/*************************************************************************/
/*  rasterizer_canvas_gles3.cpp                                          */
/*************************************************************************/
/*                       This file is part of:                           */
/*                           GODOT ENGINE                                */
/*                      https://godotengine.org                          */
/*************************************************************************/
/* Copyright (c) 2007-2022 Juan Linietsky, Ariel Manzur.                 */
/* Copyright (c) 2014-2022 Godot Engine contributors (cf. AUTHORS.md).   */
/*                                                                       */
/* 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.                */
/*************************************************************************/

#include "rasterizer_canvas_gles3.h"

#ifdef GLES3_ENABLED

#include "core/os/os.h"
#include "rasterizer_scene_gles3.h"

#include "core/config/project_settings.h"
#include "servers/rendering/rendering_server_default.h"
#include "storage/config.h"
#include "storage/material_storage.h"
#include "storage/mesh_storage.h"
#include "storage/texture_storage.h"

#ifndef GLES_OVER_GL
#define glClearDepth glClearDepthf
#endif

//static const GLenum gl_primitive[] = {
//	GL_POINTS,
//	GL_LINES,
//	GL_LINE_STRIP,
//	GL_LINE_LOOP,
//	GL_TRIANGLES,
//	GL_TRIANGLE_STRIP,
//	GL_TRIANGLE_FAN
//};

void RasterizerCanvasGLES3::_update_transform_2d_to_mat4(const Transform2D &p_transform, float *p_mat4) {
	p_mat4[0] = p_transform.columns[0][0];
	p_mat4[1] = p_transform.columns[0][1];
	p_mat4[2] = 0;
	p_mat4[3] = 0;
	p_mat4[4] = p_transform.columns[1][0];
	p_mat4[5] = p_transform.columns[1][1];
	p_mat4[6] = 0;
	p_mat4[7] = 0;
	p_mat4[8] = 0;
	p_mat4[9] = 0;
	p_mat4[10] = 1;
	p_mat4[11] = 0;
	p_mat4[12] = p_transform.columns[2][0];
	p_mat4[13] = p_transform.columns[2][1];
	p_mat4[14] = 0;
	p_mat4[15] = 1;
}

void RasterizerCanvasGLES3::_update_transform_2d_to_mat2x4(const Transform2D &p_transform, float *p_mat2x4) {
	p_mat2x4[0] = p_transform.columns[0][0];
	p_mat2x4[1] = p_transform.columns[1][0];
	p_mat2x4[2] = 0;
	p_mat2x4[3] = p_transform.columns[2][0];

	p_mat2x4[4] = p_transform.columns[0][1];
	p_mat2x4[5] = p_transform.columns[1][1];
	p_mat2x4[6] = 0;
	p_mat2x4[7] = p_transform.columns[2][1];
}

void RasterizerCanvasGLES3::_update_transform_2d_to_mat2x3(const Transform2D &p_transform, float *p_mat2x3) {
	p_mat2x3[0] = p_transform.columns[0][0];
	p_mat2x3[1] = p_transform.columns[0][1];
	p_mat2x3[2] = p_transform.columns[1][0];
	p_mat2x3[3] = p_transform.columns[1][1];
	p_mat2x3[4] = p_transform.columns[2][0];
	p_mat2x3[5] = p_transform.columns[2][1];
}

void RasterizerCanvasGLES3::_update_transform_to_mat4(const Transform3D &p_transform, float *p_mat4) {
	p_mat4[0] = p_transform.basis.rows[0][0];
	p_mat4[1] = p_transform.basis.rows[1][0];
	p_mat4[2] = p_transform.basis.rows[2][0];
	p_mat4[3] = 0;
	p_mat4[4] = p_transform.basis.rows[0][1];
	p_mat4[5] = p_transform.basis.rows[1][1];
	p_mat4[6] = p_transform.basis.rows[2][1];
	p_mat4[7] = 0;
	p_mat4[8] = p_transform.basis.rows[0][2];
	p_mat4[9] = p_transform.basis.rows[1][2];
	p_mat4[10] = p_transform.basis.rows[2][2];
	p_mat4[11] = 0;
	p_mat4[12] = p_transform.origin.x;
	p_mat4[13] = p_transform.origin.y;
	p_mat4[14] = p_transform.origin.z;
	p_mat4[15] = 1;
}

void RasterizerCanvasGLES3::canvas_render_items(RID p_to_render_target, Item *p_item_list, const Color &p_modulate, Light *p_light_list, Light *p_directional_light_list, const Transform2D &p_canvas_transform, RS::CanvasItemTextureFilter p_default_filter, RS::CanvasItemTextureRepeat p_default_repeat, bool p_snap_2d_vertices_to_pixel, bool &r_sdf_used) {
	GLES3::TextureStorage *texture_storage = GLES3::TextureStorage::get_singleton();
	GLES3::MaterialStorage *material_storage = GLES3::MaterialStorage::get_singleton();

