-Rewrote GLES2 lighting and shadows and optimized state changes, did many optimizations, added vertex lighting.

-Did some fixes to GLES3 too

5 files changed, 62 insertions, 26 deletions

diff --git a/drivers/gles3/rasterizer_scene_gles3.cpp b/drivers/gles3/rasterizer_scene_gles3.cpp
index 1e43651d54..086829f9ba 100644
--- a/drivers/gles3/rasterizer_scene_gles3.cpp
+++ b/drivers/gles3/rasterizer_scene_gles3.cpp
@@ -1557,8 +1557,11 @@ void RasterizerSceneGLES3::_render_geometry(RenderList::Element *e) {
 			RasterizerStorageGLES3::MultiMesh *multi_mesh = static_cast<RasterizerStorageGLES3::MultiMesh *>(e->owner);
 			RasterizerStorageGLES3::Surface *s = static_cast<RasterizerStorageGLES3::Surface *>(e->geometry);
 
-			int amount = MAX(multi_mesh->size, multi_mesh->visible_instances);
+			int amount = MIN(multi_mesh->size, multi_mesh->visible_instances);
 
+			if (amount == -1) {
+				amount = multi_mesh->size;
+			}
 #ifdef DEBUG_ENABLED
 
 			if (state.debug_draw == VS::VIEWPORT_DEBUG_DRAW_WIREFRAME && s->array_wireframe_id) {
@@ -2364,14 +2367,15 @@ void RasterizerSceneGLES3::_add_geometry_with_material(RasterizerStorageGLES3::G
 	e->sort_key |= uint64_t(e->geometry->index) << RenderList::SORT_KEY_GEOMETRY_INDEX_SHIFT;
 	e->sort_key |= uint64_t(e->instance->base_type) << RenderList::SORT_KEY_GEOMETRY_TYPE_SHIFT;
 
-	if (!p_depth_pass) {
+	if (e->material->last_pass != render_pass) {
+		e->material->last_pass = render_pass;
+		e->material->index = current_material_index++;
+	}
 
-		if (e->material->last_pass != render_pass) {
-			e->material->last_pass = render_pass;
-			e->material->index = current_material_index++;
-		}
+	e->sort_key |= uint64_t(e->material->index) << RenderList::SORT_KEY_MATERIAL_INDEX_SHIFT;
+	e->sort_key |= uint64_t(e->instance->depth_layer) << RenderList::SORT_KEY_OPAQUE_DEPTH_LAYER_SHIFT;
 
-		e->sort_key |= uint64_t(e->material->index) << RenderList::SORT_KEY_MATERIAL_INDEX_SHIFT;
+	if (!p_depth_pass) {
 
 		if (e->instance->gi_probe_instances.size()) {
 			e->sort_key |= SORT_KEY_GI_PROBES_FLAG;
@@ -2386,9 +2390,6 @@ void RasterizerSceneGLES3::_add_geometry_with_material(RasterizerStorageGLES3::G
 		}
 
 		e->sort_key |= uint64_t(p_material->render_priority + 128) << RenderList::SORT_KEY_PRIORITY_SHIFT;
-	} else {
-		e->sort_key |= uint64_t(e->instance->depth_layer) << RenderList::SORT_KEY_OPAQUE_DEPTH_LAYER_SHIFT;
-		e->sort_key |= uint64_t(e->material->index) << RenderList::SORT_KEY_MATERIAL_INDEX_SHIFT;
 	}
 
 	/*
@@ -2738,7 +2739,7 @@ void RasterizerSceneGLES3::_setup_directional_light(int p_index, const Transform
 
 			ubo_data.shadow_split_offsets[j] = li->shadow_transform[j].split;
 
-			Transform modelview = (p_camera_inverse_transform * li->shadow_transform[j].transform).inverse();
+			Transform modelview = (p_camera_inverse_transform * li->shadow_transform[j].transform).affine_inverse();
 
 			CameraMatrix bias;
 			bias.set_light_bias();
@@ -3131,7 +3132,6 @@ void RasterizerSceneGLES3::_fill_render_list(InstanceBase **p_cull_result, int p
 	current_material_index = 0;
 	state.used_sss = false;
 	state.used_screen_texture = false;
-
 	//fill list
 
 	for (int i = 0; i < p_cull_count; i++) {
@@ -5147,13 +5147,13 @@ void RasterizerSceneGLES3::initialize() {
 
