mirror/godot
2
0
Fork 0
mirror of https://github.com/godotengine/godot.git synced 2024-12-21 10:25:24 +08:00
Code Issues Packages Projects Releases Wiki Activity
85380c5207
godot/servers/rendering/renderer_storage.h

660 lines
32 KiB
C++
Raw Normal View History

RenderingServer reorganization
2020-12-05 02:26:24 +08:00
/*************************************************************************/
/* renderer_storage.h */
/*************************************************************************/
/* This file is part of: */
/* GODOT ENGINE */
/* https://godotengine.org */
/*************************************************************************/
Update copyright statements to 2021 Happy new year to the wonderful Godot community! 2020 has been a tough year for most of us personally, but a good year for Godot development nonetheless with a huge amount of work done towards Godot 4.0 and great improvements backported to the long-lived 3.2 branch. We've had close to 400 contributors to engine code this year, authoring near 7,000 commit! (And that's only for the `master` branch and for the engine code, there's a lot more when counting docs, demos and other first-party repos.) Here's to a great year 2021 for all Godot users 🎆
2021-01-02 03:13:46 +08:00
/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */
/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
/* */
/* Permission is hereby granted, free of charge, to any person obtaining */
/* a copy of this software and associated documentation files (the */
/* "Software"), to deal in the Software without restriction, including */
/* without limitation the rights to use, copy, modify, merge, publish, */
/* distribute, sublicense, and/or sell copies of the Software, and to */
/* permit persons to whom the Software is furnished to do so, subject to */
/* the following conditions: */
/* */
/* The above copyright notice and this permission notice shall be */
/* included in all copies or substantial portions of the Software. */
/* */
/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */
/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */
/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/
/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */
/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */
/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */
/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */
/*************************************************************************/
#ifndef RENDERINGSERVERSTORAGE_H
#define RENDERINGSERVERSTORAGE_H
#include "servers/rendering_server.h"
class RendererStorage {
Color default_clear_color;
public:
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
enum DependencyChangedNotification {
DEPENDENCY_CHANGED_AABB,
DEPENDENCY_CHANGED_MATERIAL,
DEPENDENCY_CHANGED_MESH,
DEPENDENCY_CHANGED_MULTIMESH,
DEPENDENCY_CHANGED_MULTIMESH_VISIBLE_INSTANCES,
Implement Particle Trails -Enable the trails and set the length in seconds -Provide a mesh with a skeleton and a skin -Or, alternatively use one of the built-in TubeTrailMesh/RibbonTrailMesh -Works deterministically -Fixed particle collisions (were broken) -Not working in 2D yet (that will happen next)
2021-04-27 23:43:49 +08:00
DEPENDENCY_CHANGED_PARTICLES,
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
DEPENDENCY_CHANGED_DECAL,
DEPENDENCY_CHANGED_SKELETON_DATA,
DEPENDENCY_CHANGED_SKELETON_BONES,
DEPENDENCY_CHANGED_LIGHT,
Use specialization constants in clustered renderer * Keep track of when projector, softshadow or directional sofshadow were enabled. * Enable them via specializaton constant where it makes sense. * Re-implements soft shadows. * Re-implements light projectors.
2021-07-13 07:32:05 +08:00
DEPENDENCY_CHANGED_LIGHT_SOFT_SHADOW_AND_PROJECTOR,
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
DEPENDENCY_CHANGED_REFLECTION_PROBE,
};
struct DependencyTracker;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
protected:
struct Dependency {
void changed_notify(DependencyChangedNotification p_notification);
void deleted_notify(const RID &p_rid);
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
~Dependency();
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
private:
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
friend struct DependencyTracker;
Map<DependencyTracker *, uint32_t> instances;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
};
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
public:
struct DependencyTracker {
void *userdata = nullptr;
typedef void (*ChangedCallback)(DependencyChangedNotification, DependencyTracker *);
typedef void (*DeletedCallback)(const RID &, DependencyTracker *);
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
ChangedCallback changed_callback = nullptr;
DeletedCallback deleted_callback = nullptr;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
void update_begin() { // call before updating dependencies
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
instance_version++;
}
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
void update_dependency(Dependency *p_dependency) { //called internally, can't be used directly, use update functions in Storage
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
dependencies.insert(p_dependency);
p_dependency->instances[this] = instance_version;
}
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
void update_end() { //call after updating dependencies
List<Pair<Dependency *, Map<DependencyTracker *, uint32_t>::Element *>> to_clean_up;
for (Set<Dependency *>::Element *E = dependencies.front(); E; E = E->next()) {
Dependency *dep = E->get();
Map<DependencyTracker *, uint32_t>::Element *F = dep->instances.find(this);
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
ERR_CONTINUE(!F);
if (F->get() != instance_version) {
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
Pair<Dependency *, Map<DependencyTracker *, uint32_t>::Element *> p;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
p.first = dep;
p.second = F;
to_clean_up.push_back(p);
}
}
while (to_clean_up.size()) {
to_clean_up.front()->get().first->instances.erase(to_clean_up.front()->get().second);
Rewrote how barriers work for faster rendering -Added more finegrained control in RenderingDevice API -Optimized barriers (use less ones for thee same) -General optimizations -Shadows render all together unbarriered -GI can render together with shadows. -SDFGI can render together with depth-preoass. -General fixes -Added GPU detection
2021-02-03 03:51:36 +08:00
dependencies.erase(to_clean_up.front()->get().first);
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
to_clean_up.pop_front();
}
}
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
void clear() { // clear all dependencies
for (Set<Dependency *>::Element *E = dependencies.front(); E; E = E->next()) {
Dependency *dep = E->get();
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
dep->instances.erase(this);
}
dependencies.clear();
}
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
~DependencyTracker() { clear(); }
private:
Threaded optimizations to cull and render -Reorganize thread work pool for rendering -Fixes to make secondary command buffers to work (disabled because they need more testing)
2021-01-05 04:00:44 +08:00
friend struct Dependency;
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
uint32_t instance_version = 0;
Set<Dependency *> dependencies;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
};
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual bool can_create_resources_async() const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
/* TEXTURE API */
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual RID texture_allocate() = 0;
virtual void texture_2d_initialize(RID p_texture, const Ref<Image> &p_image) = 0;
virtual void texture_2d_layered_initialize(RID p_texture, const Vector<Ref<Image>> &p_layers, RS::TextureLayeredType p_layered_type) = 0;
virtual void texture_3d_initialize(RID p_texture, Image::Format, int p_width, int p_height, int p_depth, bool p_mipmaps, const Vector<Ref<Image>> &p_data) = 0;
virtual void texture_proxy_initialize(RID p_texture, RID p_base) = 0; //all slices, then all the mipmaps, must be coherent
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void texture_2d_update(RID p_texture, const Ref<Image> &p_image, int p_layer = 0) = 0;
virtual void texture_3d_update(RID p_texture, const Vector<Ref<Image>> &p_data) = 0;
virtual void texture_proxy_update(RID p_proxy, RID p_base) = 0;
//these two APIs can be used together or in combination with the others.
