = (glow * 0.5) + 0.5; + color.r = (glow.r <= 0.5) ? (color.r - (1.0 - 2.0 * glow.r) * color.r * (1.0 - color.r)) : (((glow.r > 0.5) && (color.r <= 0.25)) ? (color.r + (2.0 * glow.r - 1.0) * (4.0 * color.r * (4.0 * color.r + 1.0) * (color.r - 1.0) + 7.0 * color.r)) : (color.r + (2.0 * glow.r - 1.0) * (sqrt(color.r) - color.r))); + color.g = (glow.g <= 0.5) ? (color.g - (1.0 - 2.0 * glow.g) * color.g * (1.0 - color.g)) : (((glow.g > 0.5) && (color.g <= 0.25)) ? (color.g + (2.0 * glow.g - 1.0) * (4.0 * color.g * (4.0 * color.g + 1.0) * (color.g - 1.0) + 7.0 * color.g)) : (color.g + (2.0 * glow.g - 1.0) * (sqrt(color.g) - color.g))); + color.b = (glow.b <= 0.5) ? (color.b - (1.0 - 2.0 * glow.b) * color.b * (1.0 - color.b)) : (((glow.b > 0.5) && (color.b <= 0.25)) ? (color.b + (2.0 * glow.b - 1.0) * (4.0 * color.b * (4.0 * color.b + 1.0) * (color.b - 1.0) + 7.0 * color.b)) : (color.b + (2.0 * glow.b - 1.0) * (sqrt(color.b) - color.b))); + } + +#endif + +#if defined(USING_GLOW) && !defined(USE_GLOW_SCREEN) && !defined(USE_GLOW_SOFTLIGHT) && !defined(USE_GLOW_REPLACE) + //additive + color.rgb+=glow; +#endif + +#ifdef USE_BCS + + color.rgb = mix(vec3(0.0),color.rgb,bcs.x); + color.rgb = mix(vec3(0.5),color.rgb,bcs.y); + color.rgb = mix(vec3(dot(vec3(1.0),color.rgb)*0.33333),color.rgb,bcs.z); + +#endif + +#ifdef USE_COLOR_CORRECTION + + color.r = texture(color_correction,vec2(color.r,0.0)).r; + color.g = texture(color_correction,vec2(color.g,0.0)).g; + color.b = texture(color_correction,vec2(color.b,0.0)).b; +#endif + + + frag_color=vec4(color.rgb,1.0); +} |