 void RasterizerSceneGLES3::iteration() {
 
-	shadow_filter_mode = ShadowFilterMode(int(ProjectSettings::get_singleton()->get("rendering/quality/shadows/filter_mode")));
-	subsurface_scatter_follow_surface = ProjectSettings::get_singleton()->get("rendering/quality/subsurface_scattering/follow_surface");
-	subsurface_scatter_weight_samples = ProjectSettings::get_singleton()->get("rendering/quality/subsurface_scattering/weight_samples");
-	subsurface_scatter_quality = SubSurfaceScatterQuality(int(ProjectSettings::get_singleton()->get("rendering/quality/subsurface_scattering/quality")));
-	subsurface_scatter_size = ProjectSettings::get_singleton()->get("rendering/quality/subsurface_scattering/scale");
+	shadow_filter_mode = ShadowFilterMode(int(GLOBAL_GET("rendering/quality/shadows/filter_mode")));
+	subsurface_scatter_follow_surface = GLOBAL_GET("rendering/quality/subsurface_scattering/follow_surface");
+	subsurface_scatter_weight_samples = GLOBAL_GET("rendering/quality/subsurface_scattering/weight_samples");
+	subsurface_scatter_quality = SubSurfaceScatterQuality(int(GLOBAL_GET("rendering/quality/subsurface_scattering/quality")));
+	subsurface_scatter_size = GLOBAL_GET("rendering/quality/subsurface_scattering/scale");
 
-	state.scene_shader.set_conditional(SceneShaderGLES3::VCT_QUALITY_HIGH, ProjectSettings::get_singleton()->get("rendering/quality/voxel_cone_tracing/high_quality"));
+	state.scene_shader.set_conditional(SceneShaderGLES3::VCT_QUALITY_HIGH, GLOBAL_GET("rendering/quality/voxel_cone_tracing/high_quality"));
 }
 
 void RasterizerSceneGLES3::finalize() {
diff --git a/drivers/gles3/shader_compiler_gles3.cpp b/drivers/gles3/shader_compiler_gles3.cpp
index 11c84e7db8..350107de69 100644
--- a/drivers/gles3/shader_compiler_gles3.cpp
+++ b/drivers/gles3/shader_compiler_gles3.cpp
@@ -31,6 +31,7 @@
 #include "shader_compiler_gles3.h"
 
 #include "core/os/os.h"
+#include "core/project_settings.h"
 
 #define SL ShaderLanguage
 
@@ -860,6 +861,7 @@ ShaderCompilerGLES3::ShaderCompilerGLES3() {
 	actions[VS::SHADER_SPATIAL].renames["SCREEN_TEXTURE"] = "screen_texture";
 	actions[VS::SHADER_SPATIAL].renames["DEPTH_TEXTURE"] = "depth_buffer";
 	actions[VS::SHADER_SPATIAL].renames["ALPHA_SCISSOR"] = "alpha_scissor";
+	actions[VS::SHADER_SPATIAL].renames["OUTPUT_IS_SRGB"] = "SHADER_IS_SRGB";
 
 	//for light
 	actions[VS::SHADER_SPATIAL].renames["VIEW"] = "view";
@@ -901,12 +903,24 @@ ShaderCompilerGLES3::ShaderCompilerGLES3() {
 	actions[VS::SHADER_SPATIAL].render_mode_defines["cull_front"] = "#define DO_SIDE_CHECK\n";
 	actions[VS::SHADER_SPATIAL].render_mode_defines["cull_disabled"] = "#define DO_SIDE_CHECK\n";
 
-	actions[VS::SHADER_SPATIAL].render_mode_defines["diffuse_burley"] = "#define DIFFUSE_BURLEY\n";
+	bool force_lambert = GLOBAL_GET("rendering/quality/shading/force_lambert_over_burley");
+
+	if (!force_lambert) {
+		actions[VS::SHADER_SPATIAL].render_mode_defines["diffuse_burley"] = "#define DIFFUSE_BURLEY\n";
+	}
+
 	actions[VS::SHADER_SPATIAL].render_mode_defines["diffuse_oren_nayar"] = "#define DIFFUSE_OREN_NAYAR\n";
 	actions[VS::SHADER_SPATIAL].render_mode_defines["diffuse_lambert_wrap"] = "#define DIFFUSE_LAMBERT_WRAP\n";
 	actions[VS::SHADER_SPATIAL].render_mode_defines["diffuse_toon"] = "#define DIFFUSE_TOON\n";
 
-	actions[VS::SHADER_SPATIAL].render_mode_defines["specular_schlick_ggx"] = "#define SPECULAR_SCHLICK_GGX\n";
+	bool force_blinn = GLOBAL_GET("rendering/quality/shading/force_blinn_over_ggx");
+
+	if (!force_blinn) {
+		actions[VS::SHADER_SPATIAL].render_mode_defines["specular_schlick_ggx"] = "#define SPECULAR_SCHLICK_GGX\n";
+	} else {
+		actions[VS::SHADER_SPATIAL].render_mode_defines["specular_schlick_ggx"] = "#define SPECULAR_BLINN\n";
+	}
+
 	actions[VS::SHADER_SPATIAL].render_mode_defines["specular_blinn"] = "#define SPECULAR_BLINN\n";
 	actions[VS::SHADER_SPATIAL].render_mode_defines["specular_phong"] = "#define SPECULAR_PHONG\n";
 	actions[VS::SHADER_SPATIAL].render_mode_defines["specular_toon"] = "#define SPECULAR_TOON\n";
diff --git a/drivers/gles3/shader_gles3.cpp b/drivers/gles3/shader_gles3.cpp
index 2a3b8a9b91..799179e8d4 100644
--- a/drivers/gles3/shader_gles3.cpp
+++ b/drivers/gles3/shader_gles3.cpp
@@ -122,6 +122,11 @@ bool ShaderGLES3::bind() {
 