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual void texture_2d_placeholder_initialize(RID p_texture) = 0;
virtual void texture_2d_layered_placeholder_initialize(RID p_texture, RenderingServer::TextureLayeredType p_layered_type) = 0;
virtual void texture_3d_placeholder_initialize(RID p_texture) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual Ref<Image> texture_2d_get(RID p_texture) const = 0;
virtual Ref<Image> texture_2d_layer_get(RID p_texture, int p_layer) const = 0;
virtual Vector<Ref<Image>> texture_3d_get(RID p_texture) const = 0;
virtual void texture_replace(RID p_texture, RID p_by_texture) = 0;
virtual void texture_set_size_override(RID p_texture, int p_width, int p_height) = 0;
virtual void texture_set_path(RID p_texture, const String &p_path) = 0;
virtual String texture_get_path(RID p_texture) const = 0;
virtual void texture_set_detect_3d_callback(RID p_texture, RS::TextureDetectCallback p_callback, void *p_userdata) = 0;
virtual void texture_set_detect_normal_callback(RID p_texture, RS::TextureDetectCallback p_callback, void *p_userdata) = 0;
virtual void texture_set_detect_roughness_callback(RID p_texture, RS::TextureDetectRoughnessCallback p_callback, void *p_userdata) = 0;
virtual void texture_debug_usage(List<RS::TextureInfo> *r_info) = 0;
virtual void texture_set_force_redraw_if_visible(RID p_texture, bool p_enable) = 0;
virtual Size2 texture_size_with_proxy(RID p_proxy) = 0;
virtual void texture_add_to_decal_atlas(RID p_texture, bool p_panorama_to_dp = false) = 0;
virtual void texture_remove_from_decal_atlas(RID p_texture, bool p_panorama_to_dp = false) = 0;
/* CANVAS TEXTURE API */
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual RID canvas_texture_allocate() = 0;
virtual void canvas_texture_initialize(RID p_rid) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void canvas_texture_set_channel(RID p_canvas_texture, RS::CanvasTextureChannel p_channel, RID p_texture) = 0;
virtual void canvas_texture_set_shading_parameters(RID p_canvas_texture, const Color &p_base_color, float p_shininess) = 0;
virtual void canvas_texture_set_texture_filter(RID p_item, RS::CanvasItemTextureFilter p_filter) = 0;
virtual void canvas_texture_set_texture_repeat(RID p_item, RS::CanvasItemTextureRepeat p_repeat) = 0;
/* SHADER API */
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual RID shader_allocate() = 0;
virtual void shader_initialize(RID p_rid) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void shader_set_code(RID p_shader, const String &p_code) = 0;
virtual String shader_get_code(RID p_shader) const = 0;
virtual void shader_get_param_list(RID p_shader, List<PropertyInfo> *p_param_list) const = 0;
Fix default_texture_param in shader pipeline to support uniform arrays
2021-10-17 19:38:26 +08:00
virtual void shader_set_default_texture_param(RID p_shader, const StringName &p_name, RID p_texture, int p_index) = 0;
virtual RID shader_get_default_texture_param(RID p_shader, const StringName &p_name, int p_index) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual Variant shader_get_param_default(RID p_material, const StringName &p_param) const = 0;
Added ability to visualize native shaders
2021-01-06 07:01:50 +08:00
virtual RS::ShaderNativeSourceCode shader_get_native_source_code(RID p_shader) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
/* COMMON MATERIAL API */
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual RID material_allocate() = 0;
virtual void material_initialize(RID p_rid) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void material_set_render_priority(RID p_material, int priority) = 0;
virtual void material_set_shader(RID p_shader_material, RID p_shader) = 0;
virtual void material_set_param(RID p_material, const StringName &p_param, const Variant &p_value) = 0;
virtual Variant material_get_param(RID p_material, const StringName &p_param) const = 0;
virtual void material_set_next_pass(RID p_material, RID p_next_material) = 0;
virtual bool material_is_animated(RID p_material) = 0;
virtual bool material_casts_shadows(RID p_material) = 0;
struct InstanceShaderParam {
PropertyInfo info;
int index;
Variant default_value;
};
virtual void material_get_instance_shader_parameters(RID p_material, List<InstanceShaderParam> *r_parameters) = 0;
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
virtual void material_update_dependency(RID p_material, DependencyTracker *p_instance) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
/* MESH API */
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual RID mesh_allocate() = 0;
virtual void mesh_initialize(RID p_rid) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Reimplement skeletons and blend shapes Uses compute shaders, which only once, on demand, and all in parallel.
2020-12-16 22:07:08 +08:00
virtual void mesh_set_blend_shape_count(RID p_mesh, int p_blend_shape_count) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
/// Returns stride
virtual void mesh_add_surface(RID p_mesh, const RS::SurfaceData &p_surface) = 0;
virtual int mesh_get_blend_shape_count(RID p_mesh) const = 0;
virtual void mesh_set_blend_shape_mode(RID p_mesh, RS::BlendShapeMode p_mode) = 0;
virtual RS::BlendShapeMode mesh_get_blend_shape_mode(RID p_mesh) const = 0;
Deprecate ImmediateGeometry * Removed entirely from RenderingServer. * Replaced by ImmediateMesh resource. * ImmediateMesh replaces ImmediateGeometry, but could use more optimization in the future. * Sprite3D and AnimatedSprite3D work again, ported from Godot 3.x (though a lot of work was needed to adapt them to Godot 4). * RootMotionView works again. * Polygon3D editor works again.
