diff options
Diffstat (limited to 'servers/rendering/rasterizer_rd/shaders')
14 files changed, 763 insertions, 364 deletions
diff --git a/servers/rendering/rasterizer_rd/shaders/canvas.glsl b/servers/rendering/rasterizer_rd/shaders/canvas.glsl index e33b3face9..2a0f94e733 100644 --- a/servers/rendering/rasterizer_rd/shaders/canvas.glsl +++ b/servers/rendering/rasterizer_rd/shaders/canvas.glsl @@ -26,7 +26,7 @@ layout(location = 3) out vec2 pixel_size_interp; #endif #ifdef USE_MATERIAL_UNIFORMS -layout(set = 1, binding = 1, std140) uniform MaterialUniforms{ +layout(set = 1, binding = 0, std140) uniform MaterialUniforms{ /* clang-format off */ MATERIAL_UNIFORMS /* clang-format on */ @@ -101,7 +101,7 @@ void main() { offset += 1; } else { instance_color = vec4(texelFetch(instancing_buffer, offset + 0), texelFetch(instancing_buffer, offset + 1), texelFetch(instancing_buffer, offset + 2), texelFetch(instancing_buffer, offset + 3)); - offser += 4; + offset += 4; } color *= instance_color; @@ -144,7 +144,7 @@ VERTEX_SHADER_CODE color_interp = color; - if (bool(draw_data.flags & FLAGS_USE_PIXEL_SNAP)) { + if (canvas_data.use_pixel_snap) { vertex = floor(vertex + 0.5); // precision issue on some hardware creates artifacts within texture // offset uv by a small amount to avoid @@ -226,7 +226,7 @@ layout(location = 3) in vec2 pixel_size_interp; layout(location = 0) out vec4 frag_color; #ifdef USE_MATERIAL_UNIFORMS -layout(set = 1, binding = 1, std140) uniform MaterialUniforms{ +layout(set = 1, binding = 0, std140) uniform MaterialUniforms{ /* clang-format off */ MATERIAL_UNIFORMS /* clang-format on */ @@ -249,7 +249,7 @@ vec4 light_compute( inout vec4 shadow_modulate, vec2 screen_uv, vec2 uv, - vec4 color) { + vec4 color, bool is_directional) { vec4 light = vec4(0.0); /* clang-format off */ LIGHT_SHADER_CODE @@ -302,6 +302,99 @@ float map_ninepatch_axis(float pixel, float draw_size, float tex_pixel_size, flo #endif +#ifdef USE_LIGHTING + +vec3 light_normal_compute(vec3 light_vec, vec3 normal, vec3 base_color, vec3 light_color, vec4 specular_shininess, bool specular_shininess_used) { + float cNdotL = max(0.0, dot(normal, light_vec)); + + if (specular_shininess_used) { + //blinn + vec3 view = vec3(0.0, 0.0, 1.0); // not great but good enough + vec3 half_vec = normalize(view + light_vec); + + float cNdotV = max(dot(normal, view), 0.0); + float cNdotH = max(dot(normal, half_vec), 0.0); + float cVdotH = max(dot(view, half_vec), 0.0); + float cLdotH = max(dot(light_vec, half_vec), 0.0); + float shininess = exp2(15.0 * specular_shininess.a + 1.0) * 0.25; + float blinn = pow(cNdotH, shininess); + blinn *= (shininess + 8.0) * (1.0 / (8.0 * M_PI)); + float s = (blinn) / max(4.0 * cNdotV * cNdotL, 0.75); + + return specular_shininess.rgb * light_color * s + light_color * base_color * cNdotL; + } else { + return light_color * base_color * cNdotL; + } +} + +//float distance = length(shadow_pos); +vec4 light_shadow_compute(uint light_base, vec4 light_color, vec4 shadow_uv +#ifdef LIGHT_SHADER_CODE_USED + , + vec3 shadow_modulate +#endif +) { + float shadow; + uint shadow_mode = light_array.data[light_base].flags & LIGHT_FLAGS_FILTER_MASK; + + if (shadow_mode == LIGHT_FLAGS_SHADOW_NEAREST) { + shadow = textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv, 0.0).x; + } else if (shadow_mode == LIGHT_FLAGS_SHADOW_PCF5) { + vec4 shadow_pixel_size = vec4(light_array.data[light_base].shadow_pixel_size, 0.0, 0.0, 0.0); + shadow = 0.0; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv - shadow_pixel_size * 2.0, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv - shadow_pixel_size, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv + shadow_pixel_size, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv + shadow_pixel_size * 2.0, 0.0).x; + shadow /= 5.0; + } else { //PCF13 + vec4 shadow_pixel_size = vec4(light_array.data[light_base].shadow_pixel_size, 0.0, 0.0, 0.0); + shadow = 0.0; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv - shadow_pixel_size * 6.0, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv - shadow_pixel_size * 5.0, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv - shadow_pixel_size * 4.0, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv - shadow_pixel_size * 3.0, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv - shadow_pixel_size * 2.0, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv - shadow_pixel_size, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv + shadow_pixel_size, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv + shadow_pixel_size * 2.0, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv + shadow_pixel_size * 3.0, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv + shadow_pixel_size * 4.0, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv + shadow_pixel_size * 5.0, 0.0).x; + shadow += textureProjLod(sampler2DShadow(shadow_atlas_texture, shadow_sampler), shadow_uv + shadow_pixel_size * 6.0, 0.0).x; + shadow /= 13.0; + } + + vec4 shadow_color = unpackUnorm4x8(light_array.data[light_base].shadow_color); +#ifdef LIGHT_SHADER_CODE_USED + shadow_color *= shadow_modulate; +#endif + + shadow_color.a *= light_color.a; //respect light alpha + + return mix(light_color, shadow_color, shadow); +} + +void light_blend_compute(uint light_base, vec4 light_color, inout vec3 color) { + uint blend_mode = light_array.data[light_base].flags & LIGHT_FLAGS_BLEND_MASK; + + switch (blend_mode) { + case LIGHT_FLAGS_BLEND_MODE_ADD: { + color.rgb += light_color.rgb * light_color.a; + } break; + case LIGHT_FLAGS_BLEND_MODE_SUB: { + color.rgb -= light_color.rgb * light_color.a; + } break; + case LIGHT_FLAGS_BLEND_MODE_MIX: { + color.rgb = mix(color.rgb, light_color.rgb, light_color.a); + } break; + } +} + +#endif + void main() { vec4 color = color_interp; vec2 uv = uv_interp; @@ -332,6 +425,7 @@ void main() { color *= texture(sampler2D(color_texture, texture_sampler), uv); uint light_count = (draw_data.flags >> FLAGS_LIGHT_COUNT_SHIFT) & 0xF; //max 16 lights + bool using_light = light_count > 0 || canvas_data.directional_light_count > 0; vec3 normal; @@ -341,7 +435,7 @@ void main() { bool normal_used = false; #endif - if (normal_used || (light_count > 0 && bool(draw_data.flags & FLAGS_DEFAULT_NORMAL_MAP_USED))) { + if (normal_used || (using_light && bool(draw_data.flags & FLAGS_DEFAULT_NORMAL_MAP_USED))) { normal.xy = texture(sampler2D(normal_texture, texture_sampler), uv).xy * vec2(2.0, -2.0) - vec2(1.0, -1.0); normal.z = sqrt(1.0 - dot(normal.xy, normal.xy)); normal_used = true; @@ -358,7 +452,7 @@ void main() { bool specular_shininess_used = false; #endif - if (specular_shininess_used || (light_count > 0 && normal_used && bool(draw_data.flags & FLAGS_DEFAULT_SPECULAR_MAP_USED))) { + if (specular_shininess_used || (using_light && normal_used && bool(draw_data.flags & FLAGS_DEFAULT_SPECULAR_MAP_USED))) { specular_shininess = texture(sampler2D(specular_texture, texture_sampler), uv); specular_shininess *= unpackUnorm4x8(draw_data.specular_shininess); specular_shininess_used = true; @@ -401,14 +495,53 @@ FRAGMENT_SHADER_CODE normal = normalize((canvas_data.canvas_normal_transform * vec4(normal, 0.0)).xyz); } - vec4 base_color = color; + vec3 base_color = color.rgb; if (bool(draw_data.flags & FLAGS_USING_LIGHT_MASK)) { color = vec4(0.0); //invisible by default due to using light mask } color *= canvas_data.canvas_modulation; #ifdef USE_LIGHTING - for (uint i = 0; i < MAX_LIGHT_TEXTURES; i++) { + + // Directional Lights + + for (uint i = 0; i < canvas_data.directional_light_count; i++) { + uint light_base = i; + + vec2 direction = light_array.data[light_base].position; + vec4 light_color = light_array.data[light_base].color; + +#ifdef