	Transform2D canvas_transform_inverse = p_canvas_transform.affine_inverse();

	// Clear out any state that may have been left from the 3D pass.
	reset_canvas();

	if (state.canvas_instance_data_buffers[state.current_buffer].fence != GLsync()) {
		GLint syncStatus;
		glGetSynciv(state.canvas_instance_data_buffers[state.current_buffer].fence, GL_SYNC_STATUS, sizeof(GLint), nullptr, &syncStatus);
		if (syncStatus == GL_UNSIGNALED) {
			// If older than 2 frames, wait for sync OpenGL can have up to 3 frames in flight, any more and we need to sync anyway.
			if (state.canvas_instance_data_buffers[state.current_buffer].last_frame_used < RSG::rasterizer->get_frame_number() - 2) {
#ifndef WEB_ENABLED
				// On web, we do nothing as the glSubBufferData will force a sync anyway and WebGL does not like waiting.
				glClientWaitSync(state.canvas_instance_data_buffers[state.current_buffer].fence, 0, 100000000); // wait for up to 100ms
#endif
			} else {
				// Used in last frame or frame before that. OpenGL can get up to two frames behind, so these buffers may still be in use
				// Allocate a new buffer and use that.
				_allocate_instance_data_buffer();
			}
		} else {
			// Already finished all rendering commands, we can use it.
			state.canvas_instance_data_buffers[state.current_buffer].last_frame_used = RSG::rasterizer->get_frame_number();
			glDeleteSync(state.canvas_instance_data_buffers[state.current_buffer].fence);
			state.canvas_instance_data_buffers[state.current_buffer].fence = GLsync();
		}
	}

	//setup directional lights if exist

	uint32_t light_count = 0;
	uint32_t directional_light_count = 0;
	{
		Light *l = p_directional_light_list;
		uint32_t index = 0;

		while (l) {
			if (index == data.max_lights_per_render) {
				l->render_index_cache = -1;
				l = l->next_ptr;
				continue;
			}

			CanvasLight *clight = canvas_light_owner.get_or_null(l->light_internal);
			if (!clight) { //unused or invalid texture
				l->render_index_cache = -1;
				l = l->next_ptr;
				ERR_CONTINUE(!clight);
			}

			Vector2 canvas_light_dir = l->xform_cache.columns[1].normalized();

			state.light_uniforms[index].position[0] = -canvas_light_dir.x;
			state.light_uniforms[index].position[1] = -canvas_light_dir.y;

			//_update_transform_2d_to_mat2x4(clight->shadow.directional_xform, state.light_uniforms[index].shadow_matrix);

			state.light_uniforms[index].height = l->height; //0..1 here

			for (int i = 0; i < 4; i++) {
				state.light_uniforms[index].shadow_color[i] = uint8_t(CLAMP(int32_t(l->shadow_color[i] * 255.0), 0, 255));
				state.light_uniforms[index].color[i] = l->color[i];
			}

			state.light_uniforms[index].color[3] = l->energy; //use alpha for energy, so base color can go separate

			/*
			if (state.shadow_fb.is_valid()) {
				state.light_uniforms[index].shadow_pixel_size = (1.0 / state.shadow_texture_size) * (1.0 + l->shadow_smooth);
				state.light_uniforms[index].shadow_z_far_inv = 1.0 / clight->shadow.z_far;
				state.light_uniforms[index].shadow_y_ofs = clight->shadow.y_offset;
			} else {
				state.light_uniforms[index].shadow_pixel_size = 1.0;
				state.light_uniforms[index].shadow_z_far_inv = 1.0;
				state.light_uniforms[index].shadow_y_ofs = 0;
			}
			*/

			state.light_uniforms[index].flags = l->blend_mode << LIGHT_FLAGS_BLEND_SHIFT;
			state.light_uniforms[index].flags |= l->shadow_filter << LIGHT_FLAGS_FILTER_SHIFT;
			/*
			if (clight->shadow.enabled) {
				state.light_uniforms[index].flags |= LIGHT_FLAGS_HAS_SHADOW;
			}
			*/

			l->render_index_cache = index;

			index++;
			l = l->next_ptr;
		}

		light_count = index;
		directional_light_count = light_count;
		state.using_directional_lights = directional_light_count > 0;
	}