2021-06-30 09:55:11 +08:00
virtual void mesh_surface_update_vertex_region(RID p_mesh, int p_surface, int p_offset, const Vector<uint8_t> &p_data) = 0;
virtual void mesh_surface_update_attribute_region(RID p_mesh, int p_surface, int p_offset, const Vector<uint8_t> &p_data) = 0;
virtual void mesh_surface_update_skin_region(RID p_mesh, int p_surface, int p_offset, const Vector<uint8_t> &p_data) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void mesh_surface_set_material(RID p_mesh, int p_surface, RID p_material) = 0;
virtual RID mesh_surface_get_material(RID p_mesh, int p_surface) const = 0;
virtual RS::SurfaceData mesh_get_surface(RID p_mesh, int p_surface) const = 0;
virtual int mesh_get_surface_count(RID p_mesh) const = 0;
virtual void mesh_set_custom_aabb(RID p_mesh, const AABB &p_aabb) = 0;
virtual AABB mesh_get_custom_aabb(RID p_mesh) const = 0;
virtual AABB mesh_get_aabb(RID p_mesh, RID p_skeleton = RID()) = 0;
Implement shadow meshes -When importing, a vertex-only version of the mesh is created. -This version is used when rendering shadows, and improves performance by reducing bandwidth -It's automatic, but can optionally be used by users, in case they want to make special versions of geometry for shadow casting.
2021-01-25 23:20:11 +08:00
virtual void mesh_set_shadow_mesh(RID p_mesh, RID p_shadow_mesh) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void mesh_clear(RID p_mesh) = 0;
Reimplement skeletons and blend shapes Uses compute shaders, which only once, on demand, and all in parallel.
2020-12-16 22:07:08 +08:00
virtual bool mesh_needs_instance(RID p_mesh, bool p_has_skeleton) = 0;
/* MESH INSTANCE */
virtual RID mesh_instance_create(RID p_base) = 0;
virtual void mesh_instance_set_skeleton(RID p_mesh_instance, RID p_skeleton) = 0;
virtual void mesh_instance_set_blend_shape_weight(RID p_mesh_instance, int p_shape, float p_weight) = 0;
virtual void mesh_instance_check_for_update(RID p_mesh_instance) = 0;
virtual void update_mesh_instances() = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
/* MULTIMESH API */
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual RID multimesh_allocate() = 0;
virtual void multimesh_initialize(RID p_rid) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual void multimesh_allocate_data(RID p_multimesh, int p_instances, RS::MultimeshTransformFormat p_transform_format, bool p_use_colors = false, bool p_use_custom_data = false) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual int multimesh_get_instance_count(RID p_multimesh) const = 0;
virtual void multimesh_set_mesh(RID p_multimesh, RID p_mesh) = 0;
Rename Transform to Transform3D in core
2020-10-17 13:08:21 +08:00
virtual void multimesh_instance_set_transform(RID p_multimesh, int p_index, const Transform3D &p_transform) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void multimesh_instance_set_transform_2d(RID p_multimesh, int p_index, const Transform2D &p_transform) = 0;
virtual void multimesh_instance_set_color(RID p_multimesh, int p_index, const Color &p_color) = 0;
virtual void multimesh_instance_set_custom_data(RID p_multimesh, int p_index, const Color &p_color) = 0;
virtual RID multimesh_get_mesh(RID p_multimesh) const = 0;
Rename Transform to Transform3D in core
2020-10-17 13:08:21 +08:00
virtual Transform3D multimesh_instance_get_transform(RID p_multimesh, int p_index) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual Transform2D multimesh_instance_get_transform_2d(RID p_multimesh, int p_index) const = 0;
virtual Color multimesh_instance_get_color(RID p_multimesh, int p_index) const = 0;
virtual Color multimesh_instance_get_custom_data(RID p_multimesh, int p_index) const = 0;
virtual void multimesh_set_buffer(RID p_multimesh, const Vector<float> &p_buffer) = 0;
virtual Vector<float> multimesh_get_buffer(RID p_multimesh) const = 0;
virtual void multimesh_set_visible_instances(RID p_multimesh, int p_visible) = 0;
virtual int multimesh_get_visible_instances(RID p_multimesh) const = 0;
virtual AABB multimesh_get_aabb(RID p_multimesh) const = 0;
/* SKELETON API */
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual RID skeleton_allocate() = 0;
virtual void skeleton_initialize(RID p_rid) = 0;
virtual void skeleton_allocate_data(RID p_skeleton, int p_bones, bool p_2d_skeleton = false) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual int skeleton_get_bone_count(RID p_skeleton) const = 0;
Rename Transform to Transform3D in core
2020-10-17 13:08:21 +08:00
virtual void skeleton_bone_set_transform(RID p_skeleton, int p_bone, const Transform3D &p_transform) = 0;
virtual Transform3D skeleton_bone_get_transform(RID p_skeleton, int p_bone) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void skeleton_bone_set_transform_2d(RID p_skeleton, int p_bone, const Transform2D &p_transform) = 0;
virtual Transform2D skeleton_bone_get_transform_2d(RID p_skeleton, int p_bone) const = 0;
virtual void skeleton_set_base_transform_2d(RID p_skeleton, const Transform2D &p_base_transform) = 0;
/* Light API */
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual RID directional_light_allocate() = 0;
virtual void directional_light_initialize(RID p_rid) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual RID omni_light_allocate() = 0;
virtual void omni_light_initialize(RID p_rid) = 0;
virtual RID spot_light_allocate() = 0;
virtual void spot_light_initialize(RID p_rid) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void light_set_color(RID p_light, const Color &p_color) = 0;
virtual void light_set_param(RID p_light, RS::LightParam p_param, float p_value) = 0;
virtual void light_set_shadow(RID p_light, bool p_enabled) = 0;
virtual void light_set_shadow_color(RID p_light, const Color &p_color) = 0;
virtual void light_set_projector(RID p_light, RID p_texture) = 0;
virtual void light_set_negative(RID p_light, bool p_enable) = 0;
virtual void light_set_cull_mask(RID p_light, uint32_t p_mask) = 0;
virtual void light_set_reverse_cull_face_mode(RID p_light, bool p_enabled) = 0;
virtual void light_set_bake_mode(RID p_light, RS::LightBakeMode p_bake_mode) = 0;
virtual void light_set_max_sdfgi_cascade(RID p_light, uint32_t p_cascade) = 0;
virtual void light_omni_set_shadow_mode(RID p_light, RS::LightOmniShadowMode p_mode) = 0;
virtual void light_directional_set_shadow_mode(RID p_light, RS::LightDirectionalShadowMode p_mode) = 0;
virtual void light_directional_set_blend_splits(RID p_light, bool p_enable) = 0;
virtual bool light_directional_get_blend_splits(RID p_light) const = 0;
virtual void light_directional_set_sky_only(RID p_light, bool p_sky_only) = 0;
virtual bool light_directional_is_sky_only(RID p_light) const = 0;
virtual RS::LightDirectionalShadowMode light_directional_get_shadow_mode(RID p_light) = 0;
virtual RS::LightOmniShadowMode light_omni_get_shadow_mode(RID p_light) = 0;
virtual bool light_has_shadow(RID p_light) const = 0;
Use specialization constants in clustered renderer * Keep track of when projector, softshadow or directional sofshadow were enabled. * Enable them via specializaton constant where it makes sense. * Re-implements soft shadows. * Re-implements light projectors.