LIGHT_SHADER_CODE_USED + + vec4 shadow_modulate = vec4(1.0); + light_color = light_compute(light_vertex, direction, normal, light_color, light_color.a, specular_shininess, shadow_modulate, screen_uv, color, uv, true); +#else + + if (normal_used) { + vec3 light_vec = normalize(mix(vec3(direction, 0.0), vec3(0, 0, 1), light_array.data[light_base].height)); + light_color.rgb = light_normal_compute(light_vec, normal, base_color, light_color.rgb, specular_shininess, specular_shininess_used); + } +#endif + + if (bool(light_array.data[light_base].flags & LIGHT_FLAGS_HAS_SHADOW)) { + vec2 shadow_pos = (vec4(shadow_vertex, 0.0, 1.0) * mat4(light_array.data[light_base].shadow_matrix[0], light_array.data[light_base].shadow_matrix[1], vec4(0.0, 0.0, 1.0, 0.0), vec4(0.0, 0.0, 0.0, 1.0))).xy; //multiply inverse given its transposed. Optimizer removes useless operations. + + vec4 shadow_uv = vec4(shadow_pos.x, light_array.data[light_base].shadow_y_ofs, shadow_pos.y * light_array.data[light_base].shadow_zfar_inv, 1.0); + + light_color = light_shadow_compute(light_base, light_color, shadow_uv +#ifdef LIGHT_SHADER_CODE_USED + , + shadow_modulate +#endif + ); + } + + light_blend_compute(light_base, light_color, color.rgb); + } + + // Positional Lights + + for (uint i = 0; i < MAX_LIGHTS_PER_ITEM; i++) { if (i >= light_count) { break; } @@ -430,7 +563,8 @@ FRAGMENT_SHADER_CODE light_base &= 0xFF; vec2 tex_uv = (vec4(vertex, 0.0, 1.0) * mat4(light_array.data[light_base].texture_matrix[0], light_array.data[light_base].texture_matrix[1], vec4(0.0, 0.0, 1.0, 0.0), vec4(0.0, 0.0, 0.0, 1.0))).xy; //multiply inverse given its transposed. Optimizer removes useless operations. - vec4 light_color = texture(sampler2D(light_textures[i], texture_sampler), tex_uv); + vec2 tex_uv_atlas = tex_uv * light_array.data[light_base].atlas_rect.zw + light_array.data[light_base].atlas_rect.xy; + vec4 light_color = textureLod(sampler2D(atlas_texture, texture_sampler), tex_uv_atlas, 0.0); vec4 light_base_color = light_array.data[light_base].color; #ifdef LIGHT_SHADER_CODE_USED @@ -439,7 +573,7 @@ FRAGMENT_SHADER_CODE vec3 light_position = vec3(light_array.data[light_base].position, light_array.data[light_base].height); light_color.rgb *= light_base_color.rgb; - light_color = light_compute(light_vertex, light_position, normal, light_color, light_base_color.a, specular_shininess, shadow_modulate, screen_uv, color, uv); + light_color = light_compute(light_vertex, light_position, normal, light_color, light_base_color.a, specular_shininess, shadow_modulate, screen_uv, color, uv, false); #else light_color.rgb *= light_base_color.rgb * light_base_color.a; @@ -450,24 +584,7 @@ FRAGMENT_SHADER_CODE vec3 light_vec = normalize(light_pos - pos); float cNdotL = max(0.0, dot(normal, light_vec)); - if (specular_shininess_used) { - //blinn - vec3 view = vec3(0.0, 0.0, 1.0); // not great but good enough - vec3 half_vec = normalize(view + light_vec); - - float cNdotV = max(dot(normal, view), 0.0); - float cNdotH = max(dot(normal, half_vec), 0.0); - float cVdotH = max(dot(view, half_vec), 0.0); - float cLdotH = max(dot(light_vec, half_vec), 0.0); - float shininess = exp2(15.0 * specular_shininess.a + 1.0) * 0.25; - float blinn = pow(cNdotH, shininess); - blinn *= (shininess + 8.0) * (1.0 / (8.0 * M_PI)); - float s = (blinn) / max(4.0 * cNdotV * cNdotL, 0.75); - - light_color.rgb = specular_shininess.rgb * light_base_color.rgb * s + light_color.rgb * cNdotL; - } else { - light_color.rgb *= cNdotL; - } + light_color.rgb = light_normal_compute(light_vec, normal, base_color, light_color.rgb, specular_shininess, specular_shininess_used); } #endif if (any(lessThan(tex_uv, vec2(0.0, 0.0))) || any(greaterThanEqual(tex_uv, vec2(1.0, 1.0)))) { @@ -502,66 +619,20 @@ FRAGMENT_SHADER_CODE } } + distance *= light_array.data[light_base].shadow_zfar_inv; + //float distance = length(shadow_pos); - float shadow; - uint shadow_mode = light_array.data[light_base].flags & LIGHT_FLAGS_FILTER_MASK; - - vec4 shadow_uv = vec4(tex_ofs, 0.0, distance, 1.0); - - if (shadow_mode == LIGHT_FLAGS_SHADOW_NEAREST) { - shadow = textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv).x; - } else if (shadow_mode == LIGHT_FLAGS_SHADOW_PCF5) { - vec4 shadow_pixel_size = vec4(light_array.data[light_base].shadow_pixel_size, 0.0, 0.0, 0.0); - shadow = 0.0; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv - shadow_pixel_size * 2.0).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv - shadow_pixel_size).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv + shadow_pixel_size).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv + shadow_pixel_size * 2.0).x; - shadow /= 5.0; - } else { //PCF13 - vec4 shadow_pixel_size = vec4(light_array.data[light_base].shadow_pixel_size, 0.0, 0.0, 0.0); - shadow = 0.0; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv - shadow_pixel_size * 6.0).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv - shadow_pixel_size * 5.0).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv - shadow_pixel_size * 4.0).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv - shadow_pixel_size * 3.0).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv - shadow_pixel_size * 2.0).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv - shadow_pixel_size).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv + shadow_pixel_size).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv + shadow_pixel_size * 2.0).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv + shadow_pixel_size * 3.0).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv + shadow_pixel_size * 4.0).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv + shadow_pixel_size * 5.0).x; - shadow += textureProj(sampler2DShadow(shadow_textures[i], shadow_sampler), shadow_uv + shadow_pixel_size * 6.0).x; - shadow /= 13.0; - } + vec4 shadow_uv = vec4(tex_ofs, light_array.data[light_base].shadow_y_ofs, distance, 1.0); - vec4 shadow_color = light_array.data[light_base].shadow_color; + light_color = light_shadow_compute(light_base, light_color, shadow_uv #ifdef LIGHT_SHADER_CODE_USED - shadow_color *= shadow_modulate; + , + shadow_modulate #endif - light_color = mix(light_color, shadow_color, shadow); + ); } - uint blend_mode = light_array.data[light_base].flags & LIGHT_FLAGS_BLEND_MASK; - - switch (blend_mode) { - case LIGHT_FLAGS_BLEND_MODE_ADD: { - color.rgb += light_color.rgb * light_color.a; - } break; - case LIGHT_FLAGS_BLEND_MODE_SUB: { - color.rgb -= light_color.rgb * light_color.a; - } break; - case LIGHT_FLAGS_BLEND_MODE_MIX: { - color.rgb = mix(color.rgb, light_color.rgb, light_color.a); - } break; - case LIGHT_FLAGS_BLEND_MODE_MASK: { - light_color.a *= base_color.a; - color.rgb = mix(color.rgb, light_color.rgb, light_color.a); - } break; - } + light_blend_compute(light_base, light_color, color.rgb); } #endif diff --git a/servers/rendering/rasterizer_rd/shaders/canvas_occlusion.glsl b/servers/rendering/rasterizer_rd/shaders/canvas_occlusion.glsl index 99e70a1976..421282cd4d 100644 --- a/servers/rendering/rasterizer_rd/shaders/canvas_occlusion.glsl +++ b/servers/rendering/rasterizer_rd/shaders/canvas_occlusion.glsl @@ -8,7 +8,8 @@ layout(push_constant, binding = 0, std430) uniform Constants { mat4 projection; mat2x4 modelview; vec2 direction; - vec2 pad; + float z_far; + float pad; } constants; @@ -25,9 +26,18 @@ void main() { #version 450 +layout(push_constant, binding = 0, std430) uniform Constants { + mat4 projection; + mat2x4 modelview; + vec2 direction; + float z_far; + float pad; +} +constants; + layout(location = 0) in highp float depth; layout(location = 0) out highp float distance_buf; void main() { - distance_buf = depth; + distance_buf = depth / constants.z_far; } diff --git a/servers/rendering/rasterizer_rd/shaders/canvas_uniforms_inc.glsl b/servers/rendering/rasterizer_rd/shaders/canvas_uniforms_inc.glsl index a39866004b..bb39584cbb 100644 --- a/servers/rendering/rasterizer_rd/shaders/canvas_uniforms_inc.glsl +++ b/servers/rendering/rasterizer_rd/shaders/canvas_uniforms_inc.glsl @@ -1,3 +1,6 @@ + +#define MAX_LIGHTS_PER_ITEM 16 + #define M_PI 3.14159265359 #define FLAGS_INSTANCING_STRIDE_MASK 0xF @@ -12,7 +15,6 @@ #define FLAGS_USING_LIGHT_MASK (1 << 11) #define FLAGS_NINEPACH_DRAW_CENTER (1 << 12) #define FLAGS_USING_PARTICLES (1 << 13) -#define FLAGS_USE_PIXEL_SNAP (1 << 14) #define FLAGS_NINEPATCH_H_MODE_SHIFT 16 #define FLAGS_NINEPATCH_V_MODE_SHIFT 18 @@ -22,13 +24,7 @@ #define FLAGS_DEFAULT_NORMAL_MAP_USED (1 << 26) #define FLAGS_DEFAULT_SPECULAR_MAP_USED (1 << 27) -// In vulkan, sets should always be ordered using the following logic: -// Lower Sets: Sets that change format and layout less often -// Higher sets: Sets that change format and layout very often -// This is because changing a set for another with a different layout or format, -// invalidates all the upper ones. - -/* SET0: Draw Primitive */ +// Push Constant layout(push_constant, binding = 0, std430) uniform DrawData { vec2 world_x; @@ -53,46 +49,31 @@ layout(push_constant, binding = 0, std430) uniform DrawData { } draw_data; -// The values passed per draw primitives are cached within it - -layout(set = 0, binding = 1) uniform texture2D color_texture; -layout(set = 0, binding = 2) uniform texture2D normal_texture; -layout(set = 0, binding = 3) uniform texture2D specular_texture; -layout(set = 0, binding = 4) uniform sampler texture_sampler; - -layout(set = 0, binding = 5) uniform textureBuffer instancing_buffer; - -/* SET1: Is reserved for the material */ - -#ifdef USE_MATERIAL_SAMPLERS - -layout(set = 1, binding = 0) uniform sampler material_samplers[12]; +// In vulkan, sets should always be ordered using the following logic: +// Lower Sets: Sets that change format and layout less often +// Higher sets: Sets that change format and layout very often +// This is because changing a set for another with a different layout or format, +// invalidates all the upper ones (as likely internal base offset changes) -#endif +/* SET0: Globals */ -/* SET2: Canvas Item State (including lighting) */ +// The values passed per draw primitives are cached within it -layout(set = 2, binding = 0, std140) uniform CanvasData { +layout(set = 0, binding = 1, std140) uniform CanvasData { mat4 canvas_transform; mat4 screen_transform; mat4 canvas_normal_transform; vec4 canvas_modulation; vec2 screen_pixel_size; float time; - float time_pad; - //uint light_count; -} -canvas_data; - -layout(set = 2, binding = 1) uniform textureBuffer skeleton_buffer; + bool use_pixel_snap; -layout(set = 2, binding = 2, std140) uniform SkeletonData { - mat4 skeleton_transform; //in world coordinates - mat4 skeleton_transform_inverse; + uint directional_light_count; + uint pad0; + uint pad1; + uint pad2; } -skeleton_data; - -#ifdef USE_LIGHTING +canvas_data; #define LIGHT_FLAGS_BLEND_MASK (3 << 16) #define LIGHT_FLAGS_BLEND_MODE_ADD (0 << 16) @@ -110,37 +91,52 @@ struct Light { mat2x4 texture_matrix; //light to texture coordinate matrix (transposed) mat2x4 shadow_matrix; //light to shadow coordinate matrix (transposed) vec4 color; - vec4 shadow_color; - vec2 position; + + uint shadow_color; // packed uint flags; //index to light texture - float height; float shadow_pixel_size; - float pad0; - float pad1; - float pad2; + float height; + + vec2 position; + float shadow_zfar_inv; + float shadow_y_ofs; + + vec4 atlas_rect; }; -layout(set = 2, binding = 3, std140) uniform LightData { +layout(set = 0, binding = 2, std140) uniform LightData { Light data[MAX_LIGHTS]; } light_array; -layout(set = 2, binding = 4) uniform texture2D light_textures[MAX_LIGHT_TEXTURES]; -layout(set = 2, binding = 5) uniform texture2D shadow_textures[MAX_LIGHT_TEXTURES]; +layout(set = 0, binding = 3) uniform texture2D atlas_texture; +layout(set = 0, binding = 4) uniform texture2D shadow_atlas_texture; -layout(set = 2, binding = 6) uniform sampler shadow_sampler; +layout(set = 0, binding = 5) uniform sampler shadow_sampler; -#endif +layout(set = 0, binding = 6) uniform texture2D screen_texture; -layout(set = 2, binding = 7, std430) restrict readonly buffer GlobalVariableData { +layout(set = 0, binding = 7) uniform sampler material_samplers[12]; + +layout(set = 0, binding = 8, std430) restrict readonly buffer GlobalVariableData { vec4 data[]; } global_variables; -/* SET3: Render Target Data */ +/* SET1: Is reserved for the material */ + +// -#ifdef SCREEN_TEXTURE_USED +/* SET2: Instancing and Skeleton */ -layout(set = 3, binding = 0) uniform texture2D screen_texture; +layout(set = 2, binding = 0, std430) restrict readonly buffer Transforms { + vec4 data[]; +} +transforms; -#endif +/* SET3: Texture */ + +layout(set = 3, binding = 0) uniform texture2D color_texture; +layout(set = 3, binding = 1) uniform texture2D normal_texture; +layout(set = 3, binding = 2) uniform texture2D specular_texture; +layout(set = 3, binding = 3) uniform sampler texture_sampler; diff --git a/servers/rendering/rasterizer_rd/shaders/copy.glsl b/servers/rendering/rasterizer_rd/shaders/copy.glsl index e565bd8e3d..cdd35dfb3f 100644 --- a/servers/rendering/rasterizer_rd/shaders/copy.glsl +++ b/servers/rendering/rasterizer_rd/shaders/copy.glsl @@ -15,6 +15,7 @@ layout(local_size_x = 8, local_size_y = 8, local_size_z = 1) in; #define FLAG_FORCE_LUMINANCE (1 << 6) #define FLAG_COPY_ALL_SOURCE (1 << 7) #define FLAG_HIGH_QUALITY_GLOW (1 << 8) +#define FLAG_ALPHA_TO_ONE (1 << 9) layout(push_constant, binding = 1, std430) uniform Params { ivec4 section; @@ -35,6 +36,8 @@ layout(push_constant, binding = 1, std430) uniform Params { float camera_z_far; float camera_z_near; uint pad2[2]; + + vec4 set_color; } params; @@ -42,7 +45,7 @@ params; layout(set = 0, binding = 0) uniform samplerCubeArray source_color; #elif defined(MODE_CUBEMAP_TO_PANORAMA) layout(set = 0, binding = 0) uniform samplerCube source_color; -#else +#elif !defined(MODE_SET_COLOR) layout(set = 0, binding = 0) uniform sampler2D source_color; #endif @@ -58,12 +61,20 @@ layout(rgba8, set = 3, binding = 0) uniform restrict writeonly image2D dest_buff layout(rgba32f, set = 3, binding = 0) uniform restrict writeonly image2D dest_buffer; #endif +#ifdef MODE_GAUSSIAN_GLOW +shared vec4 local_cache[256]; +shared vec4 temp_cache[128]; +#endif + void main() { // Pixel being shaded ivec2 pos = ivec2(gl_GlobalInvocationID.xy); + +#ifndef MODE_GAUSSIAN_GLOW // Glow needs the extra threads if (any(greaterThanEqual(pos, params.section.zw))) { //too large, do nothing return; } +#endif #ifdef MODE_MIPMAP @@ -104,70 +115,69 @@ void main() { #ifdef MODE_GAUSSIAN_GLOW - //Glow uses larger sigma 1 for a more rounded blur effect + // First pass copy texture into 16x16 local memory for every 8x8 thread block + vec2 quad_center_uv = clamp(vec2(gl_GlobalInvocationID.xy + gl_LocalInvocationID.xy - 3.5) / params.section.zw, vec2(0.5 / params.section.zw), vec2(1.0 - 1.5 / params.section.zw)); + uint dest_index = gl_LocalInvocationID.x * 2 + gl_LocalInvocationID.y * 2 * 16; -#define GLOW_ADD(m_ofs, m_mult) \ - { \ - ivec2 ofs = base_pos + m_ofs; \ - if (all(greaterThanEqual(ofs, section_begin)) && all(lessThan(ofs, section_end))) { \ - color += texelFetch(source_color, ofs, 0) * m_mult; \ - } \ + if (bool(params.flags & FLAG_HIGH_QUALITY_GLOW)) { + vec2 quad_offset_uv = clamp((vec2(gl_GlobalInvocationID.xy + gl_LocalInvocationID.xy - 3.0)) / params.section.zw, vec2(0.5 / params.section.zw), vec2(1.0 - 1.5 / params.section.zw)); + + local_cache[dest_index] = (textureLod(source_color, quad_center_uv, 0) + textureLod(source_color, quad_offset_uv, 0)) * 0.5; + local_cache[dest_index + 1] = (textureLod(source_color, quad_center_uv + vec2(1.0 / params.section.z, 0.0), 0) + textureLod(source_color, quad_offset_uv + vec2(1.0 / params.section.z, 0.0), 0)) * 0.5; + local_cache[dest_index + 16] = (textureLod(source_color, quad_center_uv + vec2(0.0, 1.0 / params.section.w), 0) + textureLod(source_color, quad_offset_uv + vec2(0.0, 1.0 / params.section.w), 0)) * 0.5; + local_cache[dest_index + 16 + 1] = (textureLod(source_color, quad_center_uv + vec2(1.0 / params.section.zw), 0) + textureLod(source_color, quad_offset_uv + vec2(1.0 / params.section.zw), 0)) * 0.5; + } else { + local_cache[dest_index] = textureLod(source_color, quad_center_uv, 0); + local_cache[dest_index + 1] = textureLod(source_color, quad_center_uv + vec2(1.0 / params.section.z, 0.0), 0); + local_cache[dest_index + 16] = textureLod(source_color, quad_center_uv + vec2(0.0, 1.0 / params.section.w), 0); + local_cache[dest_index + 16 + 1] = textureLod(source_color, quad_center_uv + vec2(1.0 / params.section.zw), 0); } + memoryBarrierShared(); + barrier(); + + // Horizontal pass. Needs to copy into 8x16 chunk of local memory so vertical pass has full resolution + uint read_index = gl_LocalInvocationID.x + gl_LocalInvocationID.y * 32 + 4; + vec4 color_top = vec4(0.0); + color_top += local_cache[read_index] * 0.174938; + color_top += local_cache[read_index + 1] * 0.165569; + color_top += local_cache[read_index + 2] * 0.140367; + color_top += local_cache[read_index + 3] * 0.106595; + color_top += local_cache[read_index - 1] * 0.165569; + color_top += local_cache[read_index - 2] * 0.140367; + color_top += local_cache[read_index - 3] * 0.106595; + + vec4 color_bottom = vec4(0.0); + color_bottom += local_cache[read_index + 16] * 0.174938; + color_bottom += local_cache[read_index + 1 + 16] * 0.165569; + color_bottom += local_cache[read_index + 2 + 16] * 0.140367; + color_bottom += local_cache[read_index + 3 + 16] * 0.106595; + color_bottom += local_cache[read_index - 1 + 16] * 0.165569; + color_bottom += local_cache[read_index - 2 + 16] * 0.140367; + color_bottom += local_cache[read_index - 3 + 16] * 0.106595; + + // rotate samples to take advantage of cache coherency + uint write_index = gl_LocalInvocationID.y * 2 + gl_LocalInvocationID.x * 16; + + temp_cache[write_index] = color_top; + temp_cache[write_index + 1] = color_bottom; + + memoryBarrierShared(); + barrier(); + + // Vertical pass + uint index = gl_LocalInvocationID.y + gl_LocalInvocationID.x * 16 + 4; vec4 color = vec4(0.0); - if (bool(params.flags & FLAG_HORIZONTAL)) { - ivec2 base_pos = ((pos + params.section.xy) << 1) + ivec2(1); - ivec2 section_begin = params.section.xy << 1; - ivec2 section_end = section_begin + (params.section.zw << 1); - - if (bool(params.flags & FLAG_HIGH_QUALITY_GLOW)) { - //Sample from two lines to capture single pixel features - GLOW_ADD(ivec2(0, 0), 0.152781); - GLOW_ADD(ivec2(1, 0), 0.144599); - GLOW_ADD(ivec2(2, 0), 0.122589); - GLOW_ADD(ivec2(3, 0), 0.093095); - GLOW_ADD(ivec2(4, 0), 0.063327); - GLOW_ADD(ivec2(-1, 0), 0.144599); - GLOW_ADD(ivec2(-2, 0), 0.122589); - GLOW_ADD(ivec2(-3, 0), 0.093095); - GLOW_ADD(ivec2(-4, 0), 0.063327); - - GLOW_ADD(ivec2(0, 1), 0.152781); - GLOW_ADD(ivec2(1, 1), 0.144599); - GLOW_ADD(ivec2(2, 1), 0.122589); - GLOW_ADD(ivec2(3, 1), 0.093095); - GLOW_ADD(ivec2(4, 1), 0.063327); - GLOW_ADD(ivec2(-1, 1), 0.144599); - GLOW_ADD(ivec2(-2, 1), 0.122589); - GLOW_ADD(ivec2(-3, 1), 0.093095); - GLOW_ADD(ivec2(-4, 1), 0.063327); - color *= 0.5; - } else { - GLOW_ADD(ivec2(0, 0), 0.174938); - GLOW_ADD(ivec2(1, 0), 0.165569); - GLOW_ADD(ivec2(2, 0), 0.140367); - GLOW_ADD(ivec2(3, 0), 0.106595); - GLOW_ADD(ivec2(-1, 0), 0.165569); - GLOW_ADD(ivec2(-2, 0), 0.140367); - GLOW_ADD(ivec2(-3, 0), 0.106595); - } - - color *= params.glow_strength; - } else { - ivec2 base_pos = pos + params.section.xy; - ivec2 section_begin = params.section.xy; - ivec2 section_end = section_begin + params.section.zw; - - GLOW_ADD(ivec2(0, 0), 0.288713); - GLOW_ADD(ivec2(0, 1), 0.233062); - GLOW_ADD(ivec2(0, 2), 0.122581); - GLOW_ADD(ivec2(0, -1), 0.233062); - GLOW_ADD(ivec2(0, -2), 0.122581); - color *= params.glow_strength; - } + color += temp_cache[index] * 0.174938; + color += temp_cache[index + 1] * 0.165569; + color += temp_cache[index + 2] * 0.140367; + color += temp_cache[index + 3] * 0.106595; + color += temp_cache[index - 1] * 0.165569; + color += temp_cache[index - 2] * 0.140367; + color += temp_cache[index - 3] * 0.106595; -#undef GLOW_ADD + color *= params.glow_strength; if (bool(params.flags & FLAG_GLOW_FIRST_PASS)) { #ifdef GLOW_USE_AUTO_EXPOSURE @@ -196,25 +206,24 @@ void main() { } color = textureLod(source_color, uv, 0.0); - if (bool(params.flags & FLAG_FORCE_LUMINANCE)) { - color.rgb = vec3(max(max(color.r, color.g), color.b)); - } - imageStore(dest_buffer, pos + params.target, color); - } else { color = texelFetch(source_color, pos + params.section.xy, 0); - if (bool(params.flags & FLAG_FORCE_LUMINANCE)) { - color.rgb = vec3(max(max(color.r, color.g), color.b)); - } - if (bool(params.flags & FLAG_FLIP_Y)) { pos.y = params.section.w - pos.y - 1; } + } - imageStore(dest_buffer, pos + params.target, color); + if (bool(params.flags & FLAG_FORCE_LUMINANCE)) { + color.rgb = vec3(max(max(color.r, color.g), color.b)); } + if (bool(params.flags & FLAG_ALPHA_TO_ONE)) { + color.a = 1.0; + } + + imageStore(dest_buffer, pos + params.target, color); + #endif #ifdef MODE_SIMPLE_COPY_DEPTH @@ -263,4 +272,8 @@ void main() { #endif imageStore(dest_buffer, pos + params.target, color); #endif + +#ifdef MODE_SET_COLOR + imageStore(dest_buffer, pos + params.target, params.set_color); +#endif } diff --git a/servers/rendering/rasterizer_rd/shaders/particles.glsl b/servers/rendering/rasterizer_rd/shaders/particles.glsl index 3de807b57c..926c7ef9fc 100644 --- a/servers/rendering/rasterizer_rd/shaders/particles.glsl +++ b/servers/rendering/rasterizer_rd/shaders/particles.glsl @@ -31,6 +31,40 @@ global_variables; /* Set 1: FRAME AND PARTICLE DATA */ // a frame history is kept for trail deterministic behavior + +#define MAX_ATTRACTORS 32 + +#define ATTRACTOR_TYPE_SPHERE 0 +#define ATTRACTOR_TYPE_BOX 1 +#define ATTRACTOR_TYPE_VECTOR_FIELD 2 + +struct Attractor { + mat4 transform; + vec3 extents; //exents or radius + uint type; + uint texture_index; //texture index for vector field + float strength; + float attenuation; + float directionality; +}; + +#define MAX_COLLIDERS 32 + +#define COLLIDER_TYPE_SPHERE 0 +#define COLLIDER_TYPE_BOX 1 +#define COLLIDER_TYPE_SDF 2 +#define COLLIDER_TYPE_HEIGHT_FIELD 3 + +struct Collider { + mat4 transform; + vec3 extents; //exents or radius + uint type; + + uint texture_index; //texture index for vector field + float scale; + uint pad[2]; +}; + struct FrameParams { bool emitting; float system_phase; @@ -43,9 +77,14 @@ struct FrameParams { float delta; uint random_seed; - uint pad[3]; + uint attractor_count; + uint collider_count; + float particle_size; mat4 emission_transform; + + Attractor attractors[MAX_ATTRACTORS]; + Collider colliders[MAX_COLLIDERS]; }; layout(set = 1, binding = 0, std430) restrict buffer FrameHistory { @@ -80,7 +119,7 @@ struct ParticleEmission { vec4 custom; }; -layout(set = 1, binding = 2, std430) restrict volatile coherent buffer SourceEmission { +layout(set = 1, binding = 2, std430) restrict buffer SourceEmission { int particle_count; uint pad0; uint pad1; @@ -89,7 +128,7 @@ layout(set = 1, binding = 2, std430) restrict volatile coherent buffer SourceEmi } src_particles; -layout(set = 1, binding = 3, std430) restrict