	//setup lights if exist

	{
		Light *l = p_light_list;
		uint32_t index = light_count;

		while (l) {
			if (index == data.max_lights_per_render) {
				l->render_index_cache = -1;
				l = l->next_ptr;
				continue;
			}

			CanvasLight *clight = canvas_light_owner.get_or_null(l->light_internal);
			if (!clight) { //unused or invalid texture
				l->render_index_cache = -1;
				l = l->next_ptr;
				ERR_CONTINUE(!clight);
			}
			Transform2D to_light_xform = (p_canvas_transform * l->light_shader_xform).affine_inverse();

			Vector2 canvas_light_pos = p_canvas_transform.xform(l->xform.get_origin()); //convert light position to canvas coordinates, as all computation is done in canvas coords to avoid precision loss
			state.light_uniforms[index].position[0] = canvas_light_pos.x;
			state.light_uniforms[index].position[1] = canvas_light_pos.y;

			_update_transform_2d_to_mat2x4(to_light_xform, state.light_uniforms[index].matrix);
			_update_transform_2d_to_mat2x4(l->xform_cache.affine_inverse(), state.light_uniforms[index].shadow_matrix);

			state.light_uniforms[index].height = l->height * (p_canvas_transform.columns[0].length() + p_canvas_transform.columns[1].length()) * 0.5; //approximate height conversion to the canvas size, since all calculations are done in canvas coords to avoid precision loss
			for (int i = 0; i < 4; i++) {
				state.light_uniforms[index].shadow_color[i] = uint8_t(CLAMP(int32_t(l->shadow_color[i] * 255.0), 0, 255));
				state.light_uniforms[index].color[i] = l->color[i];
			}

			state.light_uniforms[index].color[3] = l->energy; //use alpha for energy, so base color can go separate

			/*
				if (state.shadow_fb.is_valid()) {
					state.light_uniforms[index].shadow_pixel_size = (1.0 / state.shadow_texture_size) * (1.0 + l->shadow_smooth);
					state.light_uniforms[index].shadow_z_far_inv = 1.0 / clight->shadow.z_far;
					state.light_uniforms[index].shadow_y_ofs = clight->shadow.y_offset;
				} else {
					state.light_uniforms[index].shadow_pixel_size = 1.0;
					state.light_uniforms[index].shadow_z_far_inv = 1.0;
					state.light_uniforms[index].shadow_y_ofs = 0;
				}
			*/
			state.light_uniforms[index].flags = l->blend_mode << LIGHT_FLAGS_BLEND_SHIFT;
			state.light_uniforms[index].flags |= l->shadow_filter << LIGHT_FLAGS_FILTER_SHIFT;
			/*
			if (clight->shadow.enabled) {
				state.light_uniforms[index].flags |= LIGHT_FLAGS_HAS_SHADOW;
			}
			*/

			if (clight->texture.is_valid()) {
				Rect2 atlas_rect = GLES3::TextureStorage::get_singleton()->texture_atlas_get_texture_rect(clight->texture);
				state.light_uniforms[index].atlas_rect[0] = atlas_rect.position.x;
				state.light_uniforms[index].atlas_rect[1] = atlas_rect.position.y;
				state.light_uniforms[index].atlas_rect[2] = atlas_rect.size.width;
				state.light_uniforms[index].atlas_rect[3] = atlas_rect.size.height;

			} else {
				state.light_uniforms[index].atlas_rect[0] = 0;
				state.light_uniforms[index].atlas_rect[1] = 0;
				state.light_uniforms[index].atlas_rect[2] = 0;
				state.light_uniforms[index].atlas_rect[3] = 0;
			}

			l->render_index_cache = index;

			index++;
			l = l->next_ptr;
		}

		light_count = index;
	}

	if (light_count > 0) {
		glBindBufferBase(GL_UNIFORM_BUFFER, LIGHT_UNIFORM_LOCATION, state.canvas_instance_data_buffers[state.current_buffer].light_ubo);

