diff options
Diffstat (limited to 'servers/rendering')
55 files changed, 590 insertions, 422 deletions
diff --git a/servers/rendering/dummy/storage/texture_storage.h b/servers/rendering/dummy/storage/texture_storage.h index 62e1eb326d..fd36e7ac10 100644 --- a/servers/rendering/dummy/storage/texture_storage.h +++ b/servers/rendering/dummy/storage/texture_storage.h @@ -154,6 +154,7 @@ public: virtual RID decal_instance_create(RID p_decal) override { return RID(); } virtual void decal_instance_free(RID p_decal_instance) override {} virtual void decal_instance_set_transform(RID p_decal, const Transform3D &p_transform) override {} + virtual void decal_instance_set_sorting_offset(RID p_decal_instance, float p_sorting_offset) override {} /* RENDER TARGET */ diff --git a/servers/rendering/renderer_canvas_cull.cpp b/servers/rendering/renderer_canvas_cull.cpp index 3510b415d8..b9e3c4f303 100644 --- a/servers/rendering/renderer_canvas_cull.cpp +++ b/servers/rendering/renderer_canvas_cull.cpp @@ -597,7 +597,7 @@ void RendererCanvasCull::canvas_item_add_line(RID p_item, const Point2 &p_from, Vector2 end_left; Vector2 end_right; - if (p_width > 1.001) { + if (p_width >= 0.0) { begin_left = p_from + t; begin_right = p_from - t; end_left = p_to + t; @@ -628,7 +628,7 @@ void RendererCanvasCull::canvas_item_add_line(RID p_item, const Point2 &p_from, // This value is empirically determined to provide good antialiasing quality // while not making lines appear too soft. float border_size = 1.25f; - if (p_width < 1.0f) { + if (0.0f <= p_width && p_width < 1.0f) { border_size *= p_width; } Vector2 dir2 = diff.normalized(); @@ -769,6 +769,49 @@ void RendererCanvasCull::canvas_item_add_line(RID p_item, const Point2 &p_from, } } +static Vector2 compute_polyline_segment_dir(const Vector<Point2> &p_points, int p_index, const Vector2 &p_prev_segment_dir) { + int point_count = p_points.size(); + + bool is_last_point = (p_index == point_count - 1); + + Vector2 segment_dir; + + if (is_last_point) { + segment_dir = p_prev_segment_dir; + } else { + segment_dir = (p_points[p_index + 1] - p_points[p_index]).normalized(); + + if (segment_dir.is_zero_approx()) { + segment_dir = p_prev_segment_dir; + } + } + + return segment_dir; +} + +static Vector2 compute_polyline_edge_offset_clamped(const Vector2 &p_segment_dir, const Vector2 &p_prev_segment_dir) { + Vector2 bisector; + float length = 1.0f; + + bisector = (p_prev_segment_dir * p_segment_dir.length() - p_segment_dir * p_prev_segment_dir.length()).normalized(); + + float angle = atan2f(bisector.cross(p_prev_segment_dir), bisector.dot(p_prev_segment_dir)); + float sin_angle = sinf(angle); + + if (!Math::is_zero_approx(sin_angle) && !p_segment_dir.is_equal_approx(p_prev_segment_dir)) { + length = 1.0f / sin_angle; + length = CLAMP(length, -3.0f, 3.0f); + } else { + bisector = p_segment_dir.orthogonal(); + } + + if (bisector.is_zero_approx()) { + bisector = p_segment_dir.orthogonal(); + } + + return bisector * length; +} + void RendererCanvasCull::canvas_item_add_polyline(RID p_item, const Vector<Point2> &p_points, const Vector<Color> &p_colors, float p_width, bool p_antialiased) { ERR_FAIL_COND(p_points.size() < 2); Item *canvas_item = canvas_item_owner.get_or_null(p_item); @@ -777,20 +820,68 @@ void RendererCanvasCull::canvas_item_add_polyline(RID p_item, const Vector<Point Color color = Color(1, 1, 1, 1); Vector<int> indices; - int pc = p_points.size(); - int pc2 = pc * 2; - - Vector2 prev_t; - int j2; + int point_count = p_points.size(); Item::CommandPolygon *pline = canvas_item->alloc_command<Item::CommandPolygon>(); ERR_FAIL_COND(!pline); + if (p_width < 0) { + if (p_antialiased) { + WARN_PRINT("Antialiasing is not supported for thin polylines drawn using line strips (`p_width < 0`)."); + } + + pline->primitive = RS::PRIMITIVE_LINE_STRIP; + + if (p_colors.size() == 1 || p_colors.size() == point_count) { + pline->polygon.create(indices, p_points, p_colors); + } else { + Vector<Color> colors; + if (p_colors.is_empty()) { + colors.push_back(color); + } else { + colors.resize(point_count); + Color *colors_ptr = colors.ptrw(); + for (int i = 0; i < point_count; i++) { + if (i < p_colors.size()) { + color = p_colors[i]; + } + colors_ptr[i] = color; + } + } + pline->polygon.create(indices, p_points, colors); + } + return; + } + + int polyline_point_count = point_count * 2; + + bool loop = p_points[0].is_equal_approx(p_points[point_count - 1]); + Vector2 first_segment_dir; + Vector2 last_segment_dir; + + // Search for first non-zero vector between two segments. + for (int i = 1; i < point_count; i++) { + first_segment_dir = (p_points[i] - p_points[i - 1]).normalized(); + + if (!first_segment_dir.is_zero_approx()) { + break; + } + } + + // Search for last non-zero vector between two segments. + for (int i = point_count - 1; i >= 1; i--) { + last_segment_dir = (p_points[i] - p_points[i - 1]).normalized(); + + if (!last_segment_dir.is_zero_approx()) { + break; + } + } + PackedColorArray colors; PackedVector2Array points; - colors.resize(pc2); - points.resize(pc2); + colors.resize(polyline_point_count); + points.resize(polyline_point_count); Vector2 *points_ptr = points.ptrw(); Color *colors_ptr = colors.ptrw(); @@ -845,14 +936,14 @@ void RendererCanvasCull::canvas_item_add_polyline(RID p_item, const Vector<Point PackedColorArray colors_left; PackedVector2Array points_left; - colors_left.resize(pc2); - points_left.resize(pc2); + colors_left.resize(polyline_point_count); + points_left.resize(polyline_point_count); PackedColorArray colors_right; PackedVector2Array points_right; - colors_right.resize(pc2); - points_right.resize(pc2); + colors_right.resize(polyline_point_count); + points_right.resize(polyline_point_count); Item::CommandPolygon *pline_begin = canvas_item->alloc_command<Item::CommandPolygon>(); ERR_FAIL_COND(!pline_begin); @@ -898,79 +989,81 @@ void RendererCanvasCull::canvas_item_add_polyline(RID p_item, const Vector<Point Color *colors_left_ptr = colors_left.ptrw(); Color *colors_right_ptr = colors_right.ptrw(); - for (int i = 0, j = 0; i < pc; i++, j += 2) { - bool is_begin = i == 0; - bool is_end = i == pc - 1; + Vector2 prev_segment_dir; + for (int i = 0; i < point_count; i++) { + bool is_first_point = (i == 0); + bool is_last_point = (i == point_count - 1); - Vector2 t; - Vector2 end_border; - Vector2 begin_border; - if (is_end) { - t = prev_t; - end_border = (p_points[i] - p_points[i - 1]).normalized() * border_size; - } else { - t = (p_points[i + 1] - p_points[i]).normalized().orthogonal(); - if (is_begin) { - prev_t = t; - begin_border = (p_points[i] - p_points[i + 1]).normalized() * border_size; - } + Vector2 segment_dir = compute_polyline_segment_dir(p_points, i, prev_segment_dir); + if (is_first_point && loop) { + prev_segment_dir = last_segment_dir; + } else if (is_last_point && loop) { + prev_segment_dir = first_segment_dir; } - j2 = j + 1; + Vector2 base_edge_offset; + if (is_first_point && !loop) { + base_edge_offset = first_segment_dir.orthogonal(); + } else if (is_last_point && !loop) { + base_edge_offset = last_segment_dir.orthogonal(); + } else { + base_edge_offset = compute_polyline_edge_offset_clamped(segment_dir, prev_segment_dir); + } - Vector2 dir = (t + prev_t).normalized(); - Vector2 tangent = dir * p_width * 0.5; - Vector2 border = dir * border_size; + Vector2 edge_offset = base_edge_offset * (p_width * 0.5f); + Vector2 border = base_edge_offset * border_size; Vector2 pos = p_points[i]; - points_ptr[j] = pos + tangent; - points_ptr[j2] = pos - tangent; + points_ptr[i * 2 + 0] = pos + edge_offset; + points_ptr[i * 2 + 1] = pos - edge_offset; - points_left_ptr[j] = pos + tangent + border; - points_left_ptr[j2] = pos + tangent; + points_left_ptr[i * 2 + 0] = pos + edge_offset + border; + points_left_ptr[i * 2 + 1] = pos + edge_offset; - points_right_ptr[j] = pos - tangent; - points_right_ptr[j2] = pos - tangent - border; + points_right_ptr[i * 2 + 0] = pos - edge_offset; + points_right_ptr[i * 2 + 1] = pos - edge_offset - border; if (i < p_colors.size()) { color = p_colors[i]; color2 = Color(color.r, color.g, color.b, 0); } - colors_ptr[j] = color; - colors_ptr[j2] = color; + colors_ptr[i * 2 + 0] = color; + colors_ptr[i * 2 + 1] = color; + + colors_left_ptr[i * 2 + 0] = color2; + colors_left_ptr[i * 2 + 1] = color; - colors_left_ptr[j] = color2; - colors_left_ptr[j2] = color; + colors_right_ptr[i * 2 + 0] = color; + colors_right_ptr[i * 2 + 1] = color2; - colors_right_ptr[j] = color; - colors_right_ptr[j2] = color2; + if (is_first_point) { + Vector2 begin_border = loop ? Vector2() : -segment_dir * border_size; - if (is_begin) { - points_begin_ptr[0] = pos + tangent + begin_border; - points_begin_ptr[1] = pos - tangent + begin_border; - points_begin_ptr[2] = pos + tangent; - points_begin_ptr[3] = pos - tangent; + points_begin_ptr[0] = pos + edge_offset + begin_border; + points_begin_ptr[1] = pos - edge_offset + begin_border; + points_begin_ptr[2] = pos + edge_offset; + points_begin_ptr[3] = pos - edge_offset; colors_begin_ptr[0] = color2; colors_begin_ptr[1] = color2; colors_begin_ptr[2] = color; colors_begin_ptr[3] = color; - points_begin_left_corner_ptr[0] = pos - tangent - border; - points_begin_left_corner_ptr[1] = pos - tangent + begin_border - border; - points_begin_left_corner_ptr[2] = pos - tangent; - points_begin_left_corner_ptr[3] = pos - tangent + begin_border; + points_begin_left_corner_ptr[0] = pos - edge_offset - border; + points_begin_left_corner_ptr[1] = pos - edge_offset + begin_border - border; + points_begin_left_corner_ptr[2] = pos - edge_offset; + points_begin_left_corner_ptr[3] = pos - edge_offset + begin_border; colors_begin_left_corner_ptr[0] = color2; colors_begin_left_corner_ptr[1] = color2; colors_begin_left_corner_ptr[2] = color; colors_begin_left_corner_ptr[3] = color2; - points_begin_right_corner_ptr[0] = pos + tangent + begin_border; - points_begin_right_corner_ptr[1] = pos + tangent + begin_border + border; - points_begin_right_corner_ptr[2] = pos + tangent; - points_begin_right_corner_ptr[3] = pos + tangent + border; + points_begin_right_corner_ptr[0] = pos + edge_offset + begin_border; + points_begin_right_corner_ptr[1] = pos + edge_offset + begin_border + border; + points_begin_right_corner_ptr[2] = pos + edge_offset; + points_begin_right_corner_ptr[3] = pos + edge_offset + border; colors_begin_right_corner_ptr[0] = color2; colors_begin_right_corner_ptr[1] = color2; @@ -978,31 +1071,33 @@ void RendererCanvasCull::canvas_item_add_polyline(RID p_item, const Vector<Point colors_begin_right_corner_ptr[3] = color2; } - if (is_end) { - points_end_ptr[0] = pos + tangent + end_border; - points_end_ptr[1] = pos - tangent + end_border; - points_end_ptr[2] = pos + tangent; - points_end_ptr[3] = pos - tangent; + if (is_last_point) { + Vector2 end_border = loop ? Vector2() : prev_segment_dir * border_size; + + points_end_ptr[0] = pos + edge_offset + end_border; + points_end_ptr[1] = pos - edge_offset + end_border; + points_end_ptr[2] = pos + edge_offset; + points_end_ptr[3] = pos - edge_offset; colors_end_ptr[0] = color2; colors_end_ptr[1] = color2; colors_end_ptr[2] = color; colors_end_ptr[3] = color; - points_end_left_corner_ptr[0] = pos - tangent - border; - points_end_left_corner_ptr[1] = pos - tangent + end_border - border; - points_end_left_corner_ptr[2] = pos - tangent; - points_end_left_corner_ptr[3] = pos - tangent + end_border; + points_end_left_corner_ptr[0] = pos - edge_offset - border; + points_end_left_corner_ptr[1] = pos - edge_offset + end_border - border; + points_end_left_corner_ptr[2] = pos - edge_offset; + points_end_left_corner_ptr[3] = pos - edge_offset + end_border; colors_end_left_corner_ptr[0] = color2; colors_end_left_corner_ptr[1] = color2; colors_end_left_corner_ptr[2] = color; colors_end_left_corner_ptr[3] = color2; - points_end_right_corner_ptr[0] = pos + tangent + end_border; - points_end_right_corner_ptr[1] = pos + tangent + end_border + border; - points_end_right_corner_ptr[2] = pos + tangent; - points_end_right_corner_ptr[3] = pos + tangent + border; + points_end_right_corner_ptr[0] = pos + edge_offset + end_border; + points_end_right_corner_ptr[1] = pos + edge_offset + end_border + border; + points_end_right_corner_ptr[2] = pos + edge_offset; + points_end_right_corner_ptr[3] = pos + edge_offset + border; colors_end_right_corner_ptr[0] = color2; colors_end_right_corner_ptr[1] = color2; @@ -1010,7 +1105,7 @@ void RendererCanvasCull::canvas_item_add_polyline(RID p_item, const Vector<Point colors_end_right_corner_ptr[3] = color2; } - prev_t = t; + prev_segment_dir = segment_dir; } pline_begin->primitive = RS::PRIMITIVE_TRIANGLE_STRIP; @@ -1039,33 +1134,41 @@ void RendererCanvasCull::canvas_item_add_polyline(RID p_item, const Vector<Point } else { // Makes a single triangle strip for drawing the line. - for (int i = 0, j = 0; i < pc; i++, j += 2) { - Vector2 t; - if (i == pc - 1) { - t = prev_t; - } else { - t = (p_points[i + 1] - p_points[i]).normalized().orthogonal(); - if (i == 0) { - prev_t = t; - } + Vector2 prev_segment_dir; + for (int i = 0; i < point_count; i++) { + bool is_first_point = (i == 0); + bool is_last_point = (i == point_count - 1); + + Vector2 segment_dir = compute_polyline_segment_dir(p_points, i, prev_segment_dir); + if (is_first_point && loop) { + prev_segment_dir = last_segment_dir; + } else if (is_last_point && loop) { + prev_segment_dir = first_segment_dir; } - j2 = j + 1; + Vector2 base_edge_offset; + if (is_first_point && !loop) { + base_edge_offset = first_segment_dir.orthogonal(); + } else if (is_last_point && !loop) { + base_edge_offset = last_segment_dir.orthogonal(); + } else { + base_edge_offset = compute_polyline_edge_offset_clamped(segment_dir, prev_segment_dir); + } - Vector2 tangent = ((t + prev_t).normalized()) * p_width * 0.5; + Vector2 edge_offset = base_edge_offset * (p_width * 0.5f); Vector2 pos = p_points[i]; - points_ptr[j] = pos + tangent; - points_ptr[j2] = pos - tangent; + points_ptr[i * 2 + 0] = pos + edge_offset; + points_ptr[i * 2 + 1] = pos - edge_offset; if (i < p_colors.size()) { color = p_colors[i]; } - colors_ptr[j] = color; - colors_ptr[j2] = color; + colors_ptr[i * 2 + 0] = color; + colors_ptr[i * 2 + 1] = color; - prev_t = t; + prev_segment_dir = segment_dir; } } @@ -1077,7 +1180,7 @@ void RendererCanvasCull::canvas_item_add_multiline(RID p_item, const Vector<Poin ERR_FAIL_COND(p_points.size() < 2); // TODO: `canvas_item_add_line`(`multiline`, `polyline`) share logic, should factor out. - if (p_width <= 1) { + if (p_width < 0) { Item *canvas_item = canvas_item_owner.get_or_null(p_item); ERR_FAIL_COND(!canvas_item); @@ -1133,20 +1236,23 @@ void RendererCanvasCull::canvas_item_add_circle(RID p_item, const Point2 &p_pos, static const int circle_points = 64; points.resize(circle_points); + Vector2 *points_ptr = points.ptrw(); const real_t circle_point_step = Math_TAU / circle_points; for (int i = 0; i < circle_points; i++) { float angle = i * circle_point_step; - points.write[i].x = Math::cos(angle) * p_radius; - points.write[i].y = Math::sin(angle) * p_radius; - points.write[i] += p_pos; + points_ptr[i].x = Math::cos(angle) * p_radius; + points_ptr[i].y = Math::sin(angle) * p_radius; + points_ptr[i] += p_pos; } + indices.resize((circle_points - 2) * 3); + int *indices_ptr = indices.ptrw(); for (int i = 0; i < circle_points - 2; i++) { - indices.write[i * 3 + 0] = 0; - indices.write[i * 3 + 1] = i + 1; - indices.write[i * 3 + 2] = i + 2; + indices_ptr[i * 3 + 0] = 0; + indices_ptr[i * 3 + 1] = i + 1; + indices_ptr[i * 3 + 2] = i + 2; } Vector<Color> color; @@ -1313,7 +1419,7 @@ void RendererCanvasCull::canvas_item_add_nine_patch(RID p_item, const Rect2 &p_r style->axis_y = p_y_axis_mode; } -void RendererCanvasCull::canvas_item_add_primitive(RID p_item, const Vector<Point2> &p_points, const Vector<Color> &p_colors, const Vector<Point2> &p_uvs, RID p_texture, float p_width) { +void RendererCanvasCull::canvas_item_add_primitive(RID p_item, const Vector<Point2> &p_points, const Vector<Color> &p_colors, const Vector<Point2> &p_uvs, RID p_texture) { uint32_t pc = p_points.size(); ERR_FAIL_COND(pc == 0 || pc > 4); diff --git a/servers/rendering/renderer_canvas_cull.h b/servers/rendering/renderer_canvas_cull.h index 4d0be48cc0..1106fc4f1e 100644 --- a/servers/rendering/renderer_canvas_cull.h +++ b/servers/rendering/renderer_canvas_cull.h @@ -221,9 +221,9 @@ public: void canvas_item_set_update_when_visible(RID p_item, bool p_update); - void canvas_item_add_line(RID p_item, const Point2 &p_from, const Point2 &p_to, const Color &p_color, float p_width = 1.0, bool p_antialiased = false); - void canvas_item_add_polyline(RID p_item, const Vector<Point2> &p_points, const Vector<Color> &p_colors, float p_width = 1.0, bool p_antialiased = false); - void canvas_item_add_multiline(RID p_item, const Vector<Point2> &p_points, const Vector<Color> &p_colors, float p_width = 1.0); + void canvas_item_add_line(RID p_item, const Point2 &p_from, const Point2 &p_to, const Color &p_color, float p_width = -1.0, bool p_antialiased = false); + void canvas_item_add_polyline(RID p_item, const Vector<Point2> &p_points, const Vector<Color> &p_colors, float p_width = -1.0, bool p_antialiased = false); + void canvas_item_add_multiline(RID p_item, const Vector<Point2> &p_points, const Vector<Color> &p_colors, float p_width = -1.0); void canvas_item_add_rect(RID p_item, const Rect2 &p_rect, const Color &p_color); void canvas_item_add_circle(RID p_item, const Point2 &p_pos, float p_radius, const Color &p_color); void canvas_item_add_texture_rect(RID p_item, const Rect2 &p_rect, RID p_texture, bool p_tile = false, const Color &p_modulate = Color(1, 1, 1), bool p_transpose = false); @@ -231,7 +231,7 @@ public: void canvas_item_add_msdf_texture_rect_region(RID p_item, const Rect2 &p_rect, RID p_texture, const Rect2 &p_src_rect, const Color &p_modulate = Color(1, 1, 1), int p_outline_size = 0, float p_px_range = 1.0, float p_scale = 1.0); void canvas_item_add_lcd_texture_rect_region(RID p_item, const Rect2 &p_rect, RID p_texture, const Rect2 &p_src_rect, const Color &p_modulate = Color(1, 1, 1)); void canvas_item_add_nine_patch(RID p_item, const Rect2 &p_rect, const Rect2 &p_source, RID p_texture, const Vector2 &p_topleft, const Vector2 &p_bottomright, RS::NinePatchAxisMode p_x_axis_mode = RS::NINE_PATCH_STRETCH, RS::NinePatchAxisMode p_y_axis_mode = RS::NINE_PATCH_STRETCH, bool p_draw_center = true, const Color &p_modulate = Color(1, 1, 1)); - void canvas_item_add_primitive(RID p_item, const Vector<Point2> &p_points, const Vector<Color> &p_colors, const Vector<Point2> &p_uvs, RID p_texture, float p_width = 1.0); + void canvas_item_add_primitive(RID p_item, const Vector<Point2> &p_points, const Vector<Color> &p_colors, const Vector<Point2> &p_uvs, RID p_texture); void canvas_item_add_polygon(RID p_item, const Vector<Point2> &p_points, const Vector<Color> &p_colors, const Vector<Point2> &p_uvs = Vector<Point2>(), RID p_texture = RID()); void canvas_item_add_triangle_array(RID p_item, const Vector<int> &p_indices, const Vector<Point2> &p_points, const Vector<Color> &p_colors, const Vector<Point2> &p_uvs = Vector<Point2>(), const Vector<int> &p_bones = Vector<int>(), const Vector<float> &p_weights = Vector<float>(), RID p_texture = RID(), int p_count = -1); void canvas_item_add_mesh(RID p_item, const RID &p_mesh, const Transform2D &p_transform = Transform2D(), const Color &p_modulate = Color(1, 1, 1), RID p_texture = RID()); diff --git a/servers/rendering/renderer_rd/cluster_builder_rd.cpp b/servers/rendering/renderer_rd/cluster_builder_rd.cpp index 73a0c652a4..959a752fba 100644 --- a/servers/rendering/renderer_rd/cluster_builder_rd.cpp +++ b/servers/rendering/renderer_rd/cluster_builder_rd.cpp @@ -74,7 +74,7 @@ ClusterBuilderSharedDataRD::ClusterBuilderSharedDataRD() { cluster_debug.shader_pipeline = RD::get_singleton()->compute_pipeline_create(cluster_debug.shader); } - { // SPHERE + { // Sphere mesh data. static const uint32_t icosphere_vertex_count = 42; static const float icosphere_vertices[icosphere_vertex_count * 3] = { 0, 0, -1, 0.7236073, -0.5257253, -0.4472195, -0.276388, -0.8506492, -0.4472199, -0.8944262, 0, -0.4472156, -0.276388, 0.8506492, -0.4472199, 0.7236073, 0.5257253, -0.4472195, 0.276388, -0.8506492, 0.4472199, -0.7236073, -0.5257253, 0.4472195, -0.7236073, 0.5257253, 0.4472195, 0.276388, 0.8506492, 0.4472199, 0.8944262, 0, 0.4472156, 0, 0, 1, -0.1624555, -0.4999952, -0.8506544, 0.4253227, -0.3090114, -0.8506542, 0.2628688, -0.8090116, -0.5257377, 0.8506479, 0, -0.5257359, 0.4253227, 0.3090114, -0.8506542, -0.5257298, 0, -0.8506517, -0.6881894, -0.4999969, -0.5257362, -0.1624555, 0.4999952, -0.8506544, -0.6881894, 0.4999969, -0.5257362, 0.2628688, 0.8090116, -0.5257377, 0.9510579, -0.3090126, 0, 0.9510579, 0.3090126, 0, 0, -1, 0, 0.5877856, -0.8090167, 0, -0.9510579, -0.3090126, 0, -0.5877856, -0.8090167, 0, -0.5877856, 0.8090167, 0, -0.9510579, 0.3090126, 0, 0.5877856, 0.8090167, 0, 0, 1, 0, 0.6881894, -0.4999969, 0.5257362, -0.2628688, -0.8090116, 0.5257377, -0.8506479, 0, 0.5257359, -0.2628688, 0.8090116, 0.5257377, 0.6881894, 0.4999969, 0.5257362, 0.1624555, -0.4999952, 0.8506544, 0.5257298, 0, 0.8506517, -0.4253227, -0.3090114, 0.8506542, -0.4253227, 0.3090114, 0.8506542, 0.1624555, 0.4999952, 0.8506544 @@ -118,7 +118,7 @@ ClusterBuilderSharedDataRD::ClusterBuilderSharedDataRD() { sphere_overfit = 1.0 / min_d; } - { // CONE + { // Cone mesh data. static const uint32_t cone_vertex_count = 99; static const float cone_vertices[cone_vertex_count * 3] = { 0, 1, -1, 0.1950903, 0.9807853, -1, 0.3826835, 0.9238795, -1, 0.5555703, 0.8314696, -1, 0.7071068, 0.7071068, -1, 0.8314697, 0.5555702, -1, 0.9238795, 0.3826834, -1, 0.9807853, 0.1950903, -1, 1, 0, -1, 0.9807853, -0.1950902, -1, 0.9238796, -0.3826833, -1, 0.8314697, -0.5555702, -1, 0.7071068, -0.7071068, -1, 0.5555702, -0.8314697, -1, 0.3826833, -0.9238796, -1, 0.1950901, -0.9807853, -1, -3.25841e-7, -1, -1, -0.1950907, -0.9807852, -1, -0.3826839, -0.9238793, -1, -0.5555707, -0.8314693, -1, -0.7071073, -0.7071063, -1, -0.83147, -0.5555697, -1, -0.9238799, -0.3826827, -1, 0, 0, 0, -0.9807854, -0.1950894, -1, -1, 9.65599e-7, -1, -0.9807851, 0.1950913, -1, -0.9238791, 0.3826845, -1, -0.8314689, 0.5555713, -1, -0.7071059, 0.7071077, -1, -0.5555691, 0.8314704, -1, -0.3826821, 0.9238801, -1, -0.1950888, 0.9807856, -1 @@ -172,7 +172,7 @@ ClusterBuilderSharedDataRD::ClusterBuilderSharedDataRD() { cone_overfit = 1.0 / min_d; } - { // BOX + { // Box mesh data. static const uint32_t box_vertex_count = 8; static const float box_vertices[box_vertex_count * 3] = { -1, -1, -1, -1, -1, 1, -1, 1, -1, -1, 1, 1, 1, -1, -1, 1, -1, 1, 1, 1, -1, 1, 1, 1 @@ -219,8 +219,9 @@ ClusterBuilderSharedDataRD::~ClusterBuilderSharedDataRD() { void ClusterBuilderRD::_clear() { if (cluster_buffer.is_null()) { - return; //nothing to clear + return; } + RD::get_singleton()->free(cluster_buffer); RD::get_singleton()->free(cluster_render_buffer); RD::get_singleton()->free(element_buffer); @@ -254,7 +255,7 @@ void ClusterBuilderRD::setup(Size2i p_screen_size, uint32_t p_max_elements, RID cluster_screen_size.height = (p_screen_size.height - 1) / cluster_size + 1; max_elements_by_type = p_max_elements; - if (max_elements_by_type % 32) { //need to be 32 aligned + if (max_elements_by_type % 32) { // Needs to be aligned to 32. max_elements_by_type += 32 - (max_elements_by_type % 32); } @@ -264,7 +265,8 @@ void ClusterBuilderRD::setup(Size2i p_screen_size, uint32_t p_max_elements, RID uint32_t element_tag_bits_size = render_element_max / 32; uint32_t element_tag_depth_bits_size = render_element_max; - cluster_render_buffer_size = cluster_screen_size.x * cluster_screen_size.y * (element_tag_bits_size + element_tag_depth_bits_size) * 4; // tag bits (element was used) and tag depth (depth range in which it was used) + + cluster_render_buffer_size = cluster_screen_size.x * cluster_screen_size.y * (element_tag_bits_size + element_tag_depth_bits_size) * 4; // Tag bits (element was used) and tag depth (depth range in which it was used). cluster_render_buffer = RD::get_singleton()->storage_buffer_create(cluster_render_buffer_size); cluster_buffer = RD::get_singleton()->storage_buffer_create(cluster_buffer_size); @@ -379,9 +381,9 @@ void ClusterBuilderRD::begin(const Transform3D &p_view_transform, const Projecti projection = p_cam_projection; z_near = projection.get_z_near(); z_far = projection.get_z_far(); - orthogonal = p_cam_projection.is_orthogonal(); + camera_orthogonal = p_cam_projection.is_orthogonal(); adjusted_projection = projection; - if (!orthogonal) { + if (!camera_orthogonal) { adjusted_projection.adjust_perspective_znear(0.0001); } @@ -390,7 +392,7 @@ void ClusterBuilderRD::begin(const Transform3D &p_view_transform, const Projecti projection = correction * projection; adjusted_projection = correction * adjusted_projection; - //reset counts + // Reset counts. render_element_count = 0; for (uint32_t i = 0; i < ELEMENT_TYPE_MAX; i++) { cluster_count_by_type[i] = 0; @@ -402,14 +404,14 @@ void ClusterBuilderRD::bake_cluster() { RD::get_singleton()->draw_command_begin_label("Bake Light Cluster"); - //clear cluster buffer + // Clear cluster buffer. RD::get_singleton()->buffer_clear(cluster_buffer, 0, cluster_buffer_size, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); if (render_element_count > 0) { - //clear render buffer + // Clear render buffer. RD::get_singleton()->buffer_clear(cluster_render_buffer, 0, cluster_render_buffer_size, RD::BARRIER_MASK_RASTER); - { //fill state uniform + { // Fill state uniform. StateUniform state; @@ -425,13 +427,13 @@ void ClusterBuilderRD::bake_cluster() { RD::get_singleton()->buffer_update(state_uniform, 0, sizeof(StateUniform), &state, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); } - //update instances + // Update instances. RD::get_singleton()->buffer_update(element_buffer, 0, sizeof(RenderElementData) * render_element_count, render_elements, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); RENDER_TIMESTAMP("Render 3D Cluster Elements"); - //render elements + // Render elements. { RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(framebuffer, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD); ClusterBuilderSharedDataRD::ClusterRender::PushConstant push_constant = {}; @@ -447,8 +449,16 @@ void ClusterBuilderRD::bake_cluster() { RD::get_singleton()->draw_list_bind_index_array(draw_list, shared->sphere_index_array); } break; case ELEMENT_TYPE_SPOT_LIGHT: { - RD::get_singleton()->draw_list_bind_vertex_array(draw_list, shared->cone_vertex_array); - RD::get_singleton()->draw_list_bind_index_array(draw_list, shared->cone_index_array); + // If the spot angle is above a certain threshold, use a sphere instead of a cone for building the clusters + // since the cone gets too flat/large (spot angle close to 90 degrees) or + // can't even cover the affected area of the light (spot angle above 90 degrees). + if (render_elements[i].has_wide_spot_angle) { + RD::get_singleton()->draw_list_bind_vertex_array(draw_list, shared->sphere_vertex_array); + RD::get_singleton()->draw_list_bind_index_array(draw_list, shared->sphere_index_array); + } else { + RD::get_singleton()->draw_list_bind_vertex_array(draw_list, shared->cone_vertex_array); + RD::get_singleton()->draw_list_bind_index_array(draw_list, shared->cone_index_array); + } } break; case ELEMENT_TYPE_DECAL: case ELEMENT_TYPE_REFLECTION_PROBE: { @@ -465,7 +475,7 @@ void ClusterBuilderRD::bake_cluster() { } RD::get_singleton()->draw_list_end(RD::BARRIER_MASK_COMPUTE); } - //store elements + // Store elements. RENDER_TIMESTAMP("Pack 3D Cluster Elements"); { @@ -509,7 +519,7 @@ void ClusterBuilderRD::debug(ElementType p_element) { push_constant.cluster_screen_size[1] = cluster_screen_size.y; push_constant.cluster_shift = get_shift_from_power_of_2(cluster_size); push_constant.cluster_type = p_element; - push_constant.orthogonal = orthogonal; + push_constant.orthogonal = camera_orthogonal; push_constant.z_far = z_far; push_constant.z_near = z_near; push_constant.max_cluster_element_count_div_32 = max_elements_by_type / 32; diff --git a/servers/rendering/renderer_rd/cluster_builder_rd.h b/servers/rendering/renderer_rd/cluster_builder_rd.h index 0b20a5d7ee..a13e6c8172 100644 --- a/servers/rendering/renderer_rd/cluster_builder_rd.h +++ b/servers/rendering/renderer_rd/cluster_builder_rd.h @@ -43,13 +43,13 @@ class ClusterBuilderSharedDataRD { RID sphere_vertex_array; RID sphere_index_buffer; RID sphere_index_array; - float sphere_overfit = 0.0; //because an icosphere is not a perfect sphere, we need to enlarge it to cover the sphere area + float sphere_overfit = 0.0; // Because an icosphere is not a perfect sphere, we need to enlarge it to cover the sphere area. RID cone_vertex_buffer; RID cone_vertex_array; RID cone_index_buffer; RID cone_index_array; - float cone_overfit = 0.0; //because an cone mesh is not a perfect sphere, we need to enlarge it to cover the actual cone area + float cone_overfit = 0.0; // Because an cone mesh is not a perfect cone, we need to enlarge it to cover the actual cone area. RID box_vertex_buffer; RID box_vertex_array; @@ -73,6 +73,7 @@ class ClusterBuilderSharedDataRD { ClusterRenderShaderRD cluster_render_shader; RID shader_version; RID shader; + enum PipelineVersion { PIPELINE_NORMAL, PIPELINE_MSAA, @@ -85,10 +86,11 @@ class ClusterBuilderSharedDataRD { struct ClusterStore { struct PushConstant { uint32_t cluster_render_data_size; // how much data for a single cluster takes - uint32_t max_render_element_count_div_32; //divided by 32 + uint32_t max_render_element_count_div_32; // divided by 32 uint32_t cluster_screen_size[2]; - uint32_t render_element_count_div_32; //divided by 32 - uint32_t max_cluster_element_count_div_32; //divided by 32 + uint32_t render_element_count_div_32; // divided by 32 + uint32_t max_cluster_element_count_div_32; // divided by 32 + uint32_t pad1; uint32_t pad2; }; @@ -111,6 +113,7 @@ class ClusterBuilderSharedDataRD { uint32_t orthogonal; uint32_t max_cluster_element_count_div_32; + uint32_t pad1; uint32_t pad2; }; @@ -128,6 +131,8 @@ public: class ClusterBuilderRD { public: + static constexpr float WIDE_SPOT_ANGLE_THRESHOLD_DEG = 60.0f; + enum LightType { LIGHT_TYPE_OMNI, LIGHT_TYPE_SPOT @@ -144,21 +149,20 @@ public: ELEMENT_TYPE_DECAL, ELEMENT_TYPE_REFLECTION_PROBE, ELEMENT_TYPE_MAX, - }; private: ClusterBuilderSharedDataRD *shared = nullptr; struct RenderElementData { - uint32_t type; //0-4 + uint32_t type; // 0-4 uint32_t touches_near; uint32_t touches_far; uint32_t original_index; - float transform_inv[12]; //transposed transform for less space + float transform_inv[12]; // Transposed transform for less space. float scale[3]; - uint32_t pad; - }; + uint32_t has_wide_spot_angle; + }; // Keep aligned to 32 bytes. uint32_t cluster_count_by_type[ELEMENT_TYPE_MAX] = {}; uint32_t max_elements_by_type = 0; @@ -172,7 +176,7 @@ private: Projection projection; float z_far = 0; float z_near = 0; - bool orthogonal = false; + bool camera_orthogonal = false; enum Divisor { DIVISOR_1, @@ -188,26 +192,27 @@ private: Size2i cluster_screen_size; RID framebuffer; - RID cluster_render_buffer; //used for creating - RID cluster_buffer; //used for rendering - RID element_buffer; //used for storing, to hint element touches far plane or near plane + RID cluster_render_buffer; // Used for creating. + RID cluster_buffer; // Used for rendering. + RID element_buffer; // Used for storing, to hint element touches far plane or near plane. uint32_t cluster_render_buffer_size = 0; uint32_t cluster_buffer_size = 0; RID cluster_render_uniform_set; RID cluster_store_uniform_set; - //persistent data + // Persistent data. void _clear(); struct StateUniform { float projection[16]; float inv_z_far; - uint32_t screen_to_clusters_shift; // shift to obtain coordinates in block indices - uint32_t cluster_screen_width; // - uint32_t cluster_data_size; // how much data for a single cluster takes + uint32_t screen_to_clusters_shift; // Shift to obtain coordinates in block indices. + uint32_t cluster_screen_width; + uint32_t cluster_data_size; // How much data is needed for a single cluster. uint32_t cluster_depth_offset; + uint32_t pad0; uint32_t pad1; uint32_t pad2; @@ -224,10 +229,10 @@ public: _FORCE_INLINE_ void add_light(LightType p_type, const Transform3D &p_transform, float p_radius, float p_spot_aperture) { if (p_type == LIGHT_TYPE_OMNI && cluster_count_by_type[ELEMENT_TYPE_OMNI_LIGHT] == max_elements_by_type) { - return; //max number elements reached + return; // Max number elements reached. } if (p_type == LIGHT_TYPE_SPOT && cluster_count_by_type[ELEMENT_TYPE_SPOT_LIGHT] == max_elements_by_type) { - return; //max number elements reached + return; // Max number elements reached. } RenderElementData &e = render_elements[render_element_count]; @@ -242,15 +247,14 @@ public: radius *= p_radius; if (p_type == LIGHT_TYPE_OMNI) { - radius *= shared->sphere_overfit; // overfit icosphere + radius *= shared->sphere_overfit; // Overfit icosphere. - //omni float depth = -xform.origin.z; - if (orthogonal) { + if (camera_orthogonal) { e.touches_near = (depth - radius) < z_near; } else { - //contains camera inside light - float radius2 = radius * shared->sphere_overfit; // overfit again for outer size (camera may be outside actual sphere but behind an icosphere vertex) + // Contains camera inside light. + float radius2 = radius * shared->sphere_overfit; // Overfit again for outer size (camera may be outside actual sphere but behind an icosphere vertex) e.touches_near = xform.origin.length_squared() < radius2 * radius2; } @@ -265,12 +269,11 @@ public: cluster_count_by_type[ELEMENT_TYPE_OMNI_LIGHT]++; - } else { - //spot - radius *= shared->cone_overfit; // overfit