volatile coherent buffer DestEmission { +layout(set = 1, binding = 3, std430) restrict buffer DestEmission { int particle_count; int particle_max; uint pad1; @@ -98,10 +137,17 @@ layout(set = 1, binding = 3, std430) restrict volatile coherent buffer DestEmiss } dst_particles; -/* SET 2: MATERIAL */ +/* SET 2: COLLIDER/ATTRACTOR TEXTURES */ + +#define MAX_3D_TEXTURES 7 + +layout(set = 2, binding = 0) uniform texture3D sdf_vec_textures[MAX_3D_TEXTURES]; +layout(set = 2, binding = 1) uniform texture2D height_field_texture; + +/* SET 3: MATERIAL */ #ifdef USE_MATERIAL_UNIFORMS -layout(set = 2, binding = 0, std140) uniform MaterialUniforms{ +layout(set = 3, binding = 0, std140) uniform MaterialUniforms{ /* clang-format off */ MATERIAL_UNIFORMS /* clang-format on */ @@ -140,29 +186,7 @@ bool emit_particle(mat4 p_xform, vec3 p_velocity, vec4 p_color, vec4 p_custom, u atomicAdd(dst_particles.particle_count, -1); return false; } - /* - valid = true; - - int attempts = 256; // never trust compute - while(attempts-- > 0) { - dst_index = dst_particles.particle_count; - if (dst_index == dst_particles.particle_max) { - return false; //cant emit anymore - } - - if (atomicCompSwap(dst_particles.particle_count, dst_index, dst_index +1 ) != dst_index) { - continue; - } - valid=true; - break; - } - barrier(); - - if (!valid) { - return false; //gave up (attempts exhausted) - } -*/ dst_particles.data[dst_index].xform = p_xform; dst_particles.data[dst_index].velocity = p_velocity; dst_particles.data[dst_index].color = p_color; @@ -217,6 +241,199 @@ void main() { vec4(0.0, 0.0, 0.0, 1.0)); } + bool collided = false; + vec3 collision_normal = vec3(0.0); + float collision_depth = 0.0; + + vec3 attractor_force = vec3(0.0); + +#if !defined(DISABLE_VELOCITY) + + if (PARTICLE.is_active) { + PARTICLE.xform[3].xyz += PARTICLE.velocity * local_delta; + } +#endif + + /* Process physics if active */ + + if (PARTICLE.is_active) { + for (uint i = 0; i < FRAME.attractor_count; i++) { + vec3 dir; + float amount; + vec3 rel_vec = PARTICLE.xform[3].xyz - FRAME.attractors[i].transform[3].xyz; + vec3 local_pos = rel_vec * mat3(FRAME.attractors[i].transform); + + switch (FRAME.attractors[i].type) { + case ATTRACTOR_TYPE_SPHERE: { + dir = normalize(rel_vec); + float d = length(local_pos) / FRAME.attractors[i].extents.x; + if (d > 1.0) { + continue; + } + amount = max(0.0, 1.0 - d); + } break; + case ATTRACTOR_TYPE_BOX: { + dir = normalize(rel_vec); + + vec3 abs_pos = abs(local_pos / FRAME.attractors[i].extents); + float d = max(abs_pos.x, max(abs_pos.y, abs_pos.z)); + if (d > 1.0) { + continue; + } + amount = max(0.0, 1.0 - d); + + } break; + case ATTRACTOR_TYPE_VECTOR_FIELD: { + vec3 uvw_pos = (local_pos / FRAME.attractors[i].extents) * 2.0 - 1.0; + if (any(lessThan(uvw_pos, vec3(0.0))) || any(greaterThan(uvw_pos, vec3(1.0)))) { + continue; + } + vec3 s = texture(sampler3D(sdf_vec_textures[FRAME.attractors[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos).xyz; + dir = mat3(FRAME.attractors[i].transform) * normalize(s); //revert direction + amount = length(s); + + } break; + } + amount = pow(amount, FRAME.attractors[i].attenuation); + dir = normalize(mix(dir, FRAME.attractors[i].transform[2].xyz, FRAME.attractors[i].directionality)); + attractor_force -= amount * dir * FRAME.attractors[i].strength; + } + + float particle_size = FRAME.particle_size; + +#ifdef USE_COLLISON_SCALE + + particle_size *= dot(vec3(length(PARTICLE.xform[0].xyz), length(PARTICLE.xform[1].xyz), length(PARTICLE.xform[2].xyz)), vec3(0.33333333333)); + +#endif + + for (uint i = 0; i < FRAME.collider_count; i++) { + vec3 normal; + float depth; + bool col = false; + + vec3 rel_vec = PARTICLE.xform[3].xyz - FRAME.colliders[i].transform[3].xyz; + vec3 local_pos = rel_vec * mat3(FRAME.colliders[i].transform); + + switch (FRAME.colliders[i].type) { + case COLLIDER_TYPE_SPHERE: { + float d = length(rel_vec) - (particle_size + FRAME.colliders[i].extents.x); + + if (d < 0.0) { + col = true; + depth = -d; + normal = normalize(rel_vec); + } + + } break; + case COLLIDER_TYPE_BOX: { + vec3 abs_pos = abs(local_pos); + vec3 sgn_pos = sign(local_pos); + + if (any(greaterThan(abs_pos, FRAME.colliders[i].extents))) { + //point outside box + + vec3 closest = min(abs_pos, FRAME.colliders[i].extents); + vec3 rel = abs_pos - closest; + depth = length(rel) - particle_size; + if (depth < 0.0) { + col = true; + normal = mat3(FRAME.colliders[i].transform) * (normalize(rel) * sgn_pos); + depth = -depth; + } + } else { + //point inside box + vec3 axis_len = FRAME.colliders[i].extents - abs_pos; + // there has to be a faster way to do this? + if (all(lessThan(axis_len.xx, axis_len.yz))) { + normal = vec3(1, 0, 0); + } else if (all(lessThan(axis_len.yy, axis_len.xz))) { + normal = vec3(0, 1, 0); + } else { + normal = vec3(0, 0, 1); + } + + col = true; + depth = dot(normal * axis_len, vec3(1)) + particle_size; + normal = mat3(FRAME.colliders[i].transform) * (normal * sgn_pos); + } + + } break; + case COLLIDER_TYPE_SDF: { + vec3 apos = abs(local_pos); + float extra_dist = 0.0; + if (any(greaterThan(apos, FRAME.colliders[i].extents))) { //outside + vec3 mpos = min(apos, FRAME.colliders[i].extents); + extra_dist = distance(mpos, apos); + } + + if (extra_dist > particle_size) { + continue; + } + + vec3 uvw_pos = (local_pos / FRAME.colliders[i].extents) * 0.5 + 0.5; + float s = texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos).r; + s *= FRAME.colliders[i].scale; + s += extra_dist; + if (s < particle_size) { + col = true; + depth = particle_size - s; + const float EPSILON = 0.001; + normal = mat3(FRAME.colliders[i].transform) * + normalize( + vec3( + texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos + vec3(EPSILON, 0.0, 0.0)).r - texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos - vec3(EPSILON, 0.0, 0.0)).r, + texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos + vec3(0.0, EPSILON, 0.0)).r - texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos - vec3(0.0, EPSILON, 0.0)).r, + texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos + vec3(0.0, 0.0, EPSILON)).r - texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos - vec3(0.0, 0.0, EPSILON)).r)); + } + + } break; + case COLLIDER_TYPE_HEIGHT_FIELD: { + vec3 local_pos_bottom = local_pos; + local_pos_bottom.y -= particle_size; + + if (any(greaterThan(abs(local_pos_bottom), FRAME.colliders[i].extents))) { + continue; + } + + const float DELTA = 1.0 / 8192.0; + + vec3 uvw_pos = vec3(local_pos_bottom / FRAME.colliders[i].extents) * 0.5 + 0.5; + + float y = 1.0 - texture(sampler2D(height_field_texture, material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos.xz).r; + + if (y > uvw_pos.y) { + //inside heightfield + + vec3 pos1 = (vec3(uvw_pos.x, y, uvw_pos.z) * 2.0 - 1.0) * FRAME.colliders[i].extents; + vec3 pos2 = (vec3(uvw_pos.x + DELTA, 1.0 - texture(sampler2D(height_field_texture, material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos.xz + vec2(DELTA, 0)).r, uvw_pos.z) * 2.0 - 1.0) * FRAME.colliders[i].extents; + vec3 pos3 = (vec3(uvw_pos.x, 1.0 - texture(sampler2D(height_field_texture, material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos.xz + vec2(0, DELTA)).r, uvw_pos.z + DELTA) * 2.0 - 1.0) * FRAME.colliders[i].extents; + + normal = normalize(cross(pos1 - pos2, pos1 - pos3)); + float local_y = (vec3(local_pos / FRAME.colliders[i].extents) * 0.5 + 0.5).y; + + col = true; + depth = dot(normal, pos1) - dot(normal, local_pos_bottom); + } + + } break; + } + + if (col) { + if (!collided) { + collided = true; + collision_normal = normal; + collision_depth = depth; + } else { + vec3 c = collision_normal * collision_depth; + c += normal * max(0.0, depth - dot(normal, c)); + collision_normal = normalize(c); + collision_depth = length(c); + } + } + } + } + if (params.sub_emitter_mode) { if (!PARTICLE.is_active) { int src_index = atomicAdd(src_particles.particle_count, -1) - 1; @@ -329,66 +546,4 @@ COMPUTE_SHADER_CODE /* clang-format on */ } - -#if !defined(DISABLE_VELOCITY) - - if (PARTICLE.is_active) { - PARTICLE.xform[3].xyz += PARTICLE.velocity * local_delta; - } -#endif - -#if 0 - if (PARTICLE.is_active) { - //execute shader - - - - - //!