2021-07-13 07:32:05 +08:00
virtual bool light_has_projector(RID p_light) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual RS::LightType light_get_type(RID p_light) const = 0;
virtual AABB light_get_aabb(RID p_light) const = 0;
virtual float light_get_param(RID p_light, RS::LightParam p_param) = 0;
virtual Color light_get_color(RID p_light) = 0;
virtual RS::LightBakeMode light_get_bake_mode(RID p_light) = 0;
virtual uint32_t light_get_max_sdfgi_cascade(RID p_light) = 0;
virtual uint64_t light_get_version(RID p_light) const = 0;
/* PROBE API */
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual RID reflection_probe_allocate() = 0;
virtual void reflection_probe_initialize(RID p_rid) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void reflection_probe_set_update_mode(RID p_probe, RS::ReflectionProbeUpdateMode p_mode) = 0;
virtual void reflection_probe_set_resolution(RID p_probe, int p_resolution) = 0;
virtual void reflection_probe_set_intensity(RID p_probe, float p_intensity) = 0;
virtual void reflection_probe_set_ambient_mode(RID p_probe, RS::ReflectionProbeAmbientMode p_mode) = 0;
virtual void reflection_probe_set_ambient_color(RID p_probe, const Color &p_color) = 0;
virtual void reflection_probe_set_ambient_energy(RID p_probe, float p_energy) = 0;
virtual void reflection_probe_set_max_distance(RID p_probe, float p_distance) = 0;
virtual void reflection_probe_set_extents(RID p_probe, const Vector3 &p_extents) = 0;
virtual void reflection_probe_set_origin_offset(RID p_probe, const Vector3 &p_offset) = 0;
virtual void reflection_probe_set_as_interior(RID p_probe, bool p_enable) = 0;
virtual void reflection_probe_set_enable_box_projection(RID p_probe, bool p_enable) = 0;
virtual void reflection_probe_set_enable_shadows(RID p_probe, bool p_enable) = 0;
virtual void reflection_probe_set_cull_mask(RID p_probe, uint32_t p_layers) = 0;
Implement automatic LOD (Level of Detail) -Happens on import by default for all models -Just works (tm) -Biasing can be later adjusted per node or per viewport (as well as globally) -Disabled AABB.get_support test because its broken
2020-12-18 02:56:59 +08:00
virtual void reflection_probe_set_lod_threshold(RID p_probe, float p_ratio) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual AABB reflection_probe_get_aabb(RID p_probe) const = 0;
virtual RS::ReflectionProbeUpdateMode reflection_probe_get_update_mode(RID p_probe) const = 0;
virtual uint32_t reflection_probe_get_cull_mask(RID p_probe) const = 0;
virtual Vector3 reflection_probe_get_extents(RID p_probe) const = 0;
virtual Vector3 reflection_probe_get_origin_offset(RID p_probe) const = 0;
virtual float reflection_probe_get_origin_max_distance(RID p_probe) const = 0;
virtual bool reflection_probe_renders_shadows(RID p_probe) const = 0;
Implement automatic LOD (Level of Detail) -Happens on import by default for all models -Just works (tm) -Biasing can be later adjusted per node or per viewport (as well as globally) -Disabled AABB.get_support test because its broken
2020-12-18 02:56:59 +08:00
virtual float reflection_probe_get_lod_threshold(RID p_probe) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rewrite render code to be more cache and thread friendly.
2021-01-04 20:33:25 +08:00
virtual void base_update_dependency(RID p_base, DependencyTracker *p_instance) = 0;
virtual void skeleton_update_dependency(RID p_base, DependencyTracker *p_instance) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
/* DECAL API */
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual RID decal_allocate() = 0;
virtual void decal_initialize(RID p_rid) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void decal_set_extents(RID p_decal, const Vector3 &p_extents) = 0;
virtual void decal_set_texture(RID p_decal, RS::DecalTexture p_type, RID p_texture) = 0;
virtual void decal_set_emission_energy(RID p_decal, float p_energy) = 0;
virtual void decal_set_albedo_mix(RID p_decal, float p_mix) = 0;
virtual void decal_set_modulate(RID p_decal, const Color &p_modulate) = 0;
virtual void decal_set_cull_mask(RID p_decal, uint32_t p_layers) = 0;
virtual void decal_set_distance_fade(RID p_decal, bool p_enabled, float p_begin, float p_length) = 0;
virtual void decal_set_fade(RID p_decal, float p_above, float p_below) = 0;
virtual void decal_set_normal_fade(RID p_decal, float p_fade) = 0;
virtual AABB decal_get_aabb(RID p_decal) const = 0;
Rename GI Classes * GIProbe is now VoxelGI * BakedLightmap is now LightmapGI As godot adds more ways to provide GI (as an example, SDFGI in 4.0), the different techniques (which have different pros/cons) need to be properly named to avoid confusion.