icosphere + } else /*LIGHT_TYPE_SPOT */ { + radius *= shared->cone_overfit; // Overfit icosphere real_t len = Math::tan(Math::deg_to_rad(p_spot_aperture)) * radius; - //approximate, probably better to use a cone support function + // Approximate, probably better to use a cone support function. float max_d = -1e20; float min_d = 1e20; #define CONE_MINMAX(m_x, m_y) \ @@ -285,14 +288,13 @@ public: CONE_MINMAX(-1, -1); CONE_MINMAX(1, -1); - if (orthogonal) { + if (camera_orthogonal) { e.touches_near = min_d < z_near; } else { - //contains camera inside light Plane base_plane(-xform.basis.get_column(Vector3::AXIS_Z), xform.origin); float dist = base_plane.distance_to(Vector3()); if (dist >= 0 && dist < radius) { - //inside, check angle + // Contains camera inside light, check angle. float angle = Math::rad_to_deg(Math::acos((-xform.origin.normalized()).dot(-xform.basis.get_column(Vector3::AXIS_Z)))); e.touches_near = angle < p_spot_aperture * 1.05; //overfit aperture a little due to cone overfit } else { @@ -302,12 +304,23 @@ public: e.touches_far = max_d > z_far; - e.scale[0] = len * shared->cone_overfit; - e.scale[1] = len * shared->cone_overfit; - e.scale[2] = radius; + // If the spot angle is above the threshold, use a sphere instead of a cone for building the clusters + // since the cone gets too flat/large (spot angle close to 90 degrees) or + // can't even cover the affected area of the light (spot angle above 90 degrees). + if (p_spot_aperture > WIDE_SPOT_ANGLE_THRESHOLD_DEG) { + e.scale[0] = radius; + e.scale[1] = radius; + e.scale[2] = radius; + e.has_wide_spot_angle = true; + } else { + e.scale[0] = len * shared->cone_overfit; + e.scale[1] = len * shared->cone_overfit; + e.scale[2] = radius; + e.has_wide_spot_angle = false; + } e.type = ELEMENT_TYPE_SPOT_LIGHT; - e.original_index = cluster_count_by_type[ELEMENT_TYPE_SPOT_LIGHT]; //use omni since they share index + e.original_index = cluster_count_by_type[ELEMENT_TYPE_SPOT_LIGHT]; // Use omni light since they share index. RendererRD::MaterialStorage::store_transform_transposed_3x4(xform, e.transform_inv); @@ -319,16 +332,16 @@ public: _FORCE_INLINE_ void add_box(BoxType p_box_type, const Transform3D &p_transform, const Vector3 &p_half_extents) { if (p_box_type == BOX_TYPE_DECAL && cluster_count_by_type[ELEMENT_TYPE_DECAL] == max_elements_by_type) { - return; //max number elements reached + return; // Max number elements reached. } if (p_box_type == BOX_TYPE_REFLECTION_PROBE && cluster_count_by_type[ELEMENT_TYPE_REFLECTION_PROBE] == max_elements_by_type) { - return; //max number elements reached + return; // Max number elements reached. } RenderElementData &e = render_elements[render_element_count]; Transform3D xform = view_xform * p_transform; - //extract scale and scale the matrix by it, makes things simpler + // Extract scale and scale the matrix by it, makes things simpler. Vector3 scale = p_half_extents; for (uint32_t i = 0; i < 3; i++) { float s = xform.basis.rows[i].length(); @@ -339,10 +352,10 @@ public: float box_depth = Math::abs(xform.basis.xform_inv(Vector3(0, 0, -1)).dot(scale)); float depth = -xform.origin.z; - if (orthogonal) { + if (camera_orthogonal) { e.touches_near = depth - box_depth < z_near; } else { - //contains camera inside box + // Contains camera inside box. Vector3 inside = xform.xform_inv(Vector3(0, 0, 0)).abs(); e.touches_near = inside.x < scale.x && inside.y < scale.y && inside.z < scale.z; } diff --git a/servers/rendering/renderer_rd/environment/gi.cpp b/servers/rendering/renderer_rd/environment/gi.cpp index a52716cd78..08133bf8d6 100644 --- a/servers/rendering/renderer_rd/environment/gi.cpp +++ b/servers/rendering/renderer_rd/environment/gi.cpp @@ -540,9 +540,7 @@ void GI::SDFGI::create(RID p_env, const Vector3 &p_world_position, uint32_t p_re occlusion_texture = RD::get_singleton()->texture_create_shared(tv, occlusion_data); } - for (uint32_t i = 0; i < cascades.size(); i++) { - SDFGI::Cascade &cascade = cascades[i]; - + for (SDFGI::Cascade &cascade : cascades) { /* 3D Textures */ cascade.sdf_tex = RD::get_singleton()->texture_create(tf_sdf, RD::TextureView()); @@ -743,9 +741,7 @@ void GI::SDFGI::create(RID p_env, const Vector3 &p_world_position, uint32_t p_re } //direct light - for (uint32_t i = 0; i < cascades.size(); i++) { - SDFGI::Cascade &cascade = cascades[i]; - + for (SDFGI::Cascade &cascade : cascades) { Vector<RD::Uniform> uniforms; { RD::Uniform u; @@ -1134,8 +1130,7 @@ void GI::SDFGI::free_data() { } GI::SDFGI::~SDFGI() { - for (uint32_t i = 0; i < cascades.size(); i++) { - const SDFGI::Cascade &c = cascades[i]; + for (const SDFGI::Cascade &c : cascades) { RD::get_singleton()->free(c.light_data); RD::get_singleton()->free(c.light_aniso_0_tex); RD::get_singleton()->free(c.light_aniso_1_tex); @@ -1198,8 +1193,7 @@ void GI::SDFGI::update(RID p_env, const Vector3 &p_world_position) { int32_t drag_margin = (cascade_size / SDFGI::PROBE_DIVISOR) / 2; - for (uint32_t i = 0; i < cascades.size(); i++) { - SDFGI::Cascade &cascade = cascades[i]; + for (SDFGI::Cascade &cascade : cascades) { cascade.dirty_regions = Vector3i(); Vector3 probe_half_size = Vector3(1, 1, 1) * cascade.cell_size * float(cascade_size / SDFGI::PROBE_DIVISOR) * 0.5; diff --git a/servers/rendering/renderer_rd/environment/sky.cpp b/servers/rendering/renderer_rd/environment/sky.cpp index 863e8d6c15..1dbad79477 100644 --- a/servers/rendering/renderer_rd/environment/sky.cpp +++ b/servers/rendering/renderer_rd/environment/sky.cpp @@ -347,7 +347,10 @@ void SkyRD::ReflectionData::update_reflection_data(int p_size, int p_mipmaps, bo tf.texture_type = RD::TEXTURE_TYPE_CUBE; tf.array_layers = 6; tf.mipmaps = p_low_quality ? 7 : mipmaps - 1; - tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; + tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; + if (RendererSceneRenderRD::get_singleton()->_render_buffers_can_be_storage()) { + tf.usage_bits |= RD::TEXTURE_USAGE_STORAGE_BIT; + } downsampled_radiance_cubemap = RD::get_singleton()->texture_create(tf, RD::TextureView()); RD::get_singleton()->set_resource_name(downsampled_radiance_cubemap, "downsampled radiance cubemap"); @@ -1085,8 +1088,8 @@ void SkyRD::setup_sky(RID p_env, Ref<RenderSceneBuffersRD> p_render_buffers, con sky->reflection.dirty = true; } + sky_scene_state.ubo.directional_light_count = 0; if (shader_data->uses_light) { - sky_scene_state.ubo.directional_light_count = 0; // Run through the list of lights in the scene and pick out the Directional Lights. // This can't be done in RenderSceneRenderRD::_setup lights because that needs to be called // after the depth prepass, but this runs before the depth prepass @@ -1627,7 +1630,10 @@ void SkyRD::update_dirty_skys() { tf.mipmaps = mipmaps; tf.width = w; tf.height = h; - tf.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; + tf.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT; + if (RendererSceneRenderRD::get_singleton()->_render_buffers_can_be_storage()) { + tf.usage_bits |= RD::TEXTURE_USAGE_STORAGE_BIT; + } sky->radiance = RD::get_singleton()->texture_create(tf, RD::TextureView()); diff --git a/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp index bd34d83ea3..ad63d3fd4b 100644 --- a/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp +++ b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp @@ -1365,8 +1365,8 @@ void RenderForwardClustered::_pre_opaque_render(RenderDataRD *p_render_data, boo } //cube shadows are rendered in their own way - for (uint32_t i = 0; i < p_render_data->cube_shadows.size(); i++) { - _render_shadow_pass(p_render_data->render_shadows[p_render_data->cube_shadows[i]].light, p_render_data->shadow_atlas, p_render_data->render_shadows[p_render_data->cube_shadows[i]].pass, p_render_data->render_shadows[p_render_data->cube_shadows[i]].instances, camera_plane, lod_distance_multiplier, p_render_data->scene_data->screen_mesh_lod_threshold, true, true, true, p_render_data->render_info); + for (const int &index : p_render_data->cube_shadows) { + _render_shadow_pass(p_render_data->render_shadows[index].light, p_render_data->shadow_atlas, p_render_data->render_shadows[index].pass, p_render_data->render_shadows[index].instances, camera_plane, lod_distance_multiplier, p_render_data->scene_data->screen_mesh_lod_threshold, true, true, true, p_render_data->render_info); } if (p_render_data->directional_shadows.size()) { @@ -1480,7 +1480,7 @@ void RenderForwardClustered::_pre_opaque_render(RenderDataRD *p_render_data, boo uint32_t directional_light_count = 0; uint32_t positional_light_count = 0; light_storage->update_light_buffers(p_render_data, *p_render_data->lights, p_render_data->scene_data->cam_transform, p_render_data->shadow_atlas, using_shadows, directional_light_count, positional_light_count, p_render_data->directional_light_soft_shadows); - texture_storage->update_decal_buffer(*p_render_data->decals, p_render_data->scene_data->cam_transform.affine_inverse()); + texture_storage->update_decal_buffer(*p_render_data->decals, p_render_data->scene_data->cam_transform); p_render_data->directional_light_count = directional_light_count; @@ -1707,7 +1707,7 @@ void RenderForwardClustered::_render_scene(RenderDataRD *p_render_data, const Co RD::get_singleton()->draw_command_end_label(); - if (rb.is_valid()) { + if (rb.is_valid() && !p_render_data->reflection_probe.is_valid()) { if (using_voxelgi) { depth_pass_mode = PASS_MODE_DEPTH_NORMAL_ROUGHNESS_VOXEL_GI; } else if (p_render_data->environment.is_valid()) { @@ -2415,8 +2415,7 @@ void RenderForwardClustered::_render_shadow_process() { void RenderForwardClustered::_render_shadow_end(uint32_t p_barrier) { RD::get_singleton()->draw_command_begin_label("Shadow Render"); - for (uint32_t i = 0; i < scene_state.shadow_passes.size(); i++) { - SceneState::ShadowPass &shadow_pass = scene_state.shadow_passes[i]; + for (SceneState::ShadowPass &shadow_pass : scene_state.shadow_passes) { RenderListParameters render_list_parameters(render_list[RENDER_LIST_SECONDARY].elements.ptr() + shadow_pass.element_from, render_list[RENDER_LIST_SECONDARY].element_info.ptr() + shadow_pass.element_from, shadow_pass.element_count, shadow_pass.flip_cull, shadow_pass.pass_mode, 0, true, false, shadow_pass.rp_uniform_set, false, Vector2(), shadow_pass.lod_distance_multiplier, shadow_pass.screen_mesh_lod_threshold, 1, shadow_pass.element_from, RD::BARRIER_MASK_NO_BARRIER); _render_list_with_threads(&render_list_parameters, shadow_pass.framebuffer, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD, shadow_pass.initial_depth_action, shadow_pass.final_depth_action, Vector<Color>(), 1.0, 0, shadow_pass.rect); } @@ -3376,9 +3375,7 @@ int RenderForwardClustered::sdfgi_get_pending_region_count(const Ref<RenderScene Ref<RendererRD::GI::SDFGI> sdfgi = rb->get_custom_data(RB_SCOPE_SDFGI); int dirty_count = 0; - for (uint32_t i = 0; i < sdfgi->cascades.size(); i++) { - const RendererRD::GI::SDFGI::Cascade &c = sdfgi->cascades[i]; - + for (const RendererRD::GI::SDFGI::Cascade &c : sdfgi->cascades) { if (c.dirty_regions == RendererRD::GI::SDFGI::Cascade::DIRTY_ALL) { dirty_count++; } else { @@ -4007,11 +4004,11 @@ RenderForwardClustered::~RenderForwardClustered() { RSG::light_storage->directional_shadow_atlas_set_size(0); { - for (uint32_t i = 0; i < scene_state.uniform_buffers.size(); i++) { - RD::get_singleton()->free(scene_state.uniform_buffers[i]); + for (const RID &rid : scene_state.uniform_buffers) { + RD::get_singleton()->free(rid); } - for (uint32_t i = 0; i < scene_state.implementation_uniform_buffers.size(); i++) { - RD::get_singleton()->free(scene_state.implementation_uniform_buffers[i]); + for (const RID &rid : scene_state.implementation_uniform_buffers) { + RD::get_singleton()->free(rid); } RD::get_singleton()->free(scene_state.lightmap_buffer); RD::get_singleton()->free(scene_state.lightmap_capture_buffer); diff --git a/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.cpp b/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.cpp index 7eabce2f79..412406c0b4 100644 --- a/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.cpp +++ b/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.cpp @@ -44,7 +44,6 @@ void SceneShaderForwardClustered::ShaderData::set_code(const String &p_code) { valid = false; ubo_size = 0; uniforms.clear(); - uses_screen_texture = false; if (code.is_empty()) { return; //just invalid, but no error @@ -73,9 +72,6 @@ void SceneShaderForwardClustered::ShaderData::set_code(const String &p_code) { uses_position = false; uses_sss = false; uses_transmittance = false; - uses_screen_texture = false; - uses_depth_texture = false; - uses_normal_texture = false; uses_time = false; writes_modelview_or_projection = false; uses_world_coordinates = false; @@ -120,9 +116,6 @@ void SceneShaderForwardClustered::ShaderData::set_code(const String &p_code) { actions.usage_flag_pointers["SSS_STRENGTH"] = &uses_sss; actions.usage_flag_pointers["SSS_TRANSMITTANCE_DEPTH"] = &uses_transmittance; - actions.usage_flag_pointers["SCREEN_TEXTURE"] = &uses_screen_texture; - actions.usage_flag_pointers["DEPTH_TEXTURE"] = &uses_depth_texture; - actions.usage_flag_pointers["NORMAL_ROUGHNESS_TEXTURE"] = &uses_normal_texture; actions.usage_flag_pointers["DISCARD"] = &uses_discard; actions.usage_flag_pointers["TIME"] = &uses_time; actions.usage_flag_pointers["ROUGHNESS"] = &uses_roughness; @@ -151,6 +144,9 @@ void SceneShaderForwardClustered::ShaderData::set_code(const String &p_code) { depth_test = DepthTest(depth_testi); cull_mode = Cull(cull_modei); uses_screen_texture_mipmaps = gen_code.uses_screen_texture_mipmaps; + uses_screen_texture = gen_code.uses_screen_texture; + uses_depth_texture = gen_code.uses_depth_texture; + uses_normal_texture = gen_code.uses_normal_roughness_texture; uses_vertex_time = gen_code.uses_vertex_time; uses_fragment_time = gen_code.uses_fragment_time; @@ -607,9 +603,6 @@ void SceneShaderForwardClustered::init(const String p_defines) { actions.renames["POINT_COORD"] = "gl_PointCoord"; actions.renames["INSTANCE_CUSTOM"] = "instance_custom"; actions.renames["SCREEN_UV"] = "screen_uv"; - actions.renames["SCREEN_TEXTURE"] = "color_buffer"; - actions.renames["DEPTH_TEXTURE"] = "depth_buffer"; - actions.renames["NORMAL_ROUGHNESS_TEXTURE"] = "normal_roughness_buffer"; actions.renames["DEPTH"] = "gl_FragDepth"; actions.renames["OUTPUT_IS_SRGB"] = "true"; actions.renames["FOG"] = "fog"; @@ -674,7 +667,6 @@ void SceneShaderForwardClustered::init(const String p_defines) { actions.usage_defines["SSS_STRENGTH"] = "#define ENABLE_SSS\n"; actions.usage_defines["SSS_TRANSMITTANCE_DEPTH"] = "#define ENABLE_TRANSMITTANCE\n"; actions.usage_defines["BACKLIGHT"] = "#define LIGHT_BACKLIGHT_USED\n"; - actions.usage_defines["SCREEN_TEXTURE"] = "#define SCREEN_TEXTURE_USED\n"; actions.usage_defines["SCREEN_UV"] = "#define SCREEN_UV_USED\n"; actions.usage_defines["DIFFUSE_LIGHT"] = "#define USE_LIGHT_SHADER_CODE\n"; @@ -707,10 +699,6 @@ void SceneShaderForwardClustered::init(const String p_defines) { actions.render_mode_defines["specular_schlick_ggx"] = "#define SPECULAR_SCHLICK_GGX\n"; - actions.custom_samplers["SCREEN_TEXTURE"] = "material_samplers[3]"; // linear filter with mipmaps - actions.custom_samplers["DEPTH_TEXTURE"] = "material_samplers[3]"; - actions.custom_samplers["NORMAL_ROUGHNESS_TEXTURE"] = "material_samplers[1]"; // linear filter - actions.render_mode_defines["specular_toon"] = "#define SPECULAR_TOON\n"; actions.render_mode_defines["specular_disabled"] = "#define SPECULAR_DISABLED\n"; actions.render_mode_defines["shadows_disabled"] = "#define SHADOWS_DISABLED\n"; diff --git a/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp b/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp index 78d29e2a41..25204f1abf 100644 --- a/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp +++ b/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp @@ -251,7 +251,7 @@ RID RenderForwardMobile::RenderBufferDataForwardMobile::get_color_fbs(Framebuffe Size2i target_size = render_buffers->get_target_size(); Size2i internal_size = render_buffers->get_internal_size(); - // can't do our blit pass if resolutions don't match + // can't do our blit pass if resolutions don't match, this should already have been checked. ERR_FAIL_COND_V(target_size != internal_size, RID()); // - opaque pass @@ -579,8 +579,8 @@ void RenderForwardMobile::_pre_opaque_render(RenderDataRD *p_render_data) { } //cube shadows are rendered in their own way - for (uint32_t i = 0; i < p_render_data->cube_shadows.size(); i++) { - _render_shadow_pass(p_render_data->render_shadows[p_render_data->cube_shadows[i]].light, p_render_data->shadow_atlas, p_render_data->render_shadows[p_render_data->cube_shadows[i]].pass, p_render_data->render_shadows[p_render_data->cube_shadows[i]].instances, camera_plane, lod_distance_multiplier, p_render_data->scene_data->screen_mesh_lod_threshold, true, true, true, p_render_data->render_info); + for (const int &index : p_render_data->cube_shadows) { + _render_shadow_pass(p_render_data->render_shadows[index].light, p_render_data->shadow_atlas, p_render_data->render_shadows[index].pass, p_render_data->render_shadows[index].instances, camera_plane, lod_distance_multiplier, p_render_data->scene_data->screen_mesh_lod_threshold, true, true, true, p_render_data->render_info); } if (p_render_data->directional_shadows.size()) { @@ -632,7 +632,7 @@ void RenderForwardMobile::_pre_opaque_render(RenderDataRD *p_render_data) { uint32_t directional_light_count = 0; uint32_t positional_light_count = 0; light_storage->update_light_buffers(p_render_data, *p_render_data->lights, p_render_data->scene_data->cam_transform, p_render_data->shadow_atlas, using_shadows, directional_light_count, positional_light_count, p_render_data->directional_light_soft_shadows); - texture_storage->update_decal_buffer(*p_render_data->decals, p_render_data->scene_data->cam_transform.affine_inverse()); + texture_storage->update_decal_buffer(*p_render_data->decals, p_render_data->scene_data->cam_transform); p_render_data->directional_light_count = directional_light_count; } @@ -715,20 +715,26 @@ void RenderForwardMobile::_render_scene(RenderDataRD *p_render_data, const Color // setup rendering to render buffer screen_size = p_render_data->render_buffers->get_internal_size(); - if (rb_data->get_color_fbs(RenderBufferDataForwardMobile::FB_CONFIG_FOUR_SUBPASSES).is_null()) { - // can't do blit subpass + if (rb->get_scaling_3d_mode() != RS::VIEWPORT_SCALING_3D_MODE_OFF) { + // can't do blit subpass because we're scaling using_subpass_post_process = false; } else if (p_render_data->environment.is_valid() && (environment_get_glow_enabled(p_render_data->environment) || RSG::camera_attributes->camera_attributes_uses_auto_exposure(p_render_data->camera_attributes) || RSG::camera_attributes->camera_attributes_uses_dof(p_render_data->camera_attributes))) { - // can't do blit subpass + // can't do blit subpass because we're using post processes using_subpass_post_process = false; } if (scene_state.used_screen_texture || scene_state.used_depth_texture) { - // can't use our last two subpasses + // can't use our last two subpasses because we're reading from screen texture or depth texture using_subpass_transparent = false; using_subpass_post_process = false; } + // We do this last because our get_color_fbs creates and caches the framebuffer if we need it. + if (using_subpass_post_process && rb_data->get_color_fbs(RenderBufferDataForwardMobile::FB_CONFIG_FOUR_SUBPASSES).is_null()) { + // can't do blit subpass because we don't have all subpasses + using_subpass_post_process = false; + } + if (using_subpass_post_process) { // all as subpasses framebuffer = rb_data->get_color_fbs(RenderBufferDataForwardMobile::FB_CONFIG_FOUR_SUBPASSES); @@ -1334,8 +1340,7 @@ void RenderForwardMobile::_render_shadow_process() { void RenderForwardMobile::_render_shadow_end(uint32_t p_barrier) { RD::get_singleton()->draw_command_begin_label("Shadow Render"); - for (uint32_t i = 0; i < scene_state.shadow_passes.size(); i++) { - SceneState::ShadowPass &shadow_pass = scene_state.shadow_passes[i]; + for (SceneState::ShadowPass &shadow_pass : scene_state.shadow_passes) { RenderListParameters render_list_parameters(render_list[RENDER_LIST_SECONDARY].elements.ptr() + shadow_pass.element_from, render_list[RENDER_LIST_SECONDARY].element_info.ptr() + shadow_pass.element_from, shadow_pass.element_count, shadow_pass.flip_cull, shadow_pass.pass_mode, shadow_pass.rp_uniform_set, 0, false, Vector2(), shadow_pass.lod_distance_multiplier, shadow_pass.screen_mesh_lod_threshold, 1, shadow_pass.element_from, RD::BARRIER_MASK_NO_BARRIER); _render_list_with_threads(&render_list_parameters, shadow_pass.framebuffer, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD, shadow_pass.initial_depth_action, shadow_pass.final_depth_action, Vector<Color>(), 1.0, 0, shadow_pass.rect); } @@ -2804,8 +2809,8 @@ RenderForwardMobile::~RenderForwardMobile() { } { - for (uint32_t i = 0; i < scene_state.uniform_buffers.size(); i++) { - RD::get_singleton()->free(scene_state.uniform_buffers[i]); + for (const RID &rid : scene_state.uniform_buffers) { + RD::get_singleton()->free(rid); } RD::get_singleton()->free(scene_state.lightmap_buffer); RD::get_singleton()->free(scene_state.lightmap_capture_buffer); diff --git a/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.cpp b/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.cpp index ee4c8001eb..3d1d78c63d 100644 --- a/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.cpp +++ b/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.cpp @@ -46,7 +46,6 @@ void SceneShaderForwardMobile::ShaderData::set_code(const String &p_code) { valid = false; ubo_size = 0; uniforms.clear(); - uses_screen_texture = false; if (code.is_empty()) { return; //just invalid, but no error @@ -74,9 +73,6 @@ void SceneShaderForwardMobile::ShaderData::set_code(const String &p_code) { uses_vertex = false; uses_sss = false; uses_transmittance = false; - uses_screen_texture = false; - uses_depth_texture = false; - uses_normal_texture = false; uses_time = false; writes_modelview_or_projection = false; uses_world_coordinates = false; @@ -121,9 +117,6 @@ void SceneShaderForwardMobile::ShaderData::set_code(const String &p_code) { // actions.usage_flag_pointers["SSS_STRENGTH"] = &uses_sss; // actions.usage_flag_pointers["SSS_TRANSMITTANCE_DEPTH"] = &uses_transmittance; - actions.usage_flag_pointers["SCREEN_TEXTURE"] = &uses_screen_texture; - actions.usage_flag_pointers["DEPTH_TEXTURE"] = &uses_depth_texture; - actions.usage_flag_pointers["NORMAL_ROUGHNESS_TEXTURE"] = &uses_normal_texture; actions.usage_flag_pointers["DISCARD"] = &uses_discard; actions.usage_flag_pointers["TIME"] = &uses_time; actions.usage_flag_pointers["ROUGHNESS"] = &uses_roughness; @@ -152,6 +145,10 @@ void SceneShaderForwardMobile::ShaderData::set_code(const String &p_code) { depth_test = DepthTest(depth_testi); uses_vertex_time = gen_code.uses_vertex_time; uses_fragment_time = gen_code.uses_fragment_time; + uses_screen_texture_mipmaps = gen_code.uses_screen_texture_mipmaps; + uses_screen_texture = gen_code.uses_screen_texture; + uses_depth_texture = gen_code.uses_depth_texture; + uses_normal_texture = gen_code.uses_normal_roughness_texture; #if 0 print_line("**compiling shader:"); @@ -498,9 +495,6 @@ void SceneShaderForwardMobile::init(const String p_defines) { actions.renames["POINT_COORD"] = "gl_PointCoord"; actions.renames["INSTANCE_CUSTOM"] = "instance_custom"; actions.renames["SCREEN_UV"] = "screen_uv"; - actions.renames["SCREEN_TEXTURE"] = "color_buffer"; - actions.renames["DEPTH_TEXTURE"] = "depth_buffer"; - actions.renames["NORMAL_ROUGHNESS_TEXTURE"] = "normal_roughness_buffer"; actions.renames["DEPTH"] = "gl_FragDepth"; actions.renames["OUTPUT_IS_SRGB"] = "true"; actions.renames["FOG"] = "fog"; @@ -565,7 +559,6 @@ void SceneShaderForwardMobile::init(const String p_defines) { actions.usage_defines["SSS_STRENGTH"] = "#define ENABLE_SSS\n"; actions.usage_defines["SSS_TRANSMITTANCE_DEPTH"] = "#define ENABLE_TRANSMITTANCE\n"; actions.usage_defines["BACKLIGHT"] = "#define LIGHT_BACKLIGHT_USED\n"; - actions.usage_defines["SCREEN_TEXTURE"] = "#define SCREEN_TEXTURE_USED\n"; actions.usage_defines["SCREEN_UV"] = "#define SCREEN_UV_USED\n"; actions.usage_defines["DIFFUSE_LIGHT"] = "#define USE_LIGHT_SHADER_CODE\n"; diff --git a/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.h b/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.h index 1f92697ecc..6f1f00cedc 100644 --- a/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.h +++ b/servers/rendering/renderer_rd/forward_mobile/scene_shader_forward_mobile.h @@ -128,6 +128,7 @@ public: bool uses_screen_texture = false; bool uses_depth_texture = false; bool uses_normal_texture = false; + bool uses_screen_texture_mipmaps = false; bool uses_time = false; bool uses_vertex_time = false; bool uses_fragment_time = false; diff --git a/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp b/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp index 0c6dcb553a..638fe44266 100644 --- a/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp @@ -455,7 +455,7 @@ void RendererCanvasRenderRD::_render_item(RD::DrawListID p_draw_list, RID p_rend light_count++; - if (light_count == MAX_LIGHTS_PER_ITEM) { + if (light_count == MAX_LIGHTS_PER_ITEM - 1) { break; } } @@ -525,10 +525,12 @@ void RendererCanvasRenderRD::_render_item(RD::DrawListID p_draw_list, RID p_rend if (rect->flags & CANVAS_RECT_FLIP_H) { src_rect.size.x *= -1; + push_constant.flags |= FLAGS_FLIP_H; } if (rect->flags & CANVAS_RECT_FLIP_V) { src_rect.size.y *= -1; + push_constant.flags |= FLAGS_FLIP_V; } if (rect->flags & CANVAS_RECT_TRANSPOSE) { @@ -2034,7 +2036,6 @@ void RendererCanvasRenderRD::CanvasShaderData::set_code(const String &p_code) { actions.render_mode_values["blend_premul_alpha"] = Pair<int *, int>(&blend_mode, BLEND_MODE_PMALPHA); actions.render_mode_values["blend_disabled"] = Pair<int *, int>(&blend_mode, BLEND_MODE_DISABLED); - actions.usage_flag_pointers["SCREEN_TEXTURE"] = &uses_screen_texture; actions.usage_flag_pointers["texture_sdf"] = &uses_sdf; actions.usage_flag_pointers["TIME"] = &uses_time; @@ -2046,6 +2047,7 @@ void RendererCanvasRenderRD::CanvasShaderData::set_code(const String &p_code) { ERR_FAIL_COND_MSG(err != OK, "Shader compilation failed."); uses_screen_texture_mipmaps = gen_code.uses_screen_texture_mipmaps; + uses_screen_texture = gen_code.uses_screen_texture; if (version.is_null()) { version = canvas_singleton->shader.canvas_shader.version_create(); @@ -2423,7 +2425,6 @@ RendererCanvasRenderRD::RendererCanvasRenderRD() { actions.renames["SPECULAR_SHININESS_TEXTURE"] = "specular_texture"; actions.renames["SPECULAR_SHININESS"] = "specular_shininess"; actions.renames["SCREEN_UV"] = "screen_uv"; - actions.renames["SCREEN_TEXTURE"] = "screen_texture"; actions.renames["SCREEN_PIXEL_SIZE"] = "canvas_data.screen_pixel_size"; actions.renames["FRAGCOORD"] = "gl_FragCoord"; actions.renames["POINT_COORD"] = "gl_PointCoord"; @@ -2444,7 +2445,6 @@ RendererCanvasRenderRD::RendererCanvasRenderRD() { actions.renames["screen_uv_to_sdf"] = "screen_uv_to_sdf"; actions.usage_defines["COLOR"] = "#define COLOR_USED\n"; - actions.usage_defines["SCREEN_TEXTURE"] = "#define SCREEN_TEXTURE_USED\n"; actions.usage_defines["SCREEN_UV"] = "#define SCREEN_UV_USED\n"; actions.usage_defines["SCREEN_PIXEL_SIZE"] = "@SCREEN_UV"; actions.usage_defines["NORMAL"] = "#define NORMAL_USED\n"; @@ -2459,7 +2459,6 @@ RendererCanvasRenderRD::RendererCanvasRenderRD() { actions.custom_samplers["TEXTURE"] = "texture_sampler"; actions.custom_samplers["NORMAL_TEXTURE"] = "texture_sampler"; actions.custom_samplers["SPECULAR_SHININESS_TEXTURE"] = "texture_sampler"; - actions.custom_samplers["SCREEN_TEXTURE"] = "material_samplers[3]"; //mipmap and filter for screen texture actions.sampler_array_name = "material_samplers"; actions.base_texture_binding_index = 1; actions.texture_layout_set = MATERIAL_UNIFORM_SET; @@ -2632,8 +2631,10 @@ RendererCanvasRenderRD::RendererCanvasRenderRD() { shader_type canvas_item; +uniform sampler2D screen_texture : hint_screen_texture, repeat_disable, filter_nearest; + void fragment() { - vec4 c = textureLod(SCREEN_TEXTURE, SCREEN_UV, 0.0); + vec4 c = textureLod(screen_texture, SCREEN_UV, 0.0); if (c.a > 0.0001) { c.rgb /= c.a; @@ -2657,8 +2658,10 @@ void fragment() { shader_type canvas_item; +uniform sampler2D screen_texture : hint_screen_texture, repeat_disable, filter_nearest; + void fragment() { - vec4 c = textureLod(SCREEN_TEXTURE, SCREEN_UV, 0.0); + vec4 c = textureLod(screen_texture, SCREEN_UV, 0.0); COLOR.rgb = c.rgb; } )"); diff --git a/servers/rendering/renderer_rd/renderer_canvas_render_rd.h b/servers/rendering/renderer_rd/renderer_canvas_render_rd.h index 266083fc49..7dea4a1a65 100644 --- a/servers/rendering/renderer_rd/renderer_canvas_render_rd.h +++ b/servers/rendering/renderer_rd/renderer_canvas_render_rd.h @@ -86,6 +86,9 @@ class RendererCanvasRenderRD : public RendererCanvasRender { FLAGS_USE_MSDF = (1 << 28), FLAGS_USE_LCD = (1 << 29), + + FLAGS_FLIP_H = (1 << 30), + FLAGS_FLIP_V = (1 << 31), }; enum { diff --git a/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp b/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp index 1d45db8eaf..2a87c4fa8d 100644 --- a/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp @@ -539,7 +539,7 @@ void RendererSceneRenderRD::_render_buffers_post_process_and_tonemap(const Rende tonemap.view_count = rb->get_view_count(); RID dest_fb; - if (fsr && can_use_effects && (internal_size.x != target_size.x || internal_size.y != target_size.y)) { + if (fsr && can_use_effects && rb->get_scaling_3d_mode() == RS::VIEWPORT_SCALING_3D_MODE_FSR) { // If we use FSR to upscale we need to write our result into an intermediate buffer. // Note that this is cached so we only create the texture the first time. RID dest_texture = rb->create_texture(SNAME("Tonemapper"), SNAME("destination"), _render_buffers_get_color_format(), RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT); @@ -556,10 +556,7 @@ void RendererSceneRenderRD::_render_buffers_post_process_and_tonemap(const Rende RD::get_singleton()->draw_command_end_label(); } - if (fsr && can_use_effects && (internal_size.x != target_size.x || internal_size.y != target_size.y)) { - // TODO Investigate? Does this work? We never write into our render target and we've already done so up above in our tonemapper. - // I think FSR should either work before our tonemapper or as an alternative of our tonemapper. - + if (fsr && can_use_effects && rb->get_scaling_3d_mode() == RS::VIEWPORT_SCALING_3D_MODE_FSR) { RD::get_singleton()->draw_command_begin_label("FSR 1.0 Upscale"); for (uint32_t v = 0; v < rb->get_view_count(); v++) { diff --git a/servers/rendering/renderer_rd/shader_rd.cpp b/servers/rendering/renderer_rd/shader_rd.cpp index 2d5263a3e2..533a912a34 100644 --- a/servers/rendering/renderer_rd/shader_rd.cpp +++ b/servers/rendering/renderer_rd/shader_rd.cpp @@ -165,8 +165,7 @@ void ShaderRD::_clear_version(Version *p_version) { } void ShaderRD::_build_variant_code(StringBuilder &builder, uint32_t p_variant, const Version *p_version, const StageTemplate &p_template) { - for (uint32_t i = 0; i < p_template.chunks.size(); i++) { - const StageTemplate::Chunk &chunk = p_template.chunks[i]; + for (const StageTemplate::Chunk &chunk : p_template.chunks) { switch (chunk.type) { case StageTemplate::Chunk::TYPE_VERSION_DEFINES: { builder.append("\n"); //make sure defines begin at newline diff --git a/servers/rendering/renderer_rd/shaders/canvas.glsl b/servers/rendering/renderer_rd/shaders/canvas.glsl index eb5f68849e..1fb8b28b15 100644 --- a/servers/rendering/renderer_rd/shaders/canvas.glsl +++ b/servers/rendering/renderer_rd/shaders/canvas.glsl @@ -502,6 +502,12 @@ void main() { 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); + if (bool(draw_data.flags & FLAGS_FLIP_H)) { + normal.x = -normal.x; + } + if (bool(draw_data.flags & FLAGS_FLIP_V)) { + normal.y = -normal.y; + } normal.z = sqrt(1.0 - dot(normal.xy, normal.xy)); normal_used = true; } else { diff --git a/servers/rendering/renderer_rd/shaders/canvas_uniforms_inc.glsl b/servers/rendering/renderer_rd/shaders/canvas_uniforms_inc.glsl index 1b627a3e81..a904f4e0a6 100644 --- a/servers/rendering/renderer_rd/shaders/canvas_uniforms_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/canvas_uniforms_inc.glsl @@ -27,6 +27,9 @@ #define FLAGS_USE_MSDF (1 << 28) #define FLAGS_USE_LCD (1 << 29) +#define FLAGS_FLIP_H (1 << 30) +#define FLAGS_FLIP_V (1 << 31) + #define SAMPLER_NEAREST_CLAMP 0 #define SAMPLER_LINEAR_CLAMP 1 #define SAMPLER_NEAREST_WITH_MIPMAPS_CLAMP 2 @@ -134,7 +137,7 @@ layout(set = 0, binding = 4) uniform texture2D shadow_atlas_texture; layout(set = 0, binding = 5) uniform sampler shadow_sampler; -layout(set = 0, binding = 6) uniform texture2D screen_texture; +layout(set = 0, binding = 6) uniform texture2D color_buffer; layout(set = 0, binding = 7) uniform texture2D sdf_texture; layout(set = 0, binding = 8) uniform sampler material_samplers[12]; diff --git a/servers/rendering/renderer_rd/shaders/effects/ss_effects_downsample.glsl b/servers/rendering/renderer_rd/shaders/effects/ss_effects_downsample.glsl index 134aae5ce7..b1ff46dd3b 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ss_effects_downsample.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ss_effects_downsample.glsl @@ -56,7 +56,7 @@ vec4 screen_space_to_view_space_depth(vec4 p_depth) { float depth_linearize_mul = params.z_near; float depth_linearize_add = params.z_far; - // Optimised version of "-cameraClipNear / (cameraClipFar - projDepth * (cameraClipFar - cameraClipNear)) * cameraClipFar" + // Optimized version of "-cameraClipNear / (cameraClipFar - projDepth * (cameraClipFar - cameraClipNear)) * cameraClipFar" // Set your depth_linearize_mul and depth_linearize_add to: // depth_linearize_mul = ( cameraClipFar * cameraClipNear) / ( cameraClipFar - cameraClipNear ); diff --git a/servers/rendering/renderer_rd/shaders/effects/ssao.glsl b/servers/rendering/renderer_rd/shaders/effects/ssao.glsl index 2a87e273bc..ffaa6872c9 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ssao.