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; - } - -#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 } diff --git a/servers/rendering/rasterizer_rd/shaders/scene_high_end.glsl b/servers/rendering/rasterizer_rd/shaders/scene_high_end.glsl index 2a7b73d9aa..da3c60af04 100644 --- a/servers/rendering/rasterizer_rd/shaders/scene_high_end.glsl +++ b/servers/rendering/rasterizer_rd/shaders/scene_high_end.glsl @@ -361,6 +361,65 @@ layout(location = 0) out vec4 frag_color; #endif // RENDER DEPTH +#ifdef ALPHA_HASH_USED + +float hash_2d(vec2 p) { + return fract(1.0e4 * sin(17.0 * p.x + 0.1 * p.y) * + (0.1 + abs(sin(13.0 * p.y + p.x)))); +} + +float hash_3d(vec3 p) { + return hash_2d(vec2(hash_2d(p.xy), p.z)); +} + +float compute_alpha_hash_threshold(vec3 pos, float hash_scale) { + vec3 dx = dFdx(pos); + vec3 dy = dFdx(pos); + float delta_max_sqr = max(length(dx), length(dy)); + float pix_scale = 1.0 / (hash_scale * delta_max_sqr); + + vec2 pix_scales = + vec2(exp2(floor(log2(pix_scale))), exp2(ceil(log2(pix_scale)))); + + vec2 a_thresh = vec2(hash_3d(floor(pix_scales.x * pos.xyz)), + hash_3d(floor(pix_scales.y * pos.xyz))); + + float lerp_factor = fract(log2(pix_scale)); + + float a_interp = (1.0 - lerp_factor) * a_thresh.x + lerp_factor * a_thresh.y; + + float min_lerp = min(lerp_factor, 1.0 - lerp_factor); + + vec3 cases = vec3(a_interp * a_interp / (2.0 * min_lerp * (1.0 - min_lerp)), + (a_interp - 0.5 * min_lerp) / (1.0 - min_lerp), + 1.0 - ((1.0 - a_interp) * (1.0 - a_interp) / + (2.0 * min_lerp * (1.0 - min_lerp)))); + + float alpha_hash_threshold = + (lerp_factor < (1.0 - min_lerp)) ? ((lerp_factor < min_lerp) ? cases.x : cases.y) : cases.z; + + return clamp(alpha_hash_threshold, 0.0, 1.0); +} + +#endif // ALPHA_HASH_USED + +#ifdef ALPHA_ANTIALIASING_EDGE_USED + +float calc_mip_level(vec2 texture_coord) { + vec2 dx = dFdx(texture_coord); + vec2 dy = dFdy(texture_coord); + float delta_max_sqr = max(dot(dx, dx), dot(dy, dy)); + return max(0.0, 0.5 * log2(delta_max_sqr)); +} + +float compute_alpha_antialiasing_edge(float input_alpha, vec2 texture_coord, float alpha_edge) { + input_alpha *= 1.0 + max(0, calc_mip_level(texture_coord)) * 0.25; // 0.25 mip scale, magic number + input_alpha = (input_alpha - alpha_edge) / max(fwidth(input_alpha), 0.0001) + 0.5; + return clamp(input_alpha, 0.0, 1.0); +} + +#endif // ALPHA_ANTIALIASING_USED + // 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): @@ -681,7 +740,7 @@ LIGHT_SHADER_CODE #ifndef USE_NO_SHADOWS -// Produces cheap white noise, optmized for window-space +// Produces cheap white noise, optimized for window-space // Comes from: https://www.shadertoy.com/view/4djSRW // Copyright: Dave Hoskins, MIT License float quick_hash(vec2 pos) { @@ -1621,6 +1680,22 @@ vec4 volumetric_fog_process(vec2 screen_uv, float z) { vec4 fog_process(vec3 vertex) { vec3 fog_color = scene_data.fog_light_color; + if (scene_data.fog_aerial_perspective > 0.0) { + vec3 sky_fog_color = vec3(0.0); + vec3 cube_view = scene_data.radiance_inverse_xform * vertex; + // mip_level always reads from the second mipmap and higher so the fog is always slightly blurred + float mip_level = mix(1.0 / MAX_ROUGHNESS_LOD, 1.0, 1.0 - (abs(vertex.z) - scene_data.z_near) / (scene_data.z_far - scene_data.z_near)); +#ifdef USE_RADIANCE_CUBEMAP_ARRAY + float lod, blend; + blend = modf(mip_level * MAX_ROUGHNESS_LOD, lod); + sky_fog_color = texture(samplerCubeArray(radiance_cubemap, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), vec4(cube_view, lod)).rgb; + sky_fog_color = mix(sky_fog_color, texture(samplerCubeArray(radiance_cubemap, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), vec4(cube_view, lod + 1)).rgb, blend); +#else + sky_fog_color = textureLod(samplerCube(radiance_cubemap, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), cube_view, mip_level * MAX_ROUGHNESS_LOD).rgb; +#endif //USE_RADIANCE_CUBEMAP_ARRAY + fog_color = mix(fog_color, sky_fog_color, scene_data.fog_aerial_perspective); + } + if (scene_data.fog_sun_scatter > 0.001) { vec4 sun_scatter = vec4(0.0); float sun_total = 0.0; @@ -1676,6 +1751,15 @@ void main() { float clearcoat_gloss = 0.0; float anisotropy = 0.0; vec2 anisotropy_flow = vec2(1.0, 0.0); +#if defined(CUSTOM_FOG_USED) + vec4 custom_fog = vec4(0.0); +#endif +#if defined(CUSTOM_RADIANCE_USED) + vec4 custom_radiance = vec4(0.0); +#endif +#if defined(CUSTOM_IRRADIANCE_USED) + vec4 custom_irradiance = vec4(0.0); +#endif #if defined(AO_USED) float ao = 1.0; @@ -1684,10 +1768,6 @@ void main() { float alpha = 1.0; -#if defined(ALPHA_SCISSOR_USED) - float alpha_scissor = 0.5; -#endif - #if defined(TANGENT_USED) || defined(NORMALMAP_USED) || defined(LIGHT_ANISOTROPY_USED) vec3 binormal = normalize(binormal_interp); vec3 tangent = normalize(tangent_interp); @@ -1724,6 +1804,19 @@ void main() { float sss_strength = 0.0; +#ifdef ALPHA_SCISSOR_USED + float alpha_scissor_threshold = 1.0; +#endif // ALPHA_SCISSOR_USED + +#ifdef ALPHA_HASH_USED + float alpha_hash_scale = 1.0; +#endif // ALPHA_HASH_USED + +#ifdef ALPHA_ANTIALIASING_EDGE_USED + float alpha_antialiasing_edge = 0.0; + vec2 alpha_texture_coordinate = vec2(0.0, 0.0); +#endif // ALPHA_ANTIALIASING_EDGE_USED + { /* clang-format off */ @@ -1732,7 +1825,7 @@ FRAGMENT_SHADER_CODE /* clang-format on */ } -#if defined(LIGHT_TRANSMITTANCE_USED) +#ifdef LIGHT_TRANSMITTANCE_USED #ifdef SSS_MODE_SKIN transmittance_color.a = sss_strength; #else @@ -1740,25 +1833,43 @@ FRAGMENT_SHADER_CODE #endif #endif -#if !defined(USE_SHADOW_TO_OPACITY) +#ifndef USE_SHADOW_TO_OPACITY -#if defined(ALPHA_SCISSOR_USED) - if (alpha < alpha_scissor) { +#ifdef ALPHA_SCISSOR_USED + if (alpha < alpha_scissor_threshold) { discard; } #endif // ALPHA_SCISSOR_USED -#ifdef USE_OPAQUE_PREPASS +// alpha hash can be used in unison with alpha antialiasing +#ifdef ALPHA_HASH_USED + if (alpha < compute_alpha_hash_threshold(vertex, alpha_hash_scale)) { + discard; + } +#endif // ALPHA_HASH_USED + +// If we are not edge antialiasing, we need to remove the output alpha channel from scissor and hash +#if (defined(ALPHA_SCISSOR_USED) || defined(ALPHA_HASH_USED)) && !defined(ALPHA_ANTIALIASING_EDGE_USED) + alpha = 1.0; +#endif + +#ifdef ALPHA_ANTIALIASING_EDGE_USED +// If alpha scissor is used, we must further the edge threshold, otherwise we wont get any edge feather +#ifdef ALPHA_SCISSOR_USED + alpha_antialiasing_edge = clamp(alpha_scissor_threshold + alpha_antialiasing_edge, 0.0, 1.0); +#endif + alpha = compute_alpha_antialiasing_edge(alpha, alpha_texture_coordinate, alpha_antialiasing_edge); +#endif // ALPHA_ANTIALIASING_EDGE_USED +#ifdef USE_OPAQUE_PREPASS if (alpha < opaque_prepass_threshold) { discard; } - #endif // USE_OPAQUE_PREPASS #endif // !USE_SHADOW_TO_OPACITY -#if defined(NORMALMAP_USED) +#ifdef NORMALMAP_USED normalmap.xy = normalmap.xy * 2.0 - 1.0; normalmap.z = sqrt(max(0.0, 1.0 - dot(normalmap.xy, normalmap.xy))); //always ignore Z, as it can be RG packed, Z may be pos/neg, etc. @@ -1767,7 +1878,7 @@ FRAGMENT_SHADER_CODE #endif -#if defined(LIGHT_ANISOTROPY_USED) +#ifdef LIGHT_ANISOTROPY_USED if (anisotropy > 0.01) { //rotation matrix @@ -1893,6 +2004,10 @@ FRAGMENT_SHADER_CODE specular_light *= scene_data.ambient_light_color_energy.a; } +#if defined(CUSTOM_RADIANCE_USED) + specular_light = mix(specular_light, custom_radiance.rgb, custom_radiance.a); +#endif + #ifndef USE_LIGHTMAP //lightmap overrides everything if (scene_data.use_ambient_light) { @@ -1910,7 +2025,9 @@ FRAGMENT_SHADER_CODE } } #endif // USE_LIGHTMAP - +#if defined(CUSTOM_IRRADIANCE_USED) + ambient_light = mix(specular_light, custom_irradiance.rgb, custom_irradiance.a); +#endif #endif //!defined(MODE_RENDER_DEPTH) && !defined(MODE_UNSHADED) //radiance @@ -2726,18 +2843,24 @@ FRAGMENT_SHADER_CODE specular_buffer = vec4(specular_light, metallic); #endif - if (scene_data.volumetric_fog_enabled) { - vec4 fog = volumetric_fog_process(screen_uv, -vertex.z); + // Draw "fixed" fog before volumetric fog to ensure volumetric fog can appear in front of the sky. + if (scene_data.fog_enabled) { + vec4 fog = fog_process(vertex); diffuse_buffer.rgb = mix(diffuse_buffer.rgb, fog.rgb, fog.a); specular_buffer.rgb = mix(specular_buffer.rgb, vec3(0.0), fog.a); } - if (scene_data.fog_enabled) { - vec4 fog = fog_process(vertex); + if (scene_data.volumetric_fog_enabled) { + vec4 fog = volumetric_fog_process(screen_uv, -vertex.z); diffuse_buffer.rgb = mix(diffuse_buffer.rgb, fog.rgb, fog.a); specular_buffer.rgb = mix(specular_buffer.rgb, vec3(0.0), fog.a); } +#if defined(CUSTOM_FOG_USED) + diffuse_buffer.rgb = mix(diffuse_buffer.rgb, custom_fog.rgb, custom_fog.a); + specular_buffer.rgb = mix(specular_buffer.rgb, vec3(0.0), custom_fog.a); +#endif //CUSTOM_FOG_USED + #else //MODE_MULTIPLE_RENDER_TARGETS #ifdef MODE_UNSHADED @@ -2747,16 +2870,21 @@ FRAGMENT_SHADER_CODE //frag_color = vec4(1.0); #endif //USE_NO_SHADING - if (scene_data.volumetric_fog_enabled) { - vec4 fog = volumetric_fog_process(screen_uv, -vertex.z); + // Draw "fixed" fog before volumetric fog to ensure volumetric fog can appear in front of the sky. + if (scene_data.fog_enabled) { + vec4 fog = fog_process(vertex); frag_color.rgb = mix(frag_color.rgb, fog.rgb, fog.a); } - if (scene_data.fog_enabled) { - vec4 fog = fog_process(vertex); + if (scene_data.volumetric_fog_enabled) { + vec4 fog = volumetric_fog_process(screen_uv, -vertex.z); frag_color.rgb = mix(frag_color.rgb, fog.rgb, fog.a); } +#if defined(CUSTOM_FOG_USED) + frag_color.rgb = mix(frag_color.rgb, custom_fog.rgb, custom_fog.a); +#endif //CUSTOM_FOG_USED + #endif //MODE_MULTIPLE_RENDER_TARGETS #endif //MODE_RENDER_DEPTH diff --git a/servers/rendering/rasterizer_rd/shaders/scene_high_end_inc.glsl b/servers/rendering/rasterizer_rd/shaders/scene_high_end_inc.glsl index 66bfefbe89..e29a490ca1 100644 --- a/servers/rendering/rasterizer_rd/shaders/scene_high_end_inc.glsl +++ b/servers/rendering/rasterizer_rd/shaders/scene_high_end_inc.glsl @@ -43,12 +43,6 @@ layout(set = 0, binding = 3, std140) uniform SceneData { vec2 viewport_size; vec2 screen_pixel_size; - float time; - float reflection_multiplier; // one normally, zero when rendering reflections - - bool pancake_shadows; - uint pad; - //use vec4s because std140 doesnt play nice with vec2s, z and w are wasted vec4 directional_penumbra_shadow_kernel[32]; vec4 directional_soft_shadow_kernel[32]; @@ -108,6 +102,13 @@ layout(set = 0, binding = 3, std140) uniform SceneData { vec3 fog_light_color; float fog_sun_scatter; + + float fog_aerial_perspective; + + float time; + float reflection_multiplier; // one normally, zero when rendering reflections + + bool pancake_shadows; } scene_data; @@ -252,7 +253,7 @@ layout(set = 1, binding = 0) uniform textureCube radiance_cubemap; #endif -/* Set 2, Reflection and Shadow Atlases (view dependant) */ +/* Set 2, Reflection and Shadow Atlases (view dependent) */ layout(set = 2, binding = 0) uniform textureCubeArray reflection_atlas; diff --git a/servers/rendering/rasterizer_rd/shaders/screen_space_reflection.glsl b/servers/rendering/rasterizer_rd/shaders/screen_space_reflection.glsl index a8ee33a664..06dc4b13de 100644 --- a/servers/rendering/rasterizer_rd/shaders/screen_space_reflection.glsl +++ b/servers/rendering/rasterizer_rd/shaders/screen_space_reflection.glsl @@ -155,18 +155,14 @@ void main() { depth = imageLoad(source_depth, ivec2(pos - 0.5)).r; - if (-depth >= params.camera_z_far) { //went beyond camera - break; - } - z_from = z_to; z_to = z / w; if (depth > z_to) { // if depth was surpassed - if (depth <= max(z_to, z_from) + params.depth_tolerance) { - // check the depth tolerance - //check that normal is valid + if (depth <= max(z_to, z_from) + params.depth_tolerance && -depth < params.camera_z_far) { + // check the depth tolerance and far clip + // check that normal is valid found = true; } break; diff --git a/servers/rendering/rasterizer_rd/shaders/sdfgi_direct_light.glsl b/servers/rendering/rasterizer_rd/shaders/sdfgi_direct_light.glsl index c4b29216d5..61e4bf5e18 100644 --- a/servers/rendering/rasterizer_rd/shaders/sdfgi_direct_light.glsl +++ b/servers/rendering/rasterizer_rd/shaders/sdfgi_direct_light.glsl @@ -22,7 +22,7 @@ dispatch_data; struct ProcessVoxel { uint position; //xyz 7 bit packed, extra 11 bits for neigbours uint albedo; //rgb bits 0-15 albedo, bits 16-21 are normal bits (set if geometry exists toward that side), extra 11 bits for neibhbours - uint light; //rgbe8985 encoded total saved light, extra 2 bits for neighbous + uint light; //rgbe8985 encoded total saved light, extra 2 bits for neighbours uint light_aniso; //55555 light anisotropy, extra 2 bits for neighbours //total neighbours: 26 }; diff --git a/servers/rendering/rasterizer_rd/shaders/sdfgi_integrate.glsl b/servers/rendering/rasterizer_rd/shaders/sdfgi_integrate.glsl index 1ec471d204..d516ab22c3 100644 --- a/servers/rendering/rasterizer_rd/shaders/sdfgi_integrate.glsl +++ b/servers/rendering/rasterizer_rd/shaders/sdfgi_integrate.glsl @@ -336,7 +336,7 @@ void main() { #ifdef MODE_STORE - // converting to octahedral in this step is requiered because + // converting to octahedral in this step is required because // octahedral is much faster to read from the screen than spherical harmonics, // despite the very slight quality loss @@ -512,7 +512,7 @@ void main() { imageStore(lightprobe_average_scroll_texture, dst_pos, value); } } else if (params.cascade < params.max_cascades - 1) { - //cant scroll, must look for position in parent cascade + //can't scroll, must look for position in parent cascade //to global coords float probe_cell_size = float(params.grid_size.x / float(params.probe_axis_size - 1)) / cascades.data[params.cascade].to_cell; diff --git a/servers/rendering/rasterizer_rd/shaders/sdfgi_preprocess.glsl b/servers/rendering/rasterizer_rd/shaders/sdfgi_preprocess.glsl index dd0ca5c506..916c60ac89 100644 --- a/servers/rendering/rasterizer_rd/shaders/sdfgi_preprocess.glsl +++ b/servers/rendering/rasterizer_rd/shaders/sdfgi_preprocess.glsl @@ -103,7 +103,7 @@ dispatch_data; struct ProcessVoxel { uint position; //xyz 7 bit packed, extra 11 bits for neigbours uint albedo; //rgb bits 0-15 albedo, bits 16-21 are normal bits (set if geometry exists toward that side), extra 11 bits for neibhbours - uint light; //rgbe8985 encoded total saved light, extra 2 bits for neighbous + uint light; //rgbe8985 encoded total saved light, extra 2 bits for neighbours uint light_aniso; //55555 light anisotropy, extra 2 bits for neighbours //total neighbours: 26 }; @@ -136,7 +136,7 @@ dispatch_data; struct ProcessVoxel { uint position; //xyz 7 bit packed, extra 11 bits for neigbours uint albedo; //rgb bits 0-15 albedo, bits 16-21 are normal bits (set if geometry exists toward that side), extra 11 bits for neibhbours - uint light; //rgbe8985 encoded total saved light, extra 2 bits for neighbous + uint light; //rgbe8985 encoded total saved light, extra 2 bits for