2021-06-05 06:47:26 +08:00
/* VOXEL GI API */
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rename GI Classes * GIProbe is now VoxelGI * BakedLightmap is now LightmapGI As godot adds more ways to provide GI (as an example, SDFGI in 4.0), the different techniques (which have different pros/cons) need to be properly named to avoid confusion.
2021-06-05 06:47:26 +08:00
virtual RID voxel_gi_allocate() = 0;
virtual void voxel_gi_initialize(RID p_rid) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rename GI Classes * GIProbe is now VoxelGI * BakedLightmap is now LightmapGI As godot adds more ways to provide GI (as an example, SDFGI in 4.0), the different techniques (which have different pros/cons) need to be properly named to avoid confusion.
2021-06-05 06:47:26 +08:00
virtual void voxel_gi_allocate_data(RID p_voxel_gi, const Transform3D &p_to_cell_xform, const AABB &p_aabb, const Vector3i &p_octree_size, const Vector<uint8_t> &p_octree_cells, const Vector<uint8_t> &p_data_cells, const Vector<uint8_t> &p_distance_field, const Vector<int> &p_level_counts) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rename GI Classes * GIProbe is now VoxelGI * BakedLightmap is now LightmapGI As godot adds more ways to provide GI (as an example, SDFGI in 4.0), the different techniques (which have different pros/cons) need to be properly named to avoid confusion.
2021-06-05 06:47:26 +08:00
virtual AABB voxel_gi_get_bounds(RID p_voxel_gi) const = 0;
virtual Vector3i voxel_gi_get_octree_size(RID p_voxel_gi) const = 0;
virtual Vector<uint8_t> voxel_gi_get_octree_cells(RID p_voxel_gi) const = 0;
virtual Vector<uint8_t> voxel_gi_get_data_cells(RID p_voxel_gi) const = 0;
virtual Vector<uint8_t> voxel_gi_get_distance_field(RID p_voxel_gi) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rename GI Classes * GIProbe is now VoxelGI * BakedLightmap is now LightmapGI As godot adds more ways to provide GI (as an example, SDFGI in 4.0), the different techniques (which have different pros/cons) need to be properly named to avoid confusion.
2021-06-05 06:47:26 +08:00
virtual Vector<int> voxel_gi_get_level_counts(RID p_voxel_gi) const = 0;
virtual Transform3D voxel_gi_get_to_cell_xform(RID p_voxel_gi) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rename GI Classes * GIProbe is now VoxelGI * BakedLightmap is now LightmapGI As godot adds more ways to provide GI (as an example, SDFGI in 4.0), the different techniques (which have different pros/cons) need to be properly named to avoid confusion.
2021-06-05 06:47:26 +08:00
virtual void voxel_gi_set_dynamic_range(RID p_voxel_gi, float p_range) = 0;
virtual float voxel_gi_get_dynamic_range(RID p_voxel_gi) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rename GI Classes * GIProbe is now VoxelGI * BakedLightmap is now LightmapGI As godot adds more ways to provide GI (as an example, SDFGI in 4.0), the different techniques (which have different pros/cons) need to be properly named to avoid confusion.
2021-06-05 06:47:26 +08:00
virtual void voxel_gi_set_propagation(RID p_voxel_gi, float p_range) = 0;
virtual float voxel_gi_get_propagation(RID p_voxel_gi) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rename GI Classes * GIProbe is now VoxelGI * BakedLightmap is now LightmapGI As godot adds more ways to provide GI (as an example, SDFGI in 4.0), the different techniques (which have different pros/cons) need to be properly named to avoid confusion.
2021-06-05 06:47:26 +08:00
virtual void voxel_gi_set_energy(RID p_voxel_gi, float p_energy) = 0;
virtual float voxel_gi_get_energy(RID p_voxel_gi) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rename GI Classes * GIProbe is now VoxelGI * BakedLightmap is now LightmapGI As godot adds more ways to provide GI (as an example, SDFGI in 4.0), the different techniques (which have different pros/cons) need to be properly named to avoid confusion.
2021-06-05 06:47:26 +08:00
virtual void voxel_gi_set_bias(RID p_voxel_gi, float p_bias) = 0;
virtual float voxel_gi_get_bias(RID p_voxel_gi) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rename GI Classes * GIProbe is now VoxelGI * BakedLightmap is now LightmapGI As godot adds more ways to provide GI (as an example, SDFGI in 4.0), the different techniques (which have different pros/cons) need to be properly named to avoid confusion.
2021-06-05 06:47:26 +08:00
virtual void voxel_gi_set_normal_bias(RID p_voxel_gi, float p_range) = 0;
virtual float voxel_gi_get_normal_bias(RID p_voxel_gi) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rename GI Classes * GIProbe is now VoxelGI * BakedLightmap is now LightmapGI As godot adds more ways to provide GI (as an example, SDFGI in 4.0), the different techniques (which have different pros/cons) need to be properly named to avoid confusion.
2021-06-05 06:47:26 +08:00
virtual void voxel_gi_set_interior(RID p_voxel_gi, bool p_enable) = 0;
virtual bool voxel_gi_is_interior(RID p_voxel_gi) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rename GI Classes * GIProbe is now VoxelGI * BakedLightmap is now LightmapGI As godot adds more ways to provide GI (as an example, SDFGI in 4.0), the different techniques (which have different pros/cons) need to be properly named to avoid confusion.