#ifdef WEB_ENABLED
		glBufferSubData(GL_UNIFORM_BUFFER, 0, sizeof(LightUniform) * light_count, state.light_uniforms);
#else
		// On Desktop and mobile we map the memory without synchronizing for maximum speed.
		void *ubo = glMapBufferRange(GL_UNIFORM_BUFFER, 0, sizeof(LightUniform) * light_count, GL_MAP_WRITE_BIT | GL_MAP_UNSYNCHRONIZED_BIT);
		memcpy(ubo, state.light_uniforms, sizeof(LightUniform) * light_count);
		glUnmapBuffer(GL_UNIFORM_BUFFER);
#endif

		GLuint texture_atlas = texture_storage->texture_atlas_get_texture();
		if (texture_atlas == 0) {
			GLES3::Texture *tex = texture_storage->get_texture(texture_storage->texture_gl_get_default(GLES3::DEFAULT_GL_TEXTURE_WHITE));
			texture_atlas = tex->tex_id;
		}
		glActiveTexture(GL_TEXTURE0 + GLES3::Config::get_singleton()->max_texture_image_units - 2);
		glBindTexture(GL_TEXTURE_2D, texture_atlas);
	}

	{
		//update canvas state uniform buffer
		StateBuffer state_buffer;

		Size2i ssize = texture_storage->render_target_get_size(p_to_render_target);

		Transform3D screen_transform;
		screen_transform.translate_local(-(ssize.width / 2.0f), -(ssize.height / 2.0f), 0.0f);
		screen_transform.scale(Vector3(2.0f / ssize.width, 2.0f / ssize.height, 1.0f));
		_update_transform_to_mat4(screen_transform, state_buffer.screen_transform);
		_update_transform_2d_to_mat4(p_canvas_transform, state_buffer.canvas_transform);

		Transform2D normal_transform = p_canvas_transform;
		normal_transform.columns[0].normalize();
		normal_transform.columns[1].normalize();
		normal_transform.columns[2] = Vector2();
		_update_transform_2d_to_mat4(normal_transform, state_buffer.canvas_normal_transform);

		state_buffer.canvas_modulate[0] = p_modulate.r;
		state_buffer.canvas_modulate[1] = p_modulate.g;
		state_buffer.canvas_modulate[2] = p_modulate.b;
		state_buffer.canvas_modulate[3] = p_modulate.a;

		Size2 render_target_size = texture_storage->render_target_get_size(p_to_render_target);
		state_buffer.screen_pixel_size[0] = 1.0 / render_target_size.x;
		state_buffer.screen_pixel_size[1] = 1.0 / render_target_size.y;

		glViewport(0, 0, render_target_size.x, render_target_size.y);

		state_buffer.time = state.time;
		state_buffer.use_pixel_snap = p_snap_2d_vertices_to_pixel;

		state_buffer.directional_light_count = directional_light_count;

		Vector2 canvas_scale = p_canvas_transform.get_scale();

		state_buffer.sdf_to_screen[0] = render_target_size.width / canvas_scale.x;
		state_buffer.sdf_to_screen[1] = render_target_size.height / canvas_scale.y;

		state_buffer.screen_to_sdf[0] = 1.0 / state_buffer.sdf_to_screen[0];
		state_buffer.screen_to_sdf[1] = 1.0 / state_buffer.sdf_to_screen[1];

		Rect2 sdf_rect = texture_storage->render_target_get_sdf_rect(p_to_render_target);
		Rect2 sdf_tex_rect(sdf_rect.position / canvas_scale, sdf_rect.size / canvas_scale);

		state_buffer.sdf_to_tex[0] = 1.0 / sdf_tex_rect.size.width;
		state_buffer.sdf_to_tex[1] = 1.0 / sdf_tex_rect.size.height;
		state_buffer.sdf_to_tex[2] = -sdf_tex_rect.position.x / sdf_tex_rect.size.width;
		state_buffer.sdf_to_tex[3] = -sdf_tex_rect.position.y / sdf_tex_rect.size.height;

		state_buffer.tex_to_sdf = 1.0 / ((canvas_scale.x + canvas_scale.y) * 0.5);
		glBindBufferBase(GL_UNIFORM_BUFFER, BASE_UNIFORM_LOCATION, state.canvas_instance_data_buffers[state.current_buffer].state_ubo);
		glBufferData(GL_UNIFORM_BUFFER, sizeof(StateBuffer), &state_buffer, GL_STREAM_DRAW);

		GLuint global_buffer = material_storage->global_shader_parameters_get_uniform_buffer();

		glBindBufferBase(GL_UNIFORM_BUFFER, GLOBAL_UNIFORM_LOCATION, global_buffer);
		glBindBuffer(GL_UNIFORM_BUFFER, 0);
	}