 	ERR_FAIL_COND_V(!version, false);
 
+	if (!version->ok) { //broken, unable to bind (do not throw error, you saw it before already when it failed compilation).
+		glUseProgram(0);
+		return false;
+	}
+
 	glUseProgram(version->id);
 
 	DEBUG_TEST_ERROR("Use Program");
diff --git a/drivers/gles3/shader_gles3.h b/drivers/gles3/shader_gles3.h
index ca74317218..9db4942163 100644
--- a/drivers/gles3/shader_gles3.h
+++ b/drivers/gles3/shader_gles3.h
@@ -336,6 +336,7 @@ public:
 	}
 
 	uint32_t get_version() const { return new_conditional_version.version; }
+	_FORCE_INLINE_ bool is_version_valid() const { return version && version->ok; }
 
 	void set_uniform_camera(int p_idx, const CameraMatrix &p_mat) {
 
diff --git a/drivers/gles3/shaders/scene.glsl b/drivers/gles3/shaders/scene.glsl
index 12cbe02d0c..49296b7f4a 100644
--- a/drivers/gles3/shaders/scene.glsl
+++ b/drivers/gles3/shaders/scene.glsl
@@ -3,6 +3,8 @@
 
 #define M_PI 3.14159265359
 
+#define SHADER_IS_SRGB false
+
 /*
 from VisualServer:
 
@@ -514,6 +516,7 @@ VERTEX_SHADER_CODE
 /* clang-format off */
 [fragment]
 
+
 /* texture unit usage, N is max_texture_unity-N
 
 1-skeleton
@@ -533,6 +536,7 @@ uniform highp mat4 world_transform;
 /* clang-format on */
 
 #define M_PI 3.14159265359
+#define SHADER_IS_SRGB false
 
 /* Varyings */
 
@@ -1020,18 +1024,30 @@ LIGHT_SHADER_CODE
 
 #if defined(SPECULAR_BLINN)
 
+		//normalized blinn
 		vec3 H = normalize(V + L);
 		float cNdotH = max(dot(N, H), 0.0);
-		float intensity = pow(cNdotH, (1.0 - roughness) * 256.0);
+		float cVdotH = max(dot(V, H), 0.0);
+		float cLdotH = max(dot(L, H), 0.0);
+		float shininess = exp2( 15.0 * (1.0 - roughness) + 1.0 ) * 0.25;
+		float blinn = pow( cNdotH, shininess );
+		blinn *= (shininess + 8.0) / (8.0 * 3.141592654);
+		float intensity = ( blinn ) / max( 4.0 * cNdotV * cNdotL, 0.75 );
+
 		specular_light += light_color * intensity * specular_blob_intensity * attenuation;
 
 #elif defined(SPECULAR_PHONG)
 
 		vec3 R = normalize(-reflect(L, N));
 		float cRdotV = max(0.0, dot(R, V));
-		float intensity = pow(cRdotV, (1.0 - roughness) * 256.0);
+		float shininess = exp2( 15.0 * (1.0 - roughness) + 1.0 ) * 0.25;
+		float phong = pow( cRdotV, shininess );
+		phong *= (shininess + 8.0) / (8.0 * 3.141592654);
+		float intensity = ( phong ) / max( 4.0 * cNdotV * cNdotL, 0.75 );
+
 		specular_light += light_color * intensity * specular_blob_intensity * attenuation;
 
+
 #elif defined(SPECULAR_TOON)
 
 		vec3 R = normalize(-reflect(L, N));
@@ -1070,11 +1086,11 @@ LIGHT_SHADER_CODE
 		float G = G_GGX_2cos(cNdotL, alpha) * G_GGX_2cos(cNdotV, alpha);
 #endif
 		// F
-		float F0 = 1.0; // FIXME
-		float cLdotH5 = SchlickFresnel(cLdotH);
-		float F = mix(cLdotH5, 1.0, F0);
+		//float F0 = 1.0;
+		//float cLdotH5 = SchlickFresnel(cLdotH);
+		//float F = mix(cLdotH5, 1.0, F0);
 
-		float specular_brdf_NL = cNdotL * D * F * G;
+		float specular_brdf_NL = cNdotL * D /* F */ * G;
 
 		specular_light += specular_brdf_NL * light_color * specular_blob_intensity * attenuation;
 #endif