2021-06-05 06:47:26 +08:00
virtual void voxel_gi_set_use_two_bounces(RID p_voxel_gi, bool p_enable) = 0;
virtual bool voxel_gi_is_using_two_bounces(RID p_voxel_gi) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rename GI Classes * GIProbe is now VoxelGI * BakedLightmap is now LightmapGI As godot adds more ways to provide GI (as an example, SDFGI in 4.0), the different techniques (which have different pros/cons) need to be properly named to avoid confusion.
2021-06-05 06:47:26 +08:00
virtual void voxel_gi_set_anisotropy_strength(RID p_voxel_gi, float p_strength) = 0;
virtual float voxel_gi_get_anisotropy_strength(RID p_voxel_gi) const = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Rename GI Classes * GIProbe is now VoxelGI * BakedLightmap is now LightmapGI As godot adds more ways to provide GI (as an example, SDFGI in 4.0), the different techniques (which have different pros/cons) need to be properly named to avoid confusion.
2021-06-05 06:47:26 +08:00
virtual uint32_t voxel_gi_get_version(RID p_probe) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
/* LIGHTMAP */
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual RID lightmap_allocate() = 0;
virtual void lightmap_initialize(RID p_rid) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void lightmap_set_textures(RID p_lightmap, RID p_light, bool p_uses_spherical_haromics) = 0;
virtual void lightmap_set_probe_bounds(RID p_lightmap, const AABB &p_bounds) = 0;
virtual void lightmap_set_probe_interior(RID p_lightmap, bool p_interior) = 0;
virtual void lightmap_set_probe_capture_data(RID p_lightmap, const PackedVector3Array &p_points, const PackedColorArray &p_point_sh, const PackedInt32Array &p_tetrahedra, const PackedInt32Array &p_bsp_tree) = 0;
virtual PackedVector3Array lightmap_get_probe_capture_points(RID p_lightmap) const = 0;
virtual PackedColorArray lightmap_get_probe_capture_sh(RID p_lightmap) const = 0;
virtual PackedInt32Array lightmap_get_probe_capture_tetrahedra(RID p_lightmap) const = 0;
virtual PackedInt32Array lightmap_get_probe_capture_bsp_tree(RID p_lightmap) const = 0;
virtual AABB lightmap_get_aabb(RID p_lightmap) const = 0;
virtual void lightmap_tap_sh_light(RID p_lightmap, const Vector3 &p_point, Color *r_sh) = 0;
virtual bool lightmap_is_interior(RID p_lightmap) const = 0;
virtual void lightmap_set_probe_capture_update_speed(float p_speed) = 0;
virtual float lightmap_get_probe_capture_update_speed() const = 0;
/* PARTICLES */
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual RID particles_allocate() = 0;
virtual void particles_initialize(RID p_rid) = 0;
Fixes missng 2D engine bits -Mesh2D now works -MultiMesh2D now works -Polygon2D now works -Added hooks for processing 2D particles -Skeleton2D now works 2D particles still not working, but stuff needed for it is now implemented.
2021-05-11 00:12:44 +08:00
virtual void particles_set_mode(RID p_particles, RS::ParticlesMode p_mode) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void particles_set_emitting(RID p_particles, bool p_emitting) = 0;
virtual bool particles_get_emitting(RID p_particles) = 0;
virtual void particles_set_amount(RID p_particles, int p_amount) = 0;
Use doubles for time in many other places
2021-02-02 10:16:37 +08:00
virtual void particles_set_lifetime(RID p_particles, double p_lifetime) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void particles_set_one_shot(RID p_particles, bool p_one_shot) = 0;
Use doubles for time in many other places
2021-02-02 10:16:37 +08:00
virtual void particles_set_pre_process_time(RID p_particles, double p_time) = 0;
Use real_t and double where appropriate in Particles
2021-08-12 22:07:47 +08:00
virtual void particles_set_explosiveness_ratio(RID p_particles, real_t p_ratio) = 0;
virtual void particles_set_randomness_ratio(RID p_particles, real_t p_ratio) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void particles_set_custom_aabb(RID p_particles, const AABB &p_aabb) = 0;
Use doubles for time in many other places
2021-02-02 10:16:37 +08:00
virtual void particles_set_speed_scale(RID p_particles, double p_scale) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void particles_set_use_local_coordinates(RID p_particles, bool p_enable) = 0;
virtual void particles_set_process_material(RID p_particles, RID p_material) = 0;
virtual void particles_set_fixed_fps(RID p_particles, int p_fps) = 0;
Implement Particle Trails -Enable the trails and set the length in seconds -Provide a mesh with a skeleton and a skin -Or, alternatively use one of the built-in TubeTrailMesh/RibbonTrailMesh -Works deterministically -Fixed particle collisions (were broken) -Not working in 2D yet (that will happen next)
2021-04-27 23:43:49 +08:00
virtual void particles_set_interpolate(RID p_particles, bool p_enable) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void particles_set_fractional_delta(RID p_particles, bool p_enable) = 0;
Use real_t and double where appropriate in Particles
2021-08-12 22:07:47 +08:00
virtual void particles_set_collision_base_size(RID p_particles, real_t p_size) = 0;
Implement Particle Trails -Enable the trails and set the length in seconds -Provide a mesh with a skeleton and a skin -Or, alternatively use one of the built-in TubeTrailMesh/RibbonTrailMesh -Works deterministically -Fixed particle collisions (were broken) -Not working in 2D yet (that will happen next)
2021-04-27 23:43:49 +08:00
virtual void particles_set_transform_align(RID p_particles, RS::ParticlesTransformAlign p_transform_align) = 0;
Use real_t and double where appropriate in Particles
2021-08-12 22:07:47 +08:00
virtual void particles_set_trails(RID p_particles, bool p_enable, double p_length) = 0;
Rename Transform to Transform3D in core
2020-10-17 13:08:21 +08:00
virtual void particles_set_trail_bind_poses(RID p_particles, const Vector<Transform3D> &p_bind_poses) = 0;
Implement Particle Trails -Enable the trails and set the length in seconds -Provide a mesh with a skeleton and a skin -Or, alternatively use one of the built-in TubeTrailMesh/RibbonTrailMesh -Works deterministically -Fixed particle collisions (were broken) -Not working in 2D yet (that will happen next)
2021-04-27 23:43:49 +08:00
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void particles_restart(RID p_particles) = 0;