	{
		state.default_filter = p_default_filter;
		state.default_repeat = p_default_repeat;
	}

	r_sdf_used = false;
	int item_count = 0;
	bool backbuffer_cleared = false;
	bool time_used = false;
	bool material_screen_texture_cached = false;
	bool material_screen_texture_mipmaps_cached = false;
	Rect2 back_buffer_rect;
	bool backbuffer_copy = false;
	bool backbuffer_gen_mipmaps = false;

	Item *ci = p_item_list;
	Item *canvas_group_owner = nullptr;

	uint32_t starting_index = 0;

	while (ci) {
		if (ci->copy_back_buffer && canvas_group_owner == nullptr) {
			backbuffer_copy = true;

			if (ci->copy_back_buffer->full) {
				back_buffer_rect = Rect2();
			} else {
				back_buffer_rect = ci->copy_back_buffer->rect;
			}
		}

		// Check material for something that may change flow of rendering, but do not bind for now.
		RID material = ci->material_owner == nullptr ? ci->material : ci->material_owner->material;
		if (material.is_valid()) {
			GLES3::CanvasMaterialData *md = static_cast<GLES3::CanvasMaterialData *>(material_storage->material_get_data(material, RS::SHADER_CANVAS_ITEM));
			if (md && md->shader_data->valid) {
				if (md->shader_data->uses_screen_texture && canvas_group_owner == nullptr) {
					if (!material_screen_texture_cached) {
						backbuffer_copy = true;
						back_buffer_rect = Rect2();
						backbuffer_gen_mipmaps = md->shader_data->uses_screen_texture_mipmaps;
					} else if (!material_screen_texture_mipmaps_cached) {
						backbuffer_gen_mipmaps = md->shader_data->uses_screen_texture_mipmaps;
					}
				}

				if (md->shader_data->uses_sdf) {
					r_sdf_used = true;
				}
				if (md->shader_data->uses_time) {
					time_used = true;
				}
			}
		}

		if (ci->canvas_group_owner != nullptr) {
			if (canvas_group_owner == nullptr) {
				// Canvas group begins here, render until before this item
				_render_items(p_to_render_target, item_count, canvas_transform_inverse, p_light_list, starting_index, false);
				item_count = 0;

				Rect2i group_rect = ci->canvas_group_owner->global_rect_cache;

				if (ci->canvas_group_owner->canvas_group->mode == RS::CANVAS_GROUP_MODE_OPAQUE) {
					texture_storage->render_target_copy_to_back_buffer(p_to_render_target, group_rect, false);
				} else if (!backbuffer_cleared) {
					texture_storage->render_target_clear_back_buffer(p_to_render_target, Rect2i(), Color(0, 0, 0, 0));
					backbuffer_cleared = true;
				}

				backbuffer_copy = false;
				canvas_group_owner = ci->canvas_group_owner; //continue until owner found
			}

			ci->canvas_group_owner = nullptr; //must be cleared
		}

		if (!backbuffer_cleared && canvas_group_owner == nullptr && ci->canvas_group != nullptr && !backbuffer_copy) {
			texture_storage->render_target_clear_back_buffer(p_to_render_target, Rect2i(), Color(0, 0, 0, 0));
			backbuffer_cleared = true;
		}

		if (ci == canvas_group_owner) {
			_render_items(p_to_render_target, item_count, canvas_transform_inverse, p_light_list, starting_index, true);
			item_count = 0;

			if (ci->canvas_group->blur_mipmaps) {
				texture_storage->render_target_gen_back_buffer_mipmaps(p_to_render_target, ci->global_rect_cache);
			}

			canvas_group_owner = nullptr;
			// Backbuffer is dirty now and needs to be re-cleared if another CanvasGroup needs it.
			backbuffer_cleared = false;
		}

		if (backbuffer_copy) {
			//render anything pending, including clearing if no items

			_render_items(p_to_render_target, item_count, canvas_transform_inverse, p_light_list, starting_index, false);
			item_count = 0;

			texture_storage->render_target_copy_to_back_buffer(p_to_render_target, back_buffer_rect, backbuffer_gen_mipmaps);

			backbuffer_copy = false;
			backbuffer_gen_mipmaps = false;
			material_screen_texture_cached = true; // After a backbuffer copy, screen texture makes no further copies.
			material_screen_texture_mipmaps_cached = backbuffer_gen_mipmaps;
		}

		if (backbuffer_gen_mipmaps) {
			texture_storage->render_target_gen_back_buffer_mipmaps(p_to_render_target, back_buffer_rect);

			backbuffer_gen_mipmaps = false;
			material_screen_texture_mipmaps_cached = true;
		}