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ssao.glsl @@ -221,7 +221,7 @@ void SSAOTap(const int p_quality_level, inout float r_obscurance_sum, inout floa // snap to pixel center (more correct obscurance math, avoids artifacts) sample_offset = round(sample_offset); - // calculate MIP based on the sample distance from the centre, similar to as described + // calculate MIP based on the sample distance from the center, similar to as described // in http://graphics.cs.williams.edu/papers/SAOHPG12/. float mip_level = (p_quality_level < SSAO_DEPTH_MIPS_ENABLE_AT_QUALITY_PRESET) ? (0) : (sample_pow_2_len + p_mip_offset); @@ -259,7 +259,7 @@ void generate_SSAO_shadows_internal(out float r_shadow_term, out vec4 r_edges, o // get this pixel's viewspace depth pix_z = valuesUL.y; - // get left right top bottom neighbouring pixels for edge detection (gets compiled out on quality_level == 0) + // get left right top bottom neighboring pixels for edge detection (gets compiled out on quality_level == 0) pix_left_z = valuesUL.x; pix_top_z = valuesUL.z; pix_right_z = valuesBR.z; @@ -304,7 +304,7 @@ void generate_SSAO_shadows_internal(out float r_shadow_term, out vec4 r_edges, o float obscurance_sum = 0.0; float weight_sum = 0.0; - // edge mask for between this and left/right/top/bottom neighbour pixels - not used in quality level 0 so initialize to "no edge" (1 is no edge, 0 is edge) + // edge mask for between this and left/right/top/bottom neighbor pixels - not used in quality level 0 so initialize to "no edge" (1 is no edge, 0 is edge) vec4 edgesLRTB = vec4(1.0, 1.0, 1.0, 1.0); // Move center pixel slightly towards camera to avoid imprecision artifacts due to using of 16bit depth buffer; a lot smaller offsets needed when using 32bit floats @@ -318,7 +318,7 @@ void generate_SSAO_shadows_internal(out float r_shadow_term, out vec4 r_edges, o if (!p_adaptive_base && (p_quality_level >= SSAO_DETAIL_AO_ENABLE_AT_QUALITY_PRESET)) { // disable in case of quality level 4 (reference) if (p_quality_level != 4) { - //approximate neighbouring pixels positions (actually just deltas or "positions - pix_center_pos" ) + //approximate neighboring pixels positions (actually just deltas or "positions - pix_center_pos" ) vec3 normalized_viewspace_dir = vec3(pix_center_pos.xy / pix_center_pos.zz, 1.0); vec3 pixel_left_delta = vec3(-pixel_size_at_center.x, 0.0, 0.0) + normalized_viewspace_dir * (pix_left_z - pix_center_pos.z); vec3 pixel_right_delta = vec3(+pixel_size_at_center.x, 0.0, 0.0) + normalized_viewspace_dir * (pix_right_z - pix_center_pos.z); diff --git a/servers/rendering/renderer_rd/shaders/effects/ssao_importance_map.glsl b/servers/rendering/renderer_rd/shaders/effects/ssao_importance_map.glsl index 04f98964e8..d234ab4417 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ssao_importance_map.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ssao_importance_map.glsl @@ -80,7 +80,7 @@ void main() { #ifdef PROCESS_MAPA vec2 uv = (vec2(ssC) + 0.5f) * params.half_screen_pixel_size * 2.0; - float centre = textureLod(source_importance, uv, 0.0).x; + float center = textureLod(source_importance, uv, 0.0).x; vec2 half_pixel = params.half_screen_pixel_size; @@ -98,7 +98,7 @@ void main() { #ifdef PROCESS_MAPB vec2 uv = (vec2(ssC) + 0.5f) * params.half_screen_pixel_size * 2.0; - float centre = textureLod(source_importance, uv, 0.0).x; + float center = textureLod(source_importance, uv, 0.0).x; vec2 half_pixel = params.half_screen_pixel_size; diff --git a/servers/rendering/renderer_rd/shaders/effects/ssao_interleave.glsl b/servers/rendering/renderer_rd/shaders/effects/ssao_interleave.glsl index f6a9a92fac..45cc62d361 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ssao_interleave.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ssao_interleave.glsl @@ -60,8 +60,8 @@ void main() { int mx = int(pix_pos.x % 2); int my = int(pix_pos.y % 2); int index_center = mx + my * 2; // center index - int index_horizontal = (1 - mx) + my * 2; // neighbouring, horizontal - int index_vertical = mx + (1 - my) * 2; // neighbouring, vertical + int index_horizontal = (1 - mx) + my * 2; // neighboring, horizontal + int index_vertical = mx + (1 - my) * 2; // neighboring, vertical int index_diagonal = (1 - mx) + (1 - my) * 2; // diagonal vec2 center_val = texelFetch(source_texture, ivec3(pix_pos / uvec2(params.size_modifier), index_center), 0).xy; diff --git a/servers/rendering/renderer_rd/shaders/effects/ssil.glsl b/servers/rendering/renderer_rd/shaders/effects/ssil.glsl index 513791dfbf..de7b97953f 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ssil.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ssil.glsl @@ -234,7 +234,7 @@ void SSILTap(const int p_quality_level, inout vec3 r_color_sum, inout float r_ob // snap to pixel center (more correct obscurance math, avoids artifacts) sample_offset = round(sample_offset); - // calculate MIP based on the sample distance from the centre, similar to as described + // calculate MIP based on the sample distance from the center, similar to as described // in http://graphics.cs.williams.edu/papers/SAOHPG12/. float mip_level = (p_quality_level < SSIL_DEPTH_MIPS_ENABLE_AT_QUALITY_PRESET) ? (0) : (sample_pow_2_len + p_mip_offset); @@ -272,7 +272,7 @@ void generate_SSIL(out vec3 r_color, out vec4 r_edges, out float r_obscurance, o // get this pixel's viewspace depth pix_z = valuesUL.y; - // get left right top bottom neighbouring pixels for edge detection (gets compiled out on quality_level == 0) + // get left right top bottom neighboring pixels for edge detection (gets compiled out on quality_level == 0) pix_left_z = valuesUL.x; pix_top_z = valuesUL.z; pix_right_z = valuesBR.z; @@ -318,7 +318,7 @@ void generate_SSIL(out vec3 r_color, out vec4 r_edges, out float r_obscurance, o float obscurance_sum = 0.0; float weight_sum = 0.0; - // edge mask for between this and left/right/top/bottom neighbour pixels - not used in quality level 0 so initialize to "no edge" (1 is no edge, 0 is edge) + // edge mask for between this and left/right/top/bottom neighbor pixels - not used in quality level 0 so initialize to "no edge" (1 is no edge, 0 is edge) vec4 edgesLRTB = vec4(1.0, 1.0, 1.0, 1.0); // Move center pixel slightly towards camera to avoid imprecision artifacts due to using of 16bit depth buffer; a lot smaller offsets needed when using 32bit floats diff --git a/servers/rendering/renderer_rd/shaders/effects/ssil_blur.glsl b/servers/rendering/renderer_rd/shaders/effects/ssil_blur.glsl index 47c56571f6..f48e6c4341 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ssil_blur.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ssil_blur.glsl @@ -124,14 +124,14 @@ void main() { vec2 uv = (vec2(gl_GlobalInvocationID.xy) + vec2(0.5, 0.5)) * params.half_screen_pixel_size; - vec4 centre = textureLod(source_ssil, uv, 0.0); + vec4 center = textureLod(source_ssil, uv, 0.0); vec4 value = textureLod(source_ssil, vec2(uv + vec2(-half_pixel.x * 3, -half_pixel.y)), 0.0) * 0.2; value += textureLod(source_ssil, vec2(uv + vec2(+half_pixel.x, -half_pixel.y * 3)), 0.0) * 0.2; value += textureLod(source_ssil, vec2(uv + vec2(-half_pixel.x, +half_pixel.y * 3)), 0.0) * 0.2; value += textureLod(source_ssil, vec2(uv + vec2(+half_pixel.x * 3, +half_pixel.y)), 0.0) * 0.2; - vec4 sampled = value + centre * 0.2; + vec4 sampled = value + center * 0.2; #else #ifdef MODE_SMART diff --git a/servers/rendering/renderer_rd/shaders/effects/ssil_importance_map.glsl b/servers/rendering/renderer_rd/shaders/effects/ssil_importance_map.glsl index 6b6b02739d..193e3458ab 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ssil_importance_map.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ssil_importance_map.glsl @@ -82,7 +82,7 @@ void main() { #ifdef PROCESS_MAPA vec2 uv = (vec2(ssC) + 0.5) * params.half_screen_pixel_size * 2.0; - float centre = textureLod(source_importance, uv, 0.0).x; + float center = textureLod(source_importance, uv, 0.0).x; vec2 half_pixel = params.half_screen_pixel_size; @@ -100,7 +100,7 @@ void main() { #ifdef PROCESS_MAPB vec2 uv = (vec2(ssC) + 0.5f) * params.half_screen_pixel_size * 2.0; - float centre = textureLod(source_importance, uv, 0.0).x; + float center = textureLod(source_importance, uv, 0.0).x; vec2 half_pixel = params.half_screen_pixel_size; diff --git a/servers/rendering/renderer_rd/shaders/effects/ssil_interleave.glsl b/servers/rendering/renderer_rd/shaders/effects/ssil_interleave.glsl index 9e86ac0cf0..ed85b8ee4c 100644 --- a/servers/rendering/renderer_rd/shaders/effects/ssil_interleave.glsl +++ b/servers/rendering/renderer_rd/shaders/effects/ssil_interleave.glsl @@ -62,8 +62,8 @@ void main() { int mx = int(pix_pos.x % 2); int my = int(pix_pos.y % 2); int index_center = mx + my * 2; // center index - int index_horizontal = (1 - mx) + my * 2; // neighbouring, horizontal - int index_vertical = mx + (1 - my) * 2; // neighbouring, vertical + int index_horizontal = (1 - mx) + my * 2; // neighboring, horizontal + int index_vertical = mx + (1 - my) * 2; // neighboring, vertical int index_diagonal = (1 - mx) + (1 - my) * 2; // diagonal vec4 color = texelFetch(source_texture, ivec3(pix_pos / uvec2(params.size_modifier), index_center), 0); diff --git a/servers/rendering/renderer_rd/shaders/environment/sdfgi_direct_light.glsl b/servers/rendering/renderer_rd/shaders/environment/sdfgi_direct_light.glsl index 9f7449b8aa..06709f65d3 100644 --- a/servers/rendering/renderer_rd/shaders/environment/sdfgi_direct_light.glsl +++ b/servers/rendering/renderer_rd/shaders/environment/sdfgi_direct_light.glsl @@ -24,7 +24,7 @@ struct ProcessVoxel { uint albedo; // rgb bits 0-15 albedo, bits 16-21 are normal bits (set if geometry exists toward that side), extra 11 bits for neighbors. uint light; // rgbe8985 encoded total saved light, extra 2 bits for neighbors. uint light_aniso; // 55555 light anisotropy, extra 2 bits for neighbors. - //total neighbours: 26 + //total neighbors: 26 }; #ifdef MODE_PROCESS_STATIC @@ -443,10 +443,10 @@ void main() { imageStore(dst_aniso1, positioni, vec4(aniso1, 0.0, 0.0)); imageStore(dst_light, positioni, uvec4(light_total_rgbe)); - //also fill neighbours, so light interpolation during the indirect pass works + //also fill neighbors, so light interpolation during the indirect pass works - //recover the neighbour list from the leftover bits - uint neighbours = (voxel_albedo >> 21) | ((voxel_position >> 21) << 11) | ((process_voxels.data[voxel_index].light >> 30) << 22) | ((process_voxels.data[voxel_index].light_aniso >> 30) << 24); + //recover the neighbor list from the leftover bits + uint neighbors = (voxel_albedo >> 21) | ((voxel_position >> 21) << 11) | ((process_voxels.data[voxel_index].light >> 30) << 22) | ((process_voxels.data[voxel_index].light_aniso >> 30) << 24); const uint max_neighbours = 26; const ivec3 neighbour_positions[max_neighbours] = ivec3[]( @@ -478,7 +478,7 @@ void main() { ivec3(1, 1, 1)); for (uint i = 0; i < max_neighbours; i++) { - if (bool(neighbours & (1 << i))) { + if (bool(neighbors & (1 << i))) { ivec3 neighbour_pos = positioni + neighbour_positions[i]; imageStore(dst_light, neighbour_pos, uvec4(light_total_rgbe)); imageStore(dst_aniso0, neighbour_pos, aniso0); diff --git a/servers/rendering/renderer_rd/shaders/environment/sdfgi_preprocess.glsl b/servers/rendering/renderer_rd/shaders/environment/sdfgi_preprocess.glsl index bce98f4054..dd35ae3b73 100644 --- a/servers/rendering/renderer_rd/shaders/environment/sdfgi_preprocess.glsl +++ b/servers/rendering/renderer_rd/shaders/environment/sdfgi_preprocess.glsl @@ -102,10 +102,10 @@ dispatch_data; struct ProcessVoxel { uint position; // xyz 7 bit packed, extra 11 bits for neighbors. - uint albedo; //rgb bits 0-15 albedo, bits 16-21 are normal bits (set if geometry exists toward that side), extra 11 bits for neighbours - 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 + uint albedo; //rgb bits 0-15 albedo, bits 16-21 are normal bits (set if geometry exists toward that side), extra 11 bits for neighbors + uint light; //rgbe8985 encoded total saved light, extra 2 bits for neighbors + uint light_aniso; //55555 light anisotropy, extra 2 bits for neighbors + //total neighbors: 26 }; layout(set = 0, binding = 11, std430) restrict buffer writeonly ProcessVoxels { @@ -135,10 +135,10 @@ dispatch_data; struct ProcessVoxel { uint position; // xyz 7 bit packed, extra 11 bits for neighbors. - uint albedo; //rgb bits 0-15 albedo, bits 16-21 are normal bits (set if geometry exists toward that side), extra 11 bits for neighbours - 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 + uint albedo; //rgb bits 0-15 albedo, bits 16-21 are normal bits (set if geometry exists toward that side), extra 11 bits for neighbors + uint light; //rgbe8985 encoded total saved light, extra 2 bits for neighbors + uint light_aniso; //55555 light anisotropy, extra 2 bits for neighbors + //total neighbors: 26 }; layout(set = 0, binding = 6, std430) restrict buffer readonly ProcessVoxels { @@ -1016,14 +1016,14 @@ void main() { store_positions[index].albedo = rgb >> 1; //store as it comes (555) to avoid precision loss (and move away the alpha bit) store_positions[index].albedo |= (facing & 0x3F) << 15; // store facing in bits 15-21 - store_positions[index].albedo |= neighbour_bits << 21; //store lower 11 bits of neighbours with remaining albedo - store_positions[index].position |= (neighbour_bits >> 11) << 21; //store 11 bits more of neighbours with position + store_positions[index].albedo |= neighbour_bits << 21; //store lower 11 bits of neighbors with remaining albedo + store_positions[index].position |= (neighbour_bits >> 11) << 21; //store 11 bits more of neighbors with position store_positions[index].light = imageLoad(src_light, pos).r; store_positions[index].light_aniso = imageLoad(src_light_aniso, pos).r; - //add neighbours - store_positions[index].light |= (neighbour_bits >> 22) << 30; //store 2 bits more of neighbours with light - store_positions[index].light_aniso |= (neighbour_bits >> 24) << 30; //store 2 bits more of neighbours with aniso + //add neighbors + store_positions[index].light |= (neighbour_bits >> 22) << 30; //store 2 bits more of neighbors with light + store_positions[index].light_aniso |= (neighbour_bits >> 24) << 30; //store 2 bits more of neighbors with aniso } groupMemoryBarrier(); diff --git a/servers/rendering/renderer_rd/shaders/forward_clustered/scene_forward_clustered.glsl b/servers/rendering/renderer_rd/shaders/forward_clustered/scene_forward_clustered.glsl index d32e6d717f..91c2000c1a 100644 --- a/servers/rendering/renderer_rd/shaders/forward_clustered/scene_forward_clustered.glsl +++ b/servers/rendering/renderer_rd/shaders/forward_clustered/scene_forward_clustered.glsl @@ -1305,24 +1305,26 @@ void fragment_shader(in SceneData scene_data) { } if (sc_use_forward_gi && bool(instances.data[instance_index].flags & INSTANCE_FLAGS_USE_VOXEL_GI)) { // process voxel_gi_instances - uint index1 = instances.data[instance_index].gi_offset & 0xFFFF; - vec3 ref_vec = normalize(reflect(-view, normal)); - ref_vec = mix(ref_vec, normal, roughness * roughness); + // Make vertex orientation the world one, but still align to camera. + vec3 cam_pos = mat3(scene_data.inv_view_matrix) * vertex; + vec3 cam_normal = mat3(scene_data.inv_view_matrix) * normal; + vec3 ref_vec = mat3(scene_data.inv_view_matrix) * normalize(reflect(-view, normal)); + //find arbitrary tangent and bitangent, then build a matrix - vec3 v0 = abs(normal.z) < 0.999 ? vec3(0.0, 0.0, 1.0) : vec3(0.0, 1.0, 0.0); - vec3 tangent = normalize(cross(v0, normal)); - vec3 bitangent = normalize(cross(tangent, normal)); - mat3 normal_mat = mat3(tangent, bitangent, normal); + vec3 v0 = abs(cam_normal.z) < 0.999 ? vec3(0.0, 0.0, 1.0) : vec3(0.0, 1.0, 0.0); + vec3 tangent = normalize(cross(v0, cam_normal)); + vec3 bitangent = normalize(cross(tangent, cam_normal)); + mat3 normal_mat = mat3(tangent, bitangent, cam_normal); vec4 amb_accum = vec4(0.0); vec4 spec_accum = vec4(0.0); - voxel_gi_compute(index1, vertex, normal, ref_vec, normal_mat, roughness * roughness, ambient_light, specular_light, spec_accum, amb_accum); + voxel_gi_compute(index1, cam_pos, cam_normal, ref_vec, normal_mat, roughness * roughness, ambient_light, specular_light, spec_accum, amb_accum); uint index2 = instances.data[instance_index].gi_offset >> 16; if (index2 != 0xFFFF) { - voxel_gi_compute(index2, vertex, normal, ref_vec, normal_mat, roughness * roughness, ambient_light, specular_light, spec_accum, amb_accum); + voxel_gi_compute(index2, cam_pos, cam_normal, ref_vec, normal_mat, roughness * roughness, ambient_light, specular_light, spec_accum, amb_accum); } if (amb_accum.a > 0.0) { @@ -1351,8 +1353,8 @@ void fragment_shader(in SceneData scene_data) { #endif // USE_MULTIVIEW for (int i = 0; i < 4; i++) { - const vec2 neighbours[4] = vec2[](vec2(-1, 0), vec2(1, 0), vec2(0, -1), vec2(0, 1)); - vec2 neighbour_coord = base_coord + neighbours[i] * scene_data.screen_pixel_size; + const vec2 neighbors[4] = vec2[](vec2(-1, 0), vec2(1, 0), vec2(0, -1), vec2(0, 1)); + vec2 neighbour_coord = base_coord + neighbors[i] * scene_data.screen_pixel_size; #ifdef USE_MULTIVIEW float neighbour_ang = dot(normal, textureLod(sampler2DArray(normal_roughness_buffer, material_samplers[SAMPLER_LINEAR_CLAMP]), vec3(neighbour_coord, ViewIndex), 0.0).xyz * 2.0 - 1.0); #else // USE_MULTIVIEW @@ -1798,7 +1800,7 @@ void fragment_shader(in SceneData scene_data) { shadow = float(shadow1 >> ((i - 4u) * 8u) & 0xFFu) / 255.0; } - shadow = shadow * directional_lights.data[i].shadow_opacity + 1.0 - directional_lights.data[i].shadow_opacity; + shadow = mix(1.0, shadow, directional_lights.data[i].shadow_opacity); #endif blur_shadow(shadow); @@ -2082,7 +2084,7 @@ void fragment_shader(in SceneData scene_data) { float sRed = floor((cRed / pow(2.0f, exps - B - N)) + 0.5f); float sGreen = floor((cGreen / pow(2.0f, exps - B - N)) + 0.5f); float sBlue = floor((cBlue / pow(2.0f, exps - B - N)) + 0.5f); - //store as 8985 to have 2 extra neighbour bits + //store as 8985 to have 2 extra neighbor bits uint light_rgbe = ((uint(sRed) & 0x1FFu) >> 1) | ((uint(sGreen) & 0x1FFu) << 8) | (((uint(sBlue) & 0x1FFu) >> 1) << 17) | ((uint(exps) & 0x1Fu) << 25); imageStore(emission_grid, grid_pos, uvec4(light_rgbe)); diff --git a/servers/rendering/renderer_rd/shaders/forward_mobile/scene_forward_mobile.glsl b/servers/rendering/renderer_rd/shaders/forward_mobile/scene_forward_mobile.glsl index 5e64d4e651..2966a2ff65 100644 --- a/servers/rendering/renderer_rd/shaders/forward_mobile/scene_forward_mobile.glsl +++ b/servers/rendering/renderer_rd/shaders/forward_mobile/scene_forward_mobile.glsl @@ -1515,6 +1515,8 @@ void main() { } else { shadow = float(shadow1 >> ((i - 4) * 8) & 0xFF) / 255.0; } + + shadow = mix(1.0, shadow, directional_lights.data[i].shadow_opacity); #endif blur_shadow(shadow); diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_lights_inc.