Rename Transform to Transform3D in core
2020-10-17 13:08:21 +08:00
virtual void particles_emit(RID p_particles, const Transform3D &p_transform, const Vector3 &p_velocity, const Color &p_color, const Color &p_custom, uint32_t p_emit_flags) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void particles_set_subemitter(RID p_particles, RID p_subemitter_particles) = 0;
virtual bool particles_is_inactive(RID p_particles) const = 0;
virtual void particles_set_draw_order(RID p_particles, RS::ParticlesDrawOrder p_order) = 0;
virtual void particles_set_draw_passes(RID p_particles, int p_count) = 0;
virtual void particles_set_draw_pass_mesh(RID p_particles, int p_pass, RID p_mesh) = 0;
virtual void particles_request_process(RID p_particles) = 0;
virtual AABB particles_get_current_aabb(RID p_particles) = 0;
virtual AABB particles_get_aabb(RID p_particles) const = 0;
Rename Transform to Transform3D in core
2020-10-17 13:08:21 +08:00
virtual void particles_set_emission_transform(RID p_particles, const Transform3D &p_transform) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual int particles_get_draw_passes(RID p_particles) const = 0;
virtual RID particles_get_draw_pass_mesh(RID p_particles, int p_pass) const = 0;
Implement Particle Trails -Enable the trails and set the length in seconds -Provide a mesh with a skeleton and a skin -Or, alternatively use one of the built-in TubeTrailMesh/RibbonTrailMesh -Works deterministically -Fixed particle collisions (were broken) -Not working in 2D yet (that will happen next)
2021-04-27 23:43:49 +08:00
virtual void particles_set_view_axis(RID p_particles, const Vector3 &p_axis, const Vector3 &p_up_axis) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Reorganize renderer code. So it can hopefully be made more cache efficient afterwards.
2020-12-31 20:42:56 +08:00
virtual void particles_add_collision(RID p_particles, RID p_particles_collision_instance) = 0;
virtual void particles_remove_collision(RID p_particles, RID p_particles_collision_instance) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Support for 2D particles to collide against SDF -Added SDF collision support for 2D particles -Changed the SDF generation to be fully signed
2021-05-20 22:25:06 +08:00
virtual void particles_set_canvas_sdf_collision(RID p_particles, bool p_enable, const Transform2D &p_xform, const Rect2 &p_to_screen, RID p_texture) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void update_particles() = 0;
/* PARTICLES COLLISION */
Make Servers truly Thread Safe -Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread). -RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault. -Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory). -3D physics server changed to be made multithread friendly. -Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads. -Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
2021-02-10 00:19:03 +08:00
virtual RID particles_collision_allocate() = 0;
virtual void particles_collision_initialize(RID p_rid) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void particles_collision_set_collision_type(RID p_particles_collision, RS::ParticlesCollisionType p_type) = 0;
virtual void particles_collision_set_cull_mask(RID p_particles_collision, uint32_t p_cull_mask) = 0;
Use real_t and double where appropriate in Particles
2021-08-12 22:07:47 +08:00
virtual void particles_collision_set_sphere_radius(RID p_particles_collision, real_t p_radius) = 0; //for spheres
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void particles_collision_set_box_extents(RID p_particles_collision, const Vector3 &p_extents) = 0; //for non-spheres
Use real_t and double where appropriate in Particles
2021-08-12 22:07:47 +08:00
virtual void particles_collision_set_attractor_strength(RID p_particles_collision, real_t p_strength) = 0;
virtual void particles_collision_set_attractor_directionality(RID p_particles_collision, real_t p_directionality) = 0;
virtual void particles_collision_set_attractor_attenuation(RID p_particles_collision, real_t p_curve) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual void particles_collision_set_field_texture(RID p_particles_collision, RID p_texture) = 0; //for SDF and vector field, heightfield is dynamic
virtual void particles_collision_height_field_update(RID p_particles_collision) = 0; //for SDF and vector field
virtual void particles_collision_set_height_field_resolution(RID p_particles_collision, RS::ParticlesCollisionHeightfieldResolution p_resolution) = 0; //for SDF and vector field
virtual AABB particles_collision_get_aabb(RID p_particles_collision) const = 0;
virtual bool particles_collision_is_heightfield(RID p_particles_collision) const = 0;
virtual RID particles_collision_get_heightfield_framebuffer(RID p_particles_collision) const = 0;
Addition of FogVolumes, FogShaders, FogMaterial, and overhaul of VolumetricFog Co-authored-by: Brian Semrau <brian.semrau@gmail.com>
2021-10-03 19:28:55 +08:00
/* FOG VOLUMES */
virtual RID fog_volume_allocate() = 0;
virtual void fog_volume_initialize(RID p_rid) = 0;
virtual void fog_volume_set_shape(RID p_fog_volume, RS::FogVolumeShape p_shape) = 0;
virtual void fog_volume_set_extents(RID p_fog_volume, const Vector3 &p_extents) = 0;
virtual void fog_volume_set_material(RID p_fog_volume, RID p_material) = 0;
virtual AABB fog_volume_get_aabb(RID p_fog_volume) const = 0;
virtual RS::FogVolumeShape fog_volume_get_shape(RID p_fog_volume) const = 0;
/* VISIBILITY NOTIFIER */
Refactor VisibilityNotifier3D * This is the 3D counterpart to #49632 * Implemented a bit different as 3D works using instancing After merged, both 2D and 3D classes will most likely be renamed in a separate PR to DisplayNotifier2D/3D.
2021-06-17 02:43:02 +08:00
virtual RID visibility_notifier_allocate() = 0;
virtual void visibility_notifier_initialize(RID p_notifier) = 0;
virtual void visibility_notifier_set_aabb(RID p_notifier, const AABB &p_aabb) = 0;
virtual void visibility_notifier_set_callbacks(RID p_notifier, const Callable &p_enter_callbable, const Callable &p_exit_callable) = 0;
virtual AABB visibility_notifier_get_aabb(RID p_notifier) const = 0;
virtual void visibility_notifier_call(RID p_notifier, bool p_enter, bool p_deferred) = 0;
Reorganize renderer code. So it can hopefully be made more cache efficient afterwards.