neighbours uint light_aniso; //55555 light anisotropy, extra 2 bits for neighbours //total neighbours: 26 }; @@ -274,7 +274,7 @@ void main() { #ifdef MODE_JUMPFLOOD - //regular jumpflood, efficent for large steps, inefficient for small steps + //regular jumpflood, efficient for large steps, inefficient for small steps ivec3 pos = ivec3(gl_GlobalInvocationID.xyz); vec3 posf = vec3(pos); diff --git a/servers/rendering/rasterizer_rd/shaders/sky.glsl b/servers/rendering/rasterizer_rd/shaders/sky.glsl index 7b6de6a555..6c985e1f5c 100644 --- a/servers/rendering/rasterizer_rd/shaders/sky.glsl +++ b/servers/rendering/rasterizer_rd/shaders/sky.glsl @@ -62,7 +62,8 @@ layout(set = 0, binding = 2, std140) uniform SceneData { bool volumetric_fog_enabled; float volumetric_fog_inv_length; float volumetric_fog_detail_spread; - uint volumetric_fog_pad; + + float fog_aerial_perspective; vec3 fog_light_color; float fog_sun_scatter; @@ -140,8 +141,8 @@ vec4 volumetric_fog_process(vec2 screen_uv) { return texture(sampler3D(volumetric_fog_texture, material_samplers[SAMPLER_LINEAR_CLAMP]), fog_pos); } -vec4 fog_process(vec3 view) { - vec3 fog_color = scene_data.fog_light_color; +vec4 fog_process(vec3 view, vec3 sky_color) { + vec3 fog_color = mix(scene_data.fog_light_color, sky_color, scene_data.fog_aerial_perspective); if (scene_data.fog_sun_scatter > 0.001) { vec4 sun_scatter = vec4(0.0); @@ -181,6 +182,7 @@ void main() { float alpha = 1.0; // Only available to subpasses vec4 half_res_color = vec4(1.0); vec4 quarter_res_color = vec4(1.0); + vec4 custom_fog = vec4(0.0); #ifdef USE_CUBEMAP_PASS vec3 inverted_cube_normal = cube_normal; @@ -223,14 +225,19 @@ FRAGMENT_SHADER_CODE #if !defined(DISABLE_FOG) && !defined(USE_CUBEMAP_PASS) + // Draw "fixed" fog before volumetric fog to ensure volumetric fog can appear in front of the sky. + if (scene_data.fog_enabled) { + vec4 fog = fog_process(cube_normal, frag_color.rgb); + frag_color.rgb = mix(frag_color.rgb, fog.rgb, fog.a); + } + if (scene_data.volumetric_fog_enabled) { vec4 fog = volumetric_fog_process(uv); frag_color.rgb = mix(frag_color.rgb, fog.rgb, fog.a); } - if (scene_data.fog_enabled) { - vec4 fog = fog_process(cube_normal); - frag_color.rgb = mix(frag_color.rgb, fog.rgb, fog.a); + if (custom_fog.a > 0.0) { + frag_color.rgb = mix(frag_color.rgb, custom_fog.rgb, custom_fog.a); } #endif // DISABLE_FOG diff --git a/servers/rendering/rasterizer_rd/shaders/tonemap.glsl b/servers/rendering/rasterizer_rd/shaders/tonemap.glsl index b7c46a7d0e..4cc4fd3f64 100644 --- a/servers/rendering/rasterizer_rd/shaders/tonemap.glsl +++ b/servers/rendering/rasterizer_rd/shaders/tonemap.glsl @@ -37,16 +37,18 @@ layout(push_constant, binding = 1, std430) uniform Params { uvec2 glow_texture_size; float glow_intensity; - uint glow_level_flags; + uint pad3; uint glow_mode; + float glow_levels[7]; float exposure; float white; float auto_exposure_grey; + uint pad2; vec2 pixel_size; bool use_fxaa; - uint pad; + bool use_debanding; } params; @@ -155,6 +157,10 @@ vec3 tonemap_aces(vec3 color, float white) { } vec3 tonemap_reinhard(vec3 color, float white) { + // Ensure color values are positive. + // They can be negative in the case of negative lights, which leads to undesired behavior. + color = max(vec3(0.0), color); + return (white * color + color) / (color * white + white); } @@ -186,32 +192,32 @@ vec3 apply_tonemapping(vec3 color, float white) { // inputs are LINEAR, always o vec3 gather_glow(sampler2D tex, vec2 uv) { // sample all selected glow levels vec3 glow = vec3(0.0f); - if (bool(params.glow_level_flags & (1 << 0))) { - glow += GLOW_TEXTURE_SAMPLE(tex, uv, 0).rgb; + if (params.glow_levels[0] > 0.0001) { + glow += GLOW_TEXTURE_SAMPLE(tex, uv, 0).rgb * params.glow_levels[0]; } - if (bool(params.glow_level_flags & (1 << 1))) { - glow += GLOW_TEXTURE_SAMPLE(tex, uv, 1).rgb; + if (params.glow_levels[1] > 0.0001) { + glow += GLOW_TEXTURE_SAMPLE(tex, uv, 1).rgb * params.glow_levels[1]; } - if (bool(params.glow_level_flags & (1 << 2))) { - glow += GLOW_TEXTURE_SAMPLE(tex, uv, 2).rgb; + if (params.glow_levels[2] > 0.0001) { + glow += GLOW_TEXTURE_SAMPLE(tex, uv, 2).rgb * params.glow_levels[2]; } - if (bool(params.glow_level_flags & (1 << 3))) { - glow += GLOW_TEXTURE_SAMPLE(tex, uv, 3).rgb; + if (params.glow_levels[3] > 0.0001) { + glow += GLOW_TEXTURE_SAMPLE(tex, uv, 3).rgb * params.glow_levels[3]; } - if (bool(params.glow_level_flags & (1 << 4))) { - glow += GLOW_TEXTURE_SAMPLE(tex, uv, 4).rgb; + if (params.glow_levels[4] > 0.0001) { + glow += GLOW_TEXTURE_SAMPLE(tex, uv, 4).rgb * params.glow_levels[4]; } - if (bool(params.glow_level_flags & (1 << 5))) { - glow += GLOW_TEXTURE_SAMPLE(tex, uv, 5).rgb; + if (params.glow_levels[5] > 0.0001) { + glow += GLOW_TEXTURE_SAMPLE(tex, uv, 5).rgb * params.glow_levels[5]; } - if (bool(params.glow_level_flags & (1 << 6))) { - glow += GLOW_TEXTURE_SAMPLE(tex, uv, 6).rgb; + if (params.glow_levels[6] > 0.0001) { + glow += GLOW_TEXTURE_SAMPLE(tex, uv, 6).rgb * params.glow_levels[6]; } return glow; @@ -287,9 +293,8 @@ vec3 do_fxaa(vec3 color, float exposure, vec2 uv_interp) { dir * rcpDirMin)) * params.pixel_size; - vec3 rgbA = 0.5 * (textureLod(source_color, uv_interp + dir * (1.0 / 3.0 - 0.5), 0.0).xyz * exposure + textureLod(source_color, uv_interp + dir * (2.0 / 3.0 - 0.5), 0.0).xyz) * exposure; - vec3 rgbB = rgbA * 0.5 + 0.25 * (textureLod(source_color, uv_interp + dir * -0.5, 0.0).xyz * exposure + - textureLod(source_color, uv_interp + dir * 0.5, 0.0).xyz * exposure); + vec3 rgbA = 0.5 * exposure * (textureLod(source_color, uv_interp + dir * (1.0 / 3.0 - 0.5), 0.0).xyz + textureLod(source_color, uv_interp + dir * (2.0 / 3.0 - 0.5), 0.0).xyz); + vec3 rgbB = rgbA * 0.5 + 0.25 * exposure * (textureLod(source_color, uv_interp + dir * -0.5, 0.0).xyz + textureLod(source_color, uv_interp + dir * 0.5, 0.0).xyz); float lumaB = dot(rgbB, luma); if ((lumaB < lumaMin) || (lumaB > lumaMax)) { @@ -299,6 +304,18 @@ vec3 do_fxaa(vec3 color, float exposure, vec2 uv_interp) { } } +// From http://alex.vlachos.com/graphics/Alex_Vlachos_Advanced_VR_Rendering_GDC2015.pdf +// and https://www.shadertoy.com/view/MslGR8 (5th one starting from the bottom) +// NOTE: `frag_coord` is in pixels (i.e. not normalized UV). +vec3 screen_space_dither(vec2 frag_coord) { + // Iestyn's RGB dither (7 asm instructions) from Portal 2 X360, slightly modified for VR. + vec3 dither = vec3(dot(vec2(171.0, 231.0), frag_coord)); + dither.rgb = fract(dither.rgb / vec3(103.0, 71.0, 97.0)); + + // Subtract 0.5 to avoid slightly brightening the whole viewport. + return (dither.rgb - 0.5) / 255.0; +} + void main() { vec3 color = textureLod(source_color, uv_interp, 0.0f).rgb; @@ -322,6 +339,11 @@ void main() { if (params.use_fxaa) { color = do_fxaa(color, exposure, uv_interp); } + if (params.use_debanding) { + // For best results, debanding should be done before tonemapping. + // Otherwise, we're adding noise to an already-quantized image. + color += screen_space_dither(gl_FragCoord.xy); + } color = apply_tonemapping(color, params.white); color = linear_to_srgb(color); // regular linear -> SRGB conversion diff --git a/servers/rendering/rasterizer_rd/shaders/volumetric_fog.glsl b/servers/rendering/rasterizer_rd/shaders/volumetric_fog.glsl index cb19fb0b69..13b162f0c9 100644 --- a/servers/rendering/rasterizer_rd/shaders/volumetric_fog.glsl +++ b/servers/rendering/rasterizer_rd/shaders/volumetric_fog.glsl @@ -485,7 +485,7 @@ void main() { //get depth at cell pos float z = get_depth_at_pos(fog_cell_size.z, i); - //get distance from previos pos + //get distance from previous pos float d = abs(prev_z - z); //compute exinction based on beer's float extinction = t * exp(-d * fog.a); |