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_lights_inc.glsl index b30b0c8169..9dda62c28d 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_lights_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_lights_inc.glsl @@ -794,8 +794,13 @@ void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v float light_length = length(light_rel_vec); float spot_attenuation = get_omni_attenuation(light_length, spot_lights.data[idx].inv_radius, spot_lights.data[idx].attenuation); vec3 spot_dir = spot_lights.data[idx].direction; - float scos = max(dot(-normalize(light_rel_vec), spot_dir), spot_lights.data[idx].cone_angle); - float spot_rim = max(0.0001, (1.0 - scos) / (1.0 - spot_lights.data[idx].cone_angle)); + + // This conversion to a highp float is crucial to prevent light leaking + // due to precision errors in the following calculations (cone angle is mediump). + highp float cone_angle = spot_lights.data[idx].cone_angle; + float scos = max(dot(-normalize(light_rel_vec), spot_dir), cone_angle); + float spot_rim = max(0.0001, (1.0 - scos) / (1.0 - cone_angle)); + spot_attenuation *= 1.0 - pow(spot_rim, spot_lights.data[idx].cone_attenuation); float light_attenuation = spot_attenuation; vec3 color = spot_lights.data[idx].color; diff --git a/servers/rendering/renderer_rd/storage_rd/light_storage.cpp b/servers/rendering/renderer_rd/storage_rd/light_storage.cpp index 673fc25595..33e35a7a64 100644 --- a/servers/rendering/renderer_rd/storage_rd/light_storage.cpp +++ b/servers/rendering/renderer_rd/storage_rd/light_storage.cpp @@ -568,8 +568,6 @@ void LightStorage::update_light_buffers(RenderDataRD *p_render_data, const Paged r_directional_light_count = 0; r_positional_light_count = 0; - Plane camera_plane(-p_camera_transform.basis.get_column(Vector3::AXIS_Z).normalized(), p_camera_transform.origin); - omni_light_count = 0; spot_light_count = 0; @@ -720,7 +718,7 @@ void LightStorage::update_light_buffers(RenderDataRD *p_render_data, const Paged } Transform3D light_transform = light_instance->transform; - const real_t distance = camera_plane.distance_to(light_transform.origin); + const real_t distance = p_camera_transform.origin.distance_to(light_transform.origin); if (light->distance_fade) { const float fade_begin = light->distance_fade_begin; @@ -745,7 +743,7 @@ void LightStorage::update_light_buffers(RenderDataRD *p_render_data, const Paged } Transform3D light_transform = light_instance->transform; - const real_t distance = camera_plane.distance_to(light_transform.origin); + const real_t distance = p_camera_transform.origin.distance_to(light_transform.origin); if (light->distance_fade) { const float fade_begin = light->distance_fade_begin; @@ -787,6 +785,7 @@ void LightStorage::update_light_buffers(RenderDataRD *p_render_data, const Paged RS::LightType type = (i < omni_light_count) ? RS::LIGHT_OMNI : RS::LIGHT_SPOT; LightInstance *light_instance = (i < omni_light_count) ? omni_light_sort[index].light_instance : spot_light_sort[index].light_instance; Light *light = (i < omni_light_count) ? omni_light_sort[index].light : spot_light_sort[index].light; + real_t distance = (i < omni_light_count) ? omni_light_sort[index].depth : spot_light_sort[index].depth; if (using_forward_ids) { forward_id_storage->map_forward_id(type == RS::LIGHT_OMNI ? RendererRD::FORWARD_ID_TYPE_OMNI_LIGHT : RendererRD::FORWARD_ID_TYPE_SPOT_LIGHT, light_instance->forward_id, index); @@ -803,7 +802,6 @@ void LightStorage::update_light_buffers(RenderDataRD *p_render_data, const Paged float fade_begin = 0.0; float fade_shadow = 0.0; float fade_length = 0.0; - real_t distance = 0.0; float fade = 1.0; float shadow_opacity_fade = 1.0; @@ -811,7 +809,6 @@ void LightStorage::update_light_buffers(RenderDataRD *p_render_data, const Paged fade_begin = light->distance_fade_begin; fade_shadow = light->distance_fade_shadow; fade_length = light->distance_fade_length; - distance = camera_plane.distance_to(light_transform.origin); // Use `smoothstep()` to make opacity changes more gradual and less noticeable to the player. if (distance > fade_begin) { diff --git a/servers/rendering/renderer_rd/storage_rd/light_storage.h b/servers/rendering/renderer_rd/storage_rd/light_storage.h index f7c8b8833b..d359219128 100644 --- a/servers/rendering/renderer_rd/storage_rd/light_storage.h +++ b/servers/rendering/renderer_rd/storage_rd/light_storage.h @@ -226,7 +226,7 @@ private: Color ambient_color; float ambient_color_energy = 1.0; float max_distance = 0; - Vector3 extents = Vector3(1, 1, 1); + Vector3 extents = Vector3(10, 10, 10); Vector3 origin_offset; bool interior = false; bool box_projection = false; diff --git a/servers/rendering/renderer_rd/storage_rd/material_storage.cpp b/servers/rendering/renderer_rd/storage_rd/material_storage.cpp index 66ae1e8d1a..d631a89dd2 100644 --- a/servers/rendering/renderer_rd/storage_rd/material_storage.cpp +++ b/servers/rendering/renderer_rd/storage_rd/material_storage.cpp @@ -2685,6 +2685,14 @@ void MaterialStorage::material_free(RID p_rid) { Material *material = material_owner.get_or_null(p_rid); ERR_FAIL_COND(!material); + // Need to clear texture arrays to prevent spin locking of their RID's. + // This happens when the app is being closed. + for (KeyValue<StringName, Variant> &E : material->params) { + if (E.value.get_type() == Variant::ARRAY) { + Array(E.value).clear(); + } + } + material_set_shader(p_rid, RID()); //clean up shader material->dependency.deleted_notify(p_rid); diff --git a/servers/rendering/renderer_rd/storage_rd/mesh_storage.cpp b/servers/rendering/renderer_rd/storage_rd/mesh_storage.cpp index 62da62403f..46a0d85ba8 100644 --- a/servers/rendering/renderer_rd/storage_rd/mesh_storage.cpp +++ b/servers/rendering/renderer_rd/storage_rd/mesh_storage.cpp @@ -842,15 +842,15 @@ void MeshStorage::mesh_instance_set_blend_shape_weight(RID p_mesh_instance, int } void MeshStorage::_mesh_instance_clear(MeshInstance *mi) { - for (uint32_t i = 0; i < mi->surfaces.size(); i++) { - if (mi->surfaces[i].versions) { - for (uint32_t j = 0; j < mi->surfaces[i].version_count; j++) { - RD::get_singleton()->free(mi->surfaces[i].versions[j].vertex_array); + for (const RendererRD::MeshStorage::MeshInstance::Surface surface : mi->surfaces) { + if (surface.versions) { + for (uint32_t j = 0; j < surface.version_count; j++) { + RD::get_singleton()->free(surface.versions[j].vertex_array); } - memfree(mi->surfaces[i].versions); + memfree(surface.versions); } - if (mi->surfaces[i].vertex_buffer.is_valid()) { - RD::get_singleton()->free(mi->surfaces[i].vertex_buffer); + if (surface.vertex_buffer.is_valid()) { + RD::get_singleton()->free(surface.vertex_buffer); } } mi->surfaces.clear(); @@ -866,8 +866,8 @@ void MeshStorage::_mesh_instance_clear(MeshInstance *mi) { void MeshStorage::_mesh_instance_add_surface(MeshInstance *mi, Mesh *mesh, uint32_t p_surface) { if (mesh->blend_shape_count > 0 && mi->blend_weights_buffer.is_null()) { mi->blend_weights.resize(mesh->blend_shape_count); - for (uint32_t i = 0; i < mi->blend_weights.size(); i++) { - mi->blend_weights[i] = 0; + for (float &weight : mi->blend_weights) { + weight = 0; } mi->blend_weights_buffer = RD::get_singleton()->storage_buffer_create(sizeof(float) * mi->blend_weights.size(), mi->blend_weights.to_byte_array()); mi->weights_dirty = true; diff --git a/servers/rendering/renderer_rd/storage_rd/particles_storage.cpp b/servers/rendering/renderer_rd/storage_rd/particles_storage.cpp index 6401d0f5d0..00fb8acca8 100644 --- a/servers/rendering/renderer_rd/storage_rd/particles_storage.cpp +++ b/servers/rendering/renderer_rd/storage_rd/particles_storage.cpp @@ -851,9 +851,9 @@ void ParticlesStorage::_particles_process(Particles *p_particles, double p_delta collision_heightmap_texture = p_particles->sdf_collision_texture; //replace in all other history frames where used because parameters are no longer valid if screen moves - for (uint32_t i = 1; i < p_particles->frame_history.size(); i++) { - if (p_particles->frame_history[i].collider_count > 0 && p_particles->frame_history[i].colliders[0].type == ParticlesFrameParams::COLLISION_TYPE_2D_SDF) { - p_particles->frame_history[i].colliders[0] = frame_params.colliders[0]; + for (ParticlesFrameParams ¶ms : p_particles->frame_history) { + if (params.collider_count > 0 && params.colliders[0].type == ParticlesFrameParams::COLLISION_TYPE_2D_SDF) { + params.colliders[0] = frame_params.colliders[0]; } } } @@ -1145,7 +1145,7 @@ void ParticlesStorage::particles_set_view_axis(RID p_particles, const Vector3 &p return; } - if (particles->particle_buffer.is_null()) { + if (particles->particle_buffer.is_null() || particles->trail_bind_pose_uniform_set.is_null()) { return; //particles have not processed yet } @@ -1415,7 +1415,6 @@ void ParticlesStorage::update_particles() { } bool zero_time_scale = Engine::get_singleton()->get_time_scale() <= 0.0; - bool updated = false; if (particles->clear && particles->pre_process_time > 0.0) { double frame_time; @@ -1430,7 +1429,6 @@ void ParticlesStorage::update_particles() { while (todo >= 0) { _particles_process(particles, frame_time); todo -= frame_time; - updated = true; } } @@ -1452,10 +1450,9 @@ void ParticlesStorage::update_particles() { } double todo = particles->frame_remainder + delta; - while (todo >= frame_time || (particles->clear && !updated)) { + while (todo >= frame_time || particles->clear) { _particles_process(particles, frame_time); todo -= decr; - updated = true; } particles->frame_remainder = todo; @@ -1463,16 +1460,16 @@ void ParticlesStorage::update_particles() { } else { if (zero_time_scale) { _particles_process(particles, 0.0); - updated = true; } else { _particles_process(particles, RendererCompositorRD::singleton->get_frame_delta_time()); - updated = true; } } - //copy particles to instance buffer + // Ensure that memory is initialized (the code above should ensure that _particles_process is always called at least once upon clearing). + DEV_ASSERT(!particles->clear); - if (updated && particles->draw_order != RS::PARTICLES_DRAW_ORDER_VIEW_DEPTH && particles->transform_align != RS::PARTICLES_TRANSFORM_ALIGN_Z_BILLBOARD && particles->transform_align != RS::PARTICLES_TRANSFORM_ALIGN_Z_BILLBOARD_Y_TO_VELOCITY) { + // Copy particles to instance buffer. + if (particles->draw_order != RS::PARTICLES_DRAW_ORDER_VIEW_DEPTH && particles->transform_align != RS::PARTICLES_TRANSFORM_ALIGN_Z_BILLBOARD && particles->transform_align != RS::PARTICLES_TRANSFORM_ALIGN_Z_BILLBOARD_Y_TO_VELOCITY) { //does not need view dependent operation, do copy here ParticlesShader::CopyPushConstant copy_push_constant; diff --git a/servers/rendering/renderer_rd/storage_rd/render_scene_buffers_rd.cpp b/servers/rendering/renderer_rd/storage_rd/render_scene_buffers_rd.cpp index d67a848a40..7771f75a65 100644 --- a/servers/rendering/renderer_rd/storage_rd/render_scene_buffers_rd.cpp +++ b/servers/rendering/renderer_rd/storage_rd/render_scene_buffers_rd.cpp @@ -96,7 +96,7 @@ void RenderSceneBuffersRD::cleanup() { named_textures.clear(); } -void RenderSceneBuffersRD::configure(RID p_render_target, const Size2i p_internal_size, const Size2i p_target_size, float p_fsr_sharpness, float p_texture_mipmap_bias, RS::ViewportMSAA p_msaa_3d, RenderingServer::ViewportScreenSpaceAA p_screen_space_aa, bool p_use_taa, bool p_use_debanding, uint32_t p_view_count) { +void RenderSceneBuffersRD::configure(RID p_render_target, const Size2i p_internal_size, const Size2i p_target_size, RS::ViewportScaling3DMode p_scaling_3d_mode, float p_fsr_sharpness, float p_texture_mipmap_bias, RS::ViewportMSAA p_msaa_3d, RenderingServer::ViewportScreenSpaceAA p_screen_space_aa, bool p_use_taa, bool p_use_debanding, uint32_t p_view_count) { RendererRD::TextureStorage *texture_storage = RendererRD::TextureStorage::get_singleton(); RendererRD::MaterialStorage *material_storage = RendererRD::MaterialStorage::get_singleton(); @@ -104,12 +104,7 @@ void RenderSceneBuffersRD::configure(RID p_render_target, const Size2i p_interna target_size = p_target_size; internal_size = p_internal_size; - - // FIXME, right now we do this because only our clustered renderer supports FSR upscale - // this does mean that with linear upscale if we use subpasses, we could get into trouble. - if (!can_be_storage) { - internal_size = target_size; - } + scaling_3d_mode = p_scaling_3d_mode; if (p_use_taa) { // Use negative mipmap LOD bias when TAA is enabled to compensate for loss of sharpness. @@ -193,6 +188,7 @@ void RenderSceneBuffersRD::configure_for_reflections(const Size2i p_reflection_s target_size = p_reflection_size; internal_size = p_reflection_size; render_target = RID(); + scaling_3d_mode = RS::VIEWPORT_SCALING_3D_MODE_OFF; fsr_sharpness = 0.0; msaa_3d = RS::VIEWPORT_MSAA_DISABLED; screen_space_aa = RS::VIEWPORT_SCREEN_SPACE_AA_DISABLED; @@ -255,7 +251,7 @@ RID RenderSceneBuffersRD::create_texture(const StringName &p_context, const Stri } RID RenderSceneBuffersRD::create_texture_from_format(const StringName &p_context, const StringName &p_texture_name, const RD::TextureFormat &p_texture_format, RD::TextureView p_view, bool p_unique) { - // TODO p_unique, if p_unique is true, this is a texture that can be shared. This will be implemented later as an optimisation. + // TODO p_unique, if p_unique is true, this is a texture that can be shared. This will be implemented later as an optimization. NTKey key(p_context, p_texture_name); diff --git a/servers/rendering/renderer_rd/storage_rd/render_scene_buffers_rd.h b/servers/rendering/renderer_rd/storage_rd/render_scene_buffers_rd.h index ff946f410f..dc849fd56a 100644 --- a/servers/rendering/renderer_rd/storage_rd/render_scene_buffers_rd.h +++ b/servers/rendering/renderer_rd/storage_rd/render_scene_buffers_rd.h @@ -73,6 +73,7 @@ private: // The internal size of the textures we render 3D to in case we render at a lower resolution and upscale Size2i internal_size = Size2i(0, 0); + RS::ViewportScaling3DMode scaling_3d_mode = RS::VIEWPORT_SCALING_3D_MODE_OFF; float fsr_sharpness = 0.2f; // Aliassing settings @@ -139,7 +140,7 @@ public: void set_vrs(RendererRD::VRS *p_vrs) { vrs = p_vrs; } void cleanup(); - virtual void configure(RID p_render_target, const Size2i p_internal_size, const Size2i p_target_size, float p_fsr_sharpness, float p_texture_mipmap_bias, RS::ViewportMSAA p_msaa_3d, RenderingServer::ViewportScreenSpaceAA p_screen_space_aa, bool p_use_taa, bool p_use_debanding, uint32_t p_view_count) override; + virtual void configure(RID p_render_target, const Size2i p_internal_size, const Size2i p_target_size, RS::ViewportScaling3DMode p_scaling_3d_mode, float p_fsr_sharpness, float p_texture_mipmap_bias, RS::ViewportMSAA p_msaa_3d, RenderingServer::ViewportScreenSpaceAA p_screen_space_aa, bool p_use_taa, bool p_use_debanding, uint32_t p_view_count) override; void configure_for_reflections(const Size2i p_reflection_size); virtual void set_fsr_sharpness(float p_fsr_sharpness) override; virtual void set_texture_mipmap_bias(float p_texture_mipmap_bias) override; @@ -172,6 +173,7 @@ public: _FORCE_INLINE_ uint32_t get_view_count() const { return view_count; } _FORCE_INLINE_ Size2i get_internal_size() const { return internal_size; } _FORCE_INLINE_ Size2i get_target_size() const { return target_size; } + _FORCE_INLINE_ RS::ViewportScaling3DMode get_scaling_3d_mode() const { return scaling_3d_mode; } _FORCE_INLINE_ float get_fsr_sharpness() const { return fsr_sharpness; } _FORCE_INLINE_ RS::ViewportMSAA get_msaa_3d() const { return msaa_3d; } _FORCE_INLINE_ RS::ViewportScreenSpaceAA get_screen_space_aa() const { return screen_space_aa; } diff --git a/servers/rendering/renderer_rd/storage_rd/texture_storage.cpp b/servers/rendering/renderer_rd/storage_rd/texture_storage.cpp index 06fda8fa9e..2eaa7824fb 100644 --- a/servers/rendering/renderer_rd/storage_rd/texture_storage.cpp +++ b/servers/rendering/renderer_rd/storage_rd/texture_storage.cpp @@ -422,8 +422,7 @@ TextureStorage::TextureStorage() { tformat.