2020-12-31 20:42:56 +08:00
//used from 2D and 3D
virtual RID particles_collision_instance_create(RID p_collision) = 0;
Rename Transform to Transform3D in core
2020-10-17 13:08:21 +08:00
virtual void particles_collision_instance_set_transform(RID p_collision_instance, const Transform3D &p_transform) = 0;
Reorganize renderer code. So it can hopefully be made more cache efficient afterwards.
2020-12-31 20:42:56 +08:00
virtual void particles_collision_instance_set_active(RID p_collision_instance, bool p_active) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
/* GLOBAL VARIABLES */
virtual void global_variable_add(const StringName &p_name, RS::GlobalVariableType p_type, const Variant &p_value) = 0;
virtual void global_variable_remove(const StringName &p_name) = 0;
virtual Vector<StringName> global_variable_get_list() const = 0;
virtual void global_variable_set(const StringName &p_name, const Variant &p_value) = 0;
virtual void global_variable_set_override(const StringName &p_name, const Variant &p_value) = 0;
virtual Variant global_variable_get(const StringName &p_name) const = 0;
virtual RS::GlobalVariableType global_variable_get_type(const StringName &p_name) const = 0;
virtual void global_variables_load_settings(bool p_load_textures = true) = 0;
virtual void global_variables_clear() = 0;
virtual int32_t global_variables_instance_allocate(RID p_instance) = 0;
virtual void global_variables_instance_free(RID p_instance) = 0;
virtual void global_variables_instance_update(RID p_instance, int p_index, const Variant &p_value) = 0;
/* RENDER TARGET */
enum RenderTargetFlags {
RENDER_TARGET_TRANSPARENT,
RENDER_TARGET_DIRECT_TO_SCREEN,
RENDER_TARGET_FLAG_MAX
};
virtual RID render_target_create() = 0;
virtual void render_target_set_position(RID p_render_target, int p_x, int p_y) = 0;
Add stereoscopic rendering through multiview
2021-05-07 21:19:04 +08:00
virtual void render_target_set_size(RID p_render_target, int p_width, int p_height, uint32_t p_view_count) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual RID render_target_get_texture(RID p_render_target) = 0;
virtual void render_target_set_external_texture(RID p_render_target, unsigned int p_texture_id) = 0;
virtual void render_target_set_flag(RID p_render_target, RenderTargetFlags p_flag, bool p_value) = 0;
virtual bool render_target_was_used(RID p_render_target) = 0;
virtual void render_target_set_as_unused(RID p_render_target) = 0;
virtual void render_target_request_clear(RID p_render_target, const Color &p_clear_color) = 0;
virtual bool render_target_is_clear_requested(RID p_render_target) = 0;
virtual Color render_target_get_clear_request_color(RID p_render_target) = 0;
virtual void render_target_disable_clear_request(RID p_render_target) = 0;
virtual void render_target_do_clear_request(RID p_render_target) = 0;
virtual void render_target_set_sdf_size_and_scale(RID p_render_target, RS::ViewportSDFOversize p_size, RS::ViewportSDFScale p_scale) = 0;
virtual Rect2i render_target_get_sdf_rect(RID p_render_target) const = 0;
Support for 2D particles to collide against SDF -Added SDF collision support for 2D particles -Changed the SDF generation to be fully signed
2021-05-20 22:25:06 +08:00
virtual void render_target_mark_sdf_enabled(RID p_render_target, bool p_enabled) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual RS::InstanceType get_base_type(RID p_rid) const = 0;
virtual bool free(RID p_rid) = 0;
virtual bool has_os_feature(const String &p_feature) const = 0;
virtual void update_dirty_resources() = 0;
virtual void set_debug_generate_wireframes(bool p_generate) = 0;
Fix Render Info * Fixed and redone the process to obtain render information from a viewport * Some stats, such as material changes are too difficult to guess on Vulkan, were removed. * Separated visible and shadow stats, which causes confusion. * Texture, buffer and general video memory can be queried now. * Fixed the performance metrics too.
2021-07-03 07:14:19 +08:00
virtual void update_memory_info() = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
Fix Render Info * Fixed and redone the process to obtain render information from a viewport * Some stats, such as material changes are too difficult to guess on Vulkan, were removed. * Separated visible and shadow stats, which causes confusion. * Texture, buffer and general video memory can be queried now. * Fixed the performance metrics too.
2021-07-03 07:14:19 +08:00
virtual uint64_t get_rendering_info(RS::RenderingInfo p_info) = 0;
RenderingServer reorganization
2020-12-05 02:26:24 +08:00
virtual String get_video_adapter_name() const = 0;
virtual String get_video_adapter_vendor() const = 0;
static RendererStorage *base_singleton;
void set_default_clear_color(const Color &p_color) {
default_clear_color = p_color;
}
Color get_default_clear_color() const {
return default_clear_color;
}
#define TIMESTAMP_BEGIN() \
{ \
if (RSG::storage->capturing_timestamps) \
RSG::storage->capture_timestamps_begin(); \
}
#define RENDER_TIMESTAMP(m_text) \
{ \
if (RSG::storage->capturing_timestamps) \
RSG::storage->capture_timestamp(m_text); \
}
bool capturing_timestamps = false;
virtual void capture_timestamps_begin() = 0;
virtual void capture_timestamp(const String &p_name) = 0;
virtual uint32_t get_captured_timestamps_count() const = 0;
virtual uint64_t get_captured_timestamps_frame() const = 0;
virtual uint64_t get_captured_timestamp_gpu_time(uint32_t p_index) const = 0;
virtual uint64_t get_captured_timestamp_cpu_time(uint32_t p_index) const = 0;
virtual String get_captured_timestamp_name(uint32_t p_index) const = 0;
RendererStorage();
virtual ~RendererStorage() {}
};
#endif // RENDERINGSERVERSTORAGE_H
Reference in New Issue Copy Permalink