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_CAN_UPDATE_BIT | RD::TEXTURE_USAGE_VRS_ATTACHMENT_BIT; tformat.texture_type = RD::TEXTURE_TYPE_2D; if (!RD::get_singleton()->has_feature(RD::SUPPORTS_ATTACHMENT_VRS)) { - tformat.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_CAN_UPDATE_BIT; - tformat.format = RD::DATA_FORMAT_R8_UNORM; + tformat.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_CAN_UPDATE_BIT; } Vector<uint8_t> pv; @@ -2206,6 +2205,12 @@ void TextureStorage::decal_instance_set_transform(RID p_decal_instance, const Tr di->transform = p_transform; } +void TextureStorage::decal_instance_set_sorting_offset(RID p_decal_instance, float p_sorting_offset) { + DecalInstance *di = decal_instance_owner.get_or_null(p_decal_instance); + ERR_FAIL_COND(!di); + di->sorting_offset = p_sorting_offset; +} + /* DECAL DATA API */ void TextureStorage::free_decal_data() { @@ -2233,7 +2238,7 @@ void TextureStorage::set_max_decals(const uint32_t p_max_decals) { decal_buffer = RD::get_singleton()->storage_buffer_create(decal_buffer_size); } -void TextureStorage::update_decal_buffer(const PagedArray<RID> &p_decals, const Transform3D &p_camera_inverse_xform) { +void TextureStorage::update_decal_buffer(const PagedArray<RID> &p_decals, const Transform3D &p_camera_xform) { ForwardIDStorage *forward_id_storage = ForwardIDStorage::get_singleton(); Transform3D uv_xform; @@ -2257,7 +2262,7 @@ void TextureStorage::update_decal_buffer(const PagedArray<RID> &p_decals, const Transform3D xform = decal_instance->transform; - real_t distance = -p_camera_inverse_xform.xform(xform.origin).z; + real_t distance = p_camera_xform.origin.distance_to(xform.origin); if (decal->distance_fade) { float fade_begin = decal->distance_fade_begin; @@ -2272,7 +2277,7 @@ void TextureStorage::update_decal_buffer(const PagedArray<RID> &p_decals, const decal_sort[decal_count].decal_instance = decal_instance; decal_sort[decal_count].decal = decal; - decal_sort[decal_count].depth = distance; + decal_sort[decal_count].depth = distance - decal_instance->sorting_offset; decal_count++; } @@ -2292,11 +2297,10 @@ void TextureStorage::update_decal_buffer(const PagedArray<RID> &p_decals, const decal_instance->cull_mask = decal->cull_mask; - Transform3D xform = decal_instance->transform; float fade = 1.0; if (decal->distance_fade) { - const real_t distance = -p_camera_inverse_xform.xform(xform.origin).z; + const real_t distance = decal_sort[i].depth + decal_instance->sorting_offset; const float fade_begin = decal->distance_fade_begin; const float fade_length = decal->distance_fade_length; @@ -2312,11 +2316,16 @@ void TextureStorage::update_decal_buffer(const PagedArray<RID> &p_decals, const Transform3D scale_xform; scale_xform.basis.scale(decal_extents); - Transform3D to_decal_xform = (p_camera_inverse_xform * xform * scale_xform * uv_xform).affine_inverse(); + + Transform3D xform = decal_instance->transform; + + Transform3D camera_inverse_xform = p_camera_xform.affine_inverse(); + + Transform3D to_decal_xform = (camera_inverse_xform * xform * scale_xform * uv_xform).affine_inverse(); MaterialStorage::store_transform(to_decal_xform, dd.xform); Vector3 normal = xform.basis.get_column(Vector3::AXIS_Y).normalized(); - normal = p_camera_inverse_xform.basis.xform(normal); //camera is normalized, so fine + normal = camera_inverse_xform.basis.xform(normal); //camera is normalized, so fine dd.normal[0] = normal.x; dd.normal[1] = normal.y; @@ -2350,7 +2359,7 @@ void TextureStorage::update_decal_buffer(const PagedArray<RID> &p_decals, const dd.normal_rect[2] = rect.size.x; dd.normal_rect[3] = rect.size.y; - Basis normal_xform = p_camera_inverse_xform.basis * xform.basis.orthonormalized(); + Basis normal_xform = camera_inverse_xform.basis * xform.basis.orthonormalized(); MaterialStorage::store_basis_3x4(normal_xform, dd.normal_xform); } else { dd.normal_rect[0] = 0; diff --git a/servers/rendering/renderer_rd/storage_rd/texture_storage.h b/servers/rendering/renderer_rd/storage_rd/texture_storage.h index 1558342c3b..ea0df0b459 100644 --- a/servers/rendering/renderer_rd/storage_rd/texture_storage.h +++ b/servers/rendering/renderer_rd/storage_rd/texture_storage.h @@ -258,6 +258,7 @@ private: struct DecalInstance { RID decal; Transform3D transform; + float sorting_offset = 0.0; uint32_t cull_mask = 0; RendererRD::ForwardID forward_id = -1; }; @@ -646,6 +647,7 @@ public: virtual RID decal_instance_create(RID p_decal) override; virtual void decal_instance_free(RID p_decal_instance) override; virtual void decal_instance_set_transform(RID p_decal_instance, const Transform3D &p_transform) override; + virtual void decal_instance_set_sorting_offset(RID p_decal_instance, float p_sorting_offset) override; _FORCE_INLINE_ RID decal_instance_get_base(RID p_decal_instance) const { DecalInstance *di = decal_instance_owner.get_or_null(p_decal_instance); @@ -677,7 +679,7 @@ public: void free_decal_data(); void set_max_decals(const uint32_t p_max_decals); RID get_decal_buffer() { return decal_buffer; } - void update_decal_buffer(const PagedArray<RID> &p_decals, const Transform3D &p_camera_inverse_xform); + void update_decal_buffer(const PagedArray<RID> &p_decals, const Transform3D &p_camera_xform); /* RENDER TARGET API */ diff --git a/servers/rendering/renderer_scene_cull.cpp b/servers/rendering/renderer_scene_cull.cpp index 2ea813aab0..9011cdd98a 100644 --- a/servers/rendering/renderer_scene_cull.cpp +++ b/servers/rendering/renderer_scene_cull.cpp @@ -694,6 +694,7 @@ void RendererSceneCull::instance_set_base(RID p_instance, RID p_base) { instance->base_data = decal; decal->instance = RSG::texture_storage->decal_instance_create(p_base); + RSG::texture_storage->decal_instance_set_sorting_offset(decal->instance, instance->sorting_offset); } break; case RS::INSTANCE_LIGHTMAP: { InstanceLightmapData *lightmap_data = memnew(InstanceLightmapData); @@ -748,6 +749,10 @@ void RendererSceneCull::instance_set_scenario(RID p_instance, RID p_scenario) { switch (instance->base_type) { case RS::INSTANCE_LIGHT: { InstanceLightData *light = static_cast<InstanceLightData *>(instance->base_data); + if (instance->visible && RSG::light_storage->light_get_type(instance->base) != RS::LIGHT_DIRECTIONAL && light->bake_mode == RS::LIGHT_BAKE_DYNAMIC) { + instance->scenario->dynamic_lights.erase(light->instance); + } + #ifdef DEBUG_ENABLED if (light->geometries.size()) { ERR_PRINT("BUG, indexing did not unpair geometries from light."); @@ -867,6 +872,9 @@ void RendererSceneCull::instance_set_pivot_data(RID p_instance, float p_sorting_ InstanceGeometryData *geom = static_cast<InstanceGeometryData *>(instance->base_data); ERR_FAIL_NULL(geom->geometry_instance); geom->geometry_instance->set_pivot_data(p_sorting_offset, p_use_aabb_center); + } else if (instance->base_type == RS::INSTANCE_DECAL && instance->base_data) { + InstanceDecalData *decal = static_cast<InstanceDecalData *>(instance->base_data); + RSG::texture_storage->decal_instance_set_sorting_offset(decal->instance, instance->sorting_offset); } } @@ -2820,7 +2828,9 @@ void RendererSceneCull::_scene_cull(CullData &cull_data, InstanceCullResult &cul InstanceGeometryData *geom = static_cast<InstanceGeometryData *>(idata.instance->base_data); ERR_FAIL_NULL(geom->geometry_instance); + cull_data.cull->lock.lock(); geom->geometry_instance->set_softshadow_projector_pairing(geom->softshadow_count > 0, geom->projector_count > 0); + cull_data.cull->lock.unlock(); idata.flags &= ~uint32_t(InstanceData::FLAG_GEOM_PROJECTOR_SOFTSHADOW_DIRTY); } @@ -2887,7 +2897,9 @@ void RendererSceneCull::_scene_cull(CullData &cull_data, InstanceCullResult &cul sh[j] = sh[j].lerp(target_sh[j], MIN(1.0, lightmap_probe_update_speed)); } ERR_FAIL_NULL(geom->geometry_instance); + cull_data.cull->lock.lock(); geom->geometry_instance->set_lightmap_capture(sh); + cull_data.cull->lock.unlock(); idata.instance->last_frame_pass = frame_number; } @@ -3084,15 +3096,15 @@ void RendererSceneCull::_render_scene(const RendererSceneRender::CameraData *p_c #endif if (cull_to > thread_cull_threshold) { //multiple threads - for (uint32_t i = 0; i < scene_cull_result_threads.size(); i++) { - scene_cull_result_threads[i].clear(); + for (InstanceCullResult &thread : scene_cull_result_threads) { + thread.clear(); } WorkerThreadPool::GroupID group_task = WorkerThreadPool::get_singleton()->add_template_group_task(this, &RendererSceneCull::_scene_cull_threaded, &cull_data, scene_cull_result_threads.size(), -1, true, SNAME("RenderCullInstances")); WorkerThreadPool::get_singleton()->wait_for_group_task_completion(group_task); - for (uint32_t i = 0; i < scene_cull_result_threads.size(); i++) { - scene_cull_result.append_from(scene_cull_result_threads[i]); + for (InstanceCullResult &thread : scene_cull_result_threads) { + scene_cull_result.append_from(thread); } } else { @@ -4122,8 +4134,8 @@ RendererSceneCull::RendererSceneCull() { scene_cull_result.init(&rid_cull_page_pool, &geometry_instance_cull_page_pool, &instance_cull_page_pool); scene_cull_result_threads.resize(WorkerThreadPool::get_singleton()->get_thread_count()); - for (uint32_t i = 0; i < scene_cull_result_threads.size(); i++) { - scene_cull_result_threads[i].init(&rid_cull_page_pool, &geometry_instance_cull_page_pool, &instance_cull_page_pool); + for (InstanceCullResult &thread : scene_cull_result_threads) { + thread.init(&rid_cull_page_pool, &geometry_instance_cull_page_pool, &instance_cull_page_pool); } indexer_update_iterations = GLOBAL_GET("rendering/limits/spatial_indexer/update_iterations_per_frame"); @@ -4151,8 +4163,8 @@ RendererSceneCull::~RendererSceneCull() { } scene_cull_result.reset(); - for (uint32_t i = 0; i < scene_cull_result_threads.size(); i++) { - scene_cull_result_threads[i].reset(); + for (InstanceCullResult &thread : scene_cull_result_threads) { + thread.reset(); } scene_cull_result_threads.clear(); diff --git a/servers/rendering/renderer_scene_render.cpp b/servers/rendering/renderer_scene_render.cpp index e2e6ea5aa2..a389e3e767 100644 --- a/servers/rendering/renderer_scene_render.cpp +++ b/servers/rendering/renderer_scene_render.cpp @@ -84,7 +84,7 @@ void RendererSceneRender::CameraData::set_multiview_camera(uint32_t p_view_count Transform3D main_transform_inv = main_transform.inverse(); // 5. figure out far plane, this could use some improvement, we may have our far plane too close like this, not sure if this matters - Vector3 far_center = (planes[0][Projection::PLANE_FAR].center() + planes[1][Projection::PLANE_FAR].center()) * 0.5; + Vector3 far_center = (planes[0][Projection::PLANE_FAR].get_center() + planes[1][Projection::PLANE_FAR].get_center()) * 0.5; Plane far(-z, far_center); ///////////////////////////////////////////////////////////////////////////// diff --git a/servers/rendering/renderer_viewport.cpp b/servers/rendering/renderer_viewport.cpp index 05a3fccf1c..3886f5b379 100644 --- a/servers/rendering/renderer_viewport.cpp +++ b/servers/rendering/renderer_viewport.cpp @@ -115,9 +115,8 @@ void RendererViewport::_configure_3d_render_buffers(Viewport *p_viewport) { if (p_viewport->size.width == 0 || p_viewport->size.height == 0) { p_viewport->render_buffers.unref(); } else { - const float scaling_3d_scale = p_viewport->scaling_3d_scale; + float scaling_3d_scale = p_viewport->scaling_3d_scale; RS::ViewportScaling3DMode scaling_3d_mode = p_viewport->scaling_3d_mode; - bool scaling_enabled = true; if ((scaling_3d_mode == RS::VIEWPORT_SCALING_3D_MODE_FSR) && (scaling_3d_scale > 1.0)) { // FSR is not designed for downsampling. @@ -133,7 +132,7 @@ void RendererViewport::_configure_3d_render_buffers(Viewport *p_viewport) { } if (scaling_3d_scale == 1.0) { - scaling_enabled = false; + scaling_3d_mode = RS::VIEWPORT_SCALING_3D_MODE_OFF; } int width; @@ -141,36 +140,37 @@ void RendererViewport::_configure_3d_render_buffers(Viewport *p_viewport) { int render_width; int render_height; - if (scaling_enabled) { - switch (scaling_3d_mode) { - case RS::VIEWPORT_SCALING_3D_MODE_BILINEAR: - // Clamp 3D rendering resolution to reasonable values supported on most hardware. - // This prevents freezing the engine or outright crashing on lower-end GPUs. - width = CLAMP(p_viewport->size.width * scaling_3d_scale, 1, 16384); - height = CLAMP(p_viewport->size.height * scaling_3d_scale, 1, 16384); - render_width = width; - render_height = height; - break; - case RS::VIEWPORT_SCALING_3D_MODE_FSR: - width = p_viewport->size.width; - height = p_viewport->size.height; - render_width = MAX(width * scaling_3d_scale, 1.0); // width / (width * scaling) - render_height = MAX(height * scaling_3d_scale, 1.0); - break; - default: - // This is an unknown mode. - WARN_PRINT_ONCE(vformat("Unknown scaling mode: %d. Disabling 3D resolution scaling.", scaling_3d_mode)); - width = p_viewport->size.width; - height = p_viewport->size.height; - render_width = width; - render_height = height; - break; - } - } else { - width = p_viewport->size.width; - height = p_viewport->size.height; - render_width = width; - render_height = height; + switch (scaling_3d_mode) { + case RS::VIEWPORT_SCALING_3D_MODE_BILINEAR: + // Clamp 3D rendering resolution to reasonable values supported on most hardware. + // This prevents freezing the engine or outright crashing on lower-end GPUs. + width = CLAMP(p_viewport->size.width * scaling_3d_scale, 1, 16384); + height = CLAMP(p_viewport->size.height * scaling_3d_scale, 1, 16384); + render_width = width; + render_height = height; + break; + case RS::VIEWPORT_SCALING_3D_MODE_FSR: + width = p_viewport->size.width; + height = p_viewport->size.height; + render_width = MAX(width * scaling_3d_scale, 1.0); // width / (width * scaling) + render_height = MAX(height * scaling_3d_scale, 1.0); + break; + case RS::VIEWPORT_SCALING_3D_MODE_OFF: + width = p_viewport->size.width; + height = p_viewport->size.height; + render_width = width; + render_height = height; + break; + default: + // This is an unknown mode. + WARN_PRINT_ONCE(vformat("Unknown scaling mode: %d. Disabling 3D resolution scaling.", scaling_3d_mode)); + scaling_3d_mode = RS::VIEWPORT_SCALING_3D_MODE_OFF; + scaling_3d_scale = 1.0; + width = p_viewport->size.width; + height = p_viewport->size.height; + render_width = width; + render_height = height; + break; } p_viewport->internal_size = Size2(render_width, render_height); @@ -179,7 +179,7 @@ void RendererViewport::_configure_3d_render_buffers(Viewport *p_viewport) { // to compensate for the loss of sharpness. const float texture_mipmap_bias = log2f(MIN(scaling_3d_scale, 1.0)) + p_viewport->texture_mipmap_bias; - p_viewport->render_buffers->configure(p_viewport->render_target, Size2i(render_width, render_height), Size2(width, height), p_viewport->fsr_sharpness, texture_mipmap_bias, p_viewport->msaa_3d, p_viewport->screen_space_aa, p_viewport->use_taa, p_viewport->use_debanding, p_viewport->view_count); + p_viewport->render_buffers->configure(p_viewport->render_target, Size2i(render_width, render_height), Size2(width, height), scaling_3d_mode, p_viewport->fsr_sharpness, texture_mipmap_bias, p_viewport->msaa_3d, p_viewport->screen_space_aa, p_viewport->use_taa, p_viewport->use_debanding, p_viewport->view_count); } } } @@ -1138,6 +1138,7 @@ void RendererViewport::viewport_set_screen_space_aa(RID p_viewport, RS::Viewport void RendererViewport::viewport_set_use_taa(RID p_viewport, bool p_use_taa) { Viewport *viewport = viewport_owner.get_or_null(p_viewport); ERR_FAIL_COND(!viewport); + ERR_FAIL_COND_EDMSG(OS::get_singleton()->get_current_rendering_method() != "forward_plus", "TAA is only available when using the Forward+ renderer."); if (viewport->use_taa == p_use_taa) { return; diff --git a/servers/rendering/rendering_device_binds.cpp b/servers/rendering/rendering_device_binds.cpp index 5967835d03..3678b70254 100644 --- a/servers/rendering/rendering_device_binds.cpp +++ b/servers/rendering/rendering_device_binds.cpp @@ -31,6 +31,8 @@ #include "rendering_device_binds.h" Error RDShaderFile::parse_versions_from_text(const String &p_text, const String p_defines, OpenIncludeFunction p_include_func, void *p_include_func_userdata) { + ERR_FAIL_NULL_V(RenderingDevice::get_singleton(), ERR_UNAVAILABLE); + Vector<String> lines = p_text.split("\n"); bool reading_versions = false; diff --git a/servers/rendering/rendering_server_default.cpp b/servers/rendering/rendering_server_default.cpp index 4c4b3d13f9..6017eff55e 100644 --- a/servers/rendering/rendering_server_default.cpp +++ b/servers/rendering/rendering_server_default.cpp @@ -291,6 +291,10 @@ void RenderingServerDefault::set_boot_image(const Ref<Image> &p_image, const Col RSG::rasterizer->set_boot_image(p_image, p_color, p_scale, p_use_filter); } +Color RenderingServerDefault::get_default_clear_color() { + return RSG::texture_storage->get_default_clear_color(); +} + void RenderingServerDefault::set_default_clear_color(const Color &p_color) { RSG::viewport->set_default_clear_color(p_color); } diff --git a/servers/rendering/rendering_server_default.h b/servers/rendering/rendering_server_default.h index f2fadd5e1f..8ac522bafe 100644 --- a/servers/rendering/rendering_server_default.h +++ b/servers/rendering/rendering_server_default.h @@ -854,7 +854,7 @@ public: FUNC8(canvas_item_add_msdf_texture_rect_region, RID, const Rect2 &, RID, const Rect2 &, const Color &, int, float, float) FUNC5(canvas_item_add_lcd_texture_rect_region, RID, const Rect2 &, RID, const Rect2 &, const Color &) FUNC10(canvas_item_add_nine_patch, RID, const Rect2 &, const Rect2 &, RID, const Vector2 &, const Vector2 &, NinePatchAxisMode, NinePatchAxisMode, bool, const Color &) - FUNC6(canvas_item_add_primitive, RID, const Vector<Point2> &, const Vector<Color> &, const Vector<Point2> &, RID, float) + FUNC5(canvas_item_add_primitive, RID, const Vector<Point2> &, const Vector<Color> &, const Vector<Point2> &, RID) FUNC5(canvas_item_add_polygon, RID, const Vector<Point2> &, const Vector<Color> &, const Vector<Point2> &, RID) FUNC9(canvas_item_add_triangle_array, RID, const Vector<int> &, const Vector<Point2> &, const Vector<Color> &, const Vector<Point2> &, const Vector<int> &, const Vector<float> &, RID, int) FUNC5(canvas_item_add_mesh, RID, const RID &, const Transform2D &, const Color &, RID) @@ -986,6 +986,7 @@ public: virtual double get_frame_setup_time_cpu() const override; virtual void set_boot_image(const Ref<Image> &p_image, const Color &p_color, bool p_scale, bool p_use_filter = true) override; + virtual Color get_default_clear_color() override; virtual void set_default_clear_color(const Color &p_color) override; virtual bool has_feature(Features p_feature) const override; diff --git a/servers/rendering/shader_compiler.cpp b/servers/rendering/shader_compiler.cpp index 626da90168..68542f32af 100644 --- a/servers/rendering/shader_compiler.cpp +++ b/servers/rendering/shader_compiler.cpp @@ -909,9 +909,6 @@ String ShaderCompiler::_dump_node_code(const SL::Node *p_node, int p_level, Gene if (p_default_actions.renames.has(vnode->name)) { code = p_default_actions.renames[vnode->name]; - if (vnode->name == "SCREEN_TEXTURE") { - r_gen_code.uses_screen_texture_mipmaps = true; - } } else { if (shader->uniforms.has(vnode->name)) { //its a uniform! @@ -919,29 +916,22 @@ String ShaderCompiler::_dump_node_code(const SL::Node *p_node, int p_level, Gene if (u.texture_order >= 0) { StringName name = vnode->name; if (u.hint == ShaderLanguage::ShaderNode::Uniform::HINT_SCREEN_TEXTURE) { - name = "SCREEN_TEXTURE"; + name = "color_buffer"; if (u.filter >= ShaderLanguage::FILTER_NEAREST_MIPMAP) { r_gen_code.uses_screen_texture_mipmaps = true; } + r_gen_code.uses_screen_texture = true; } else if (u.hint == ShaderLanguage::ShaderNode::Uniform::HINT_NORMAL_ROUGHNESS_TEXTURE) { - name = "NORMAL_ROUGHNESS_TEXTURE"; + name = "normal_roughness_buffer"; + r_gen_code.uses_normal_roughness_texture = true; } else if (u.hint == ShaderLanguage::ShaderNode::Uniform::HINT_DEPTH_TEXTURE) { - name = "DEPTH_TEXTURE"; + name = "depth_buffer"; + r_gen_code.uses_depth_texture = true; } else { name = _mkid(vnode->name); //texture, use as is } - if (p_default_actions.renames.has(name)) { - code = p_default_actions.renames[name]; - } else { - code = name; - } - - if (p_actions.usage_flag_pointers.has(name) && !used_flag_pointers.has(name)) { - *p_actions.usage_flag_pointers[name] = true; - used_flag_pointers.insert(name); - } - + code = name; } else { //a scalar or vector if (u.scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_GLOBAL) { @@ -1251,16 +1241,20 @@ String ShaderCompiler::_dump_node_code(const SL::Node *p_node, int p_level, Gene } if (correct_texture_uniform) { - //TODO Needs to detect screen_texture hint as well - is_screen_texture = (texture_uniform == "SCREEN_TEXTURE"); - String sampler_name; + bool is_normal_roughness_texture = false; if (actions.custom_samplers.has(texture_uniform)) { sampler_name = actions.custom_samplers[texture_uniform]; } else { if (shader->uniforms.has(texture_uniform)) { - sampler_name = _get_sampler_name(shader->uniforms[texture_uniform].filter, shader->uniforms[texture_uniform].repeat); + const ShaderLanguage::ShaderNode::Uniform &u = shader->uniforms[texture_uniform]; + if (u.hint == ShaderLanguage::ShaderNode::Uniform::HINT_SCREEN_TEXTURE) { + is_screen_texture = true; + } else if (u.hint == ShaderLanguage::ShaderNode::Uniform::HINT_NORMAL_ROUGHNESS_TEXTURE) { + is_normal_roughness_texture = true; + } + sampler_name = _get_sampler_name(u.filter, u.repeat); } else { bool found = false; @@ -1287,7 +1281,7 @@ String ShaderCompiler::_dump_node_code(const SL::Node *p_node, int p_level, Gene } String data_type_name = ""; - if (texture_uniform == "NORMAL_ROUGHNESS_TEXTURE") { + if (is_normal_roughness_texture) { data_type_name = "multiviewSampler"; normal_roughness_texture_used = true; } else { @@ -1515,6 +1509,9 @@ Error ShaderCompiler::compile(RS::ShaderMode p_mode, const String &p_code, Ident r_gen_code.uses_vertex_time = false; r_gen_code.uses_global_textures = false; r_gen_code.uses_screen_texture_mipmaps = false; + r_gen_code.uses_screen_texture = false; + r_gen_code.uses_depth_texture = false; + r_gen_code.uses_normal_roughness_texture = false; used_name_defines.clear(); used_rmode_defines.clear(); diff --git a/servers/rendering/shader_compiler.h b/servers/rendering/shader_compiler.h index eeb2916160..43bea213da 100644 --- a/servers/rendering/shader_compiler.h +++ b/servers/rendering/shader_compiler.h @@ -81,6 +81,9 @@ public: bool uses_fragment_time; bool uses_vertex_time; bool uses_screen_texture_mipmaps; + bool uses_screen_texture; + bool uses_depth_texture; + bool uses_normal_roughness_texture; }; struct DefaultIdentifierActions { diff --git a/servers/rendering/shader_language.cpp b/servers/rendering/shader_language.cpp index 59a2ff66f3..ba7ffda00a 100644 --- a/servers/rendering/shader_language.cpp +++ b/servers/rendering/shader_language.cpp @@ -5399,6 +5399,12 @@ ShaderLanguage::Node *ShaderLanguage::_parse_expression(BlockNode *p_block, cons } } else { if (!_find_identifier(p_block, false, p_function_info, identifier, &data_type, &ident_type, &is_const, &array_size, &struct_name)) { + if (identifier == "SCREEN_TEXTURE" || identifier == "DEPTH_TEXTURE" || identifier == "NORMAL_ROUGHNESS_TEXTURE") { + String name = String(identifier); + String name_lower = name.to_lower(); + _set_error(vformat(RTR("%s has been removed in favor of using hint_%s with a uniform.\nTo continue with minimal code changes add 'uniform sampler2D %s : hint_%s, filter_linear_mipmap;' near the top of your shader."), name, name_lower, name, name_lower)); + return nullptr; + } _set_error(vformat(RTR("Unknown identifier in expression: '%s'."), String(identifier))); return nullptr; } @@ -8691,14 +8697,17 @@ Error ShaderLanguage::_parse_shader(const HashMap<StringName, FunctionInfo> &p_f case TK_HINT_SCREEN_TEXTURE: { new_hint = ShaderNode::Uniform::HINT_SCREEN_TEXTURE; --texture_uniforms; + --texture_binding; } break; case TK_HINT_NORMAL_ROUGHNESS_TEXTURE: { new_hint = ShaderNode::Uniform::HINT_NORMAL_ROUGHNESS_TEXTURE; --texture_uniforms; + --texture_binding; } break; case TK_HINT_DEPTH_TEXTURE: { new_hint = ShaderNode::Uniform::HINT_DEPTH_TEXTURE; --texture_uniforms; + --texture_binding; } break; case TK_FILTER_NEAREST: { new_filter = FILTER_NEAREST; diff --git a/servers/rendering/shader_preprocessor.cpp b/servers/rendering/shader_preprocessor.cpp index 40c8acffe5..97efd2df94 100644 --- a/servers/rendering/shader_preprocessor.cpp +++ b/servers/rendering/shader_preprocessor.cpp @@ -330,8 +330,8 @@ String ShaderPreprocessor::vector_to_string(const LocalVector<char32_t> &p_v, in String ShaderPreprocessor::tokens_to_string(const LocalVector<Token> &p_tokens) { LocalVector<char32_t> result; - for (uint32_t i = 0; i < p_tokens.size(); i++) { - result.push_back(p_tokens[i].text); + for (const Token &token : p_tokens) { + result.push_back(token.text); } return vector_to_string(result); } @@ -1081,21 +1081,17 @@ ShaderPreprocessor::Define *ShaderPreprocessor::create_define(const String &p_bo return define; } -void ShaderPreprocessor::clear() { - if (state_owner && state != nullptr) { +void ShaderPreprocessor::clear_state() { + if (state != nullptr) { for (const RBMap<String, Define *>::Element *E = state->defines.front(); E; E = E->next()) { memdelete(E->get()); } - - memdelete(state); + state->defines.clear(); } - state_owner = false; state = nullptr; } Error ShaderPreprocessor::preprocess(State *p_state, const String &p_code, String &r_result) { - clear(); - output.clear(); state = p_state; @@ -1242,6 +1238,9 @@ Error ShaderPreprocessor::preprocess(const String &p_code, const String &p_filen } } } + + clear_state(); + return err; } @@ -1273,5 +1272,4 @@ ShaderPreprocessor::ShaderPreprocessor() { } ShaderPreprocessor::~ShaderPreprocessor() { - clear(); } diff --git a/servers/rendering/shader_preprocessor.h b/servers/rendering/shader_preprocessor.h index f5902c64ca..6e5533c575 100644 --- a/servers/rendering/shader_preprocessor.h +++ b/servers/rendering/shader_preprocessor.h @@ -167,7 +167,6 @@ private: private: LocalVector<char32_t> output; State *state = nullptr; - bool state_owner = false; private: static bool is_char_word(char32_t p_char); @@ -211,7 +210,7 @@ private: static Define *create_define(const String &p_body); - void clear(); + void clear_state(); Error preprocess(State *p_state, const String &p_code, String &r_result); diff --git a/servers/rendering/shader_types.cpp b/servers/rendering/shader_types.cpp index 38dc806370..ba39328b2e 100644 --- a/servers/rendering/shader_types.cpp +++ b/servers/rendering/shader_types.cpp @@ -138,9 +138,6 @@ ShaderTypes::ShaderTypes() { shader_modes[RS::SHADER_SPATIAL].functions["fragment"].built_ins["AO"] = ShaderLanguage::TYPE_FLOAT; shader_modes[RS::SHADER_SPATIAL].functions["fragment"].built_ins["AO_LIGHT_AFFECT"] = ShaderLanguage::TYPE_FLOAT; shader_modes[RS::SHADER_SPATIAL].functions["fragment"].built_ins["EMISSION"] = ShaderLanguage::TYPE_VEC3; - shader_modes[RS::SHADER_SPATIAL].functions["fragment"].built_ins["SCREEN_TEXTURE"] = constt(ShaderLanguage::TYPE_SAMPLER2D); - shader_modes[RS::SHADER_SPATIAL].functions["fragment"].built_ins["NORMAL_ROUGHNESS_TEXTURE"] = constt(ShaderLanguage::TYPE_SAMPLER2D); - shader_modes[RS::SHADER_SPATIAL].functions["fragment"].built_ins["DEPTH_TEXTURE"] = constt(ShaderLanguage::TYPE_SAMPLER2D); shader_modes[RS::SHADER_SPATIAL].functions["fragment"].built_ins["DEPTH"] = ShaderLanguage::TYPE_FLOAT; shader_modes[RS::SHADER_SPATIAL].functions["fragment"].built_ins["SCREEN_UV"] = constt(ShaderLanguage::TYPE_VEC2); shader_modes[RS::SHADER_SPATIAL].functions["fragment"].built_ins["POINT_COORD"] = constt(ShaderLanguage::TYPE_VEC2); @@ -267,7 +264,6 @@ ShaderTypes::ShaderTypes() { shader_modes[RS::SHADER_CANVAS_ITEM].functions["fragment"].built_ins["SCREEN_PIXEL_SIZE"] = constt(ShaderLanguage::TYPE_VEC2); shader_modes[RS::SHADER_CANVAS_ITEM].functions["fragment"].built_ins["POINT_COORD"] = constt(ShaderLanguage::TYPE_VEC2); shader_modes[RS::SHADER_CANVAS_ITEM].functions["fragment"].built_ins["AT_LIGHT_PASS"] = constt(ShaderLanguage::TYPE_BOOL); - shader_modes[RS::SHADER_CANVAS_ITEM].functions["fragment"].built_ins["SCREEN_TEXTURE"] = constt(ShaderLanguage::TYPE_SAMPLER2D); shader_modes[RS::SHADER_CANVAS_ITEM].functions["fragment"].can_discard = true; shader_modes[RS::SHADER_CANVAS_ITEM].functions["fragment"].main_function = true; diff --git a/servers/rendering/storage/render_scene_buffers.cpp b/servers/rendering/storage/render_scene_buffers.cpp index 1320199833..6369139aa6 100644 --- a/servers/rendering/storage/render_scene_buffers.cpp +++ b/servers/rendering/storage/render_scene_buffers.cpp @@ -34,8 +34,8 @@ void RenderSceneBuffers::_bind_methods() { ClassDB::bind_method(D_METHOD("configure", "render_target", "internal_size", "target_size", "fsr_sharpness", "texture_mipmap_bias", "msaa", "screen_space_aa", "use_taa", "use_debanding", "view_count"), &RenderSceneBuffers::configure); } -void RenderSceneBuffers::configure(RID p_render_target, const Size2i p_internal_size, const Size2i p_target_size, float p_fsr_sharpness, float p_texture_mipmap_bias, RS::ViewportMSAA p_msaa, RenderingServer::ViewportScreenSpaceAA p_screen_space_aa, bool p_use_taa, bool p_use_debanding, uint32_t p_view_count) { - GDVIRTUAL_CALL(_configure, p_render_target, p_internal_size, p_target_size, p_fsr_sharpness, p_texture_mipmap_bias, p_msaa, p_screen_space_aa, p_use_taa, p_use_debanding, p_view_count); +void RenderSceneBuffers::configure(RID p_render_target, const Size2i p_internal_size, const Size2i p_target_size, RS::ViewportScaling3DMode p_scaling_3d_mode, float p_fsr_sharpness, float p_texture_mipmap_bias, RS::ViewportMSAA p_msaa, RenderingServer::ViewportScreenSpaceAA p_screen_space_aa, bool p_use_taa, bool p_use_debanding, uint32_t p_view_count) { + GDVIRTUAL_CALL(_configure, p_render_target, p_internal_size, p_target_size, p_scaling_3d_mode, p_fsr_sharpness, p_texture_mipmap_bias, p_msaa, p_screen_space_aa, p_use_taa, p_use_debanding, p_view_count); }; void RenderSceneBuffers::set_fsr_sharpness(float p_fsr_sharpness) { diff --git a/servers/rendering/storage/render_scene_buffers.h b/servers/rendering/storage/render_scene_buffers.h index 83fc78ca1c..cf96a9f372 100644 --- a/servers/rendering/storage/render_scene_buffers.h +++ b/servers/rendering/storage/render_scene_buffers.h @@ -40,7 +40,7 @@ class RenderSceneBuffers : public RefCounted { protected: static void _bind_methods(); - GDVIRTUAL10(_configure, RID, Size2i, Size2i, float, float, RS::ViewportMSAA, RenderingServer::ViewportScreenSpaceAA, bool, bool, uint32_t) + GDVIRTUAL11(_configure, RID, Size2i, Size2i, RS::ViewportScaling3DMode, float, float, RS::ViewportMSAA, RenderingServer::ViewportScreenSpaceAA, bool, bool, uint32_t) GDVIRTUAL1(_set_fsr_sharpness, float) GDVIRTUAL1(_set_texture_mipmap_bias, float) GDVIRTUAL1(_set_use_debanding, bool) @@ -49,7 +49,7 @@ public: RenderSceneBuffers(){}; virtual ~RenderSceneBuffers(){}; - virtual void configure(RID p_render_target, const Size2i p_internal_size, const Size2i p_target_size, float p_fsr_sharpness, float p_texture_mipmap_bias, RS::ViewportMSAA p_msaa_3d, RenderingServer::ViewportScreenSpaceAA p_screen_space_aa, bool p_use_taa, bool p_use_debanding, uint32_t p_view_count); + virtual void configure(RID p_render_target, const Size2i p_internal_size, const Size2i p_target_size, RS::ViewportScaling3DMode p_scaling_3d_mode, float p_fsr_sharpness, float p_texture_mipmap_bias, RS::ViewportMSAA p_msaa_3d, RenderingServer::ViewportScreenSpaceAA p_screen_space_aa, bool p_use_taa, bool p_use_debanding, uint32_t p_view_count); // for those settings that are unlikely to require buffers to be recreated, we'll add setters virtual void set_fsr_sharpness(float p_fsr_sharpness); diff --git a/servers/rendering/storage/texture_storage.h b/servers/rendering/storage/texture_storage.h index 92149b0064..4c4a84d04e 100644 --- a/servers/rendering/storage/texture_storage.h +++ b/servers/rendering/storage/texture_storage.h @@ -127,6 +127,7 @@ public: virtual RID decal_instance_create(RID p_decal) = 0; virtual void decal_instance_free(RID p_decal_instance) = 0; virtual void decal_instance_set_transform(RID p_decal_instance, const Transform3D &p_transform) = 0; + virtual void decal_instance_set_sorting_offset(RID p_decal_instance, float p_sorting_offset) = 0; /* RENDER TARGET */ |