		// just add all items for now
		items[item_count++] = ci;

		if (!ci->next || item_count == MAX_RENDER_ITEMS - 1) {
			_render_items(p_to_render_target, item_count, canvas_transform_inverse, p_light_list, starting_index, false);
			//then reset
			item_count = 0;
		}

		ci = ci->next;
	}

	if (time_used) {
		RenderingServerDefault::redraw_request();
	}

	state.canvas_instance_data_buffers[state.current_buffer].fence = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);

	// Clear out state used in 2D pass
	reset_canvas();
	state.current_buffer = (state.current_buffer + 1) % state.canvas_instance_data_buffers.size();
}

void RasterizerCanvasGLES3::_render_items(RID p_to_render_target, int p_item_count, const Transform2D &p_canvas_transform_inverse, Light *p_lights, uint32_t &r_last_index, bool p_to_backbuffer) {
	GLES3::MaterialStorage *material_storage = GLES3::MaterialStorage::get_singleton();

	canvas_begin(p_to_render_target, p_to_backbuffer);

	if (p_item_count <= 0) {
		// Nothing to draw, just call canvas_begin() to clear the render target and return.
		return;
	}

	uint32_t index = 0;
	Item *current_clip = nullptr;

	// Record Batches.
	// First item always forms its own batch.
	bool batch_broken = false;
	_new_batch(batch_broken, index);

	// Override the start position and index as we want to start from where we finished off last time.
	state.canvas_instance_batches[state.current_batch_index].start = r_last_index * sizeof(InstanceData);
	index = 0;
	_align_instance_data_buffer(index);

	for (int i = 0; i < p_item_count; i++) {
		Item *ci = items[i];

		if (ci->final_clip_owner != state.canvas_instance_batches[state.current_batch_index].clip) {
			_new_batch(batch_broken, index);
			state.canvas_instance_batches[state.current_batch_index].clip = ci->final_clip_owner;
			current_clip = ci->final_clip_owner;
		}

		RID material = ci->material_owner == nullptr ? ci->material : ci->material_owner->material;

		if (material.is_null() && ci->canvas_group != nullptr) {
			material = default_canvas_group_material;
		}

		GLES3::CanvasShaderData *shader_data_cache = nullptr;
		if (material != state.canvas_instance_batches[state.current_batch_index].material) {
			_new_batch(batch_broken, index);

			GLES3::CanvasMaterialData *material_data = nullptr;
			if (material.is_valid()) {
				material_data = static_cast<GLES3::CanvasMaterialData *>(material_storage->material_get_data(material, RS::SHADER_CANVAS_ITEM));
			}
			shader_data_cache = nullptr;
			if (material_data) {
				if (material_data->shader_data->version.is_valid() && material_data->shader_data->valid) {
					shader_data_cache = material_data->shader_data;
				}
			}

			state.canvas_instance_batches[state.current_batch_index].material = material;
			state.canvas_instance_batches[state.current_batch_index].material_data = material_data;
		}

		GLES3::CanvasShaderData::BlendMode blend_mode = shader_data_cache ? shader_data_cache->blend_mode : GLES3::CanvasShaderData::BLEND_MODE_MIX;

		_record_item_commands(ci, p_canvas_transform_inverse, current_clip, blend_mode, p_lights, index, batch_broken);
	}

	// Copy over all data needed for rendering.
	glBindBuffer(GL_UNIFORM_BUFFER, state.canvas_instance_data_buffers[state.current_buffer].ubo);
#ifdef WEB_ENABLED
	glBufferSubData(GL_UNIFORM_BUFFER, r_last_index * sizeof(InstanceData), sizeof(InstanceData) * index, state.instance_data_array);
#else
	// On Desktop and mobile we map the memory without synchronizing for maximum speed.
	void *ubo = glMapBufferRange(GL_UNIFORM_BUFFER, r_last_index * sizeof(InstanceData), index * sizeof(InstanceData), GL_MAP_WRITE_BIT | GL_MAP_UNSYNCHRONIZED_BIT);
	memcpy(ubo, state.instance_data_array, index * sizeof(InstanceData));
	glUnmapBuffer(GL_UNIFORM_BUFFER);
#endif

	glDisable(GL_SCISSOR_TEST);
	current_clip = nullptr;

	GLES3::CanvasShaderData::BlendMode last_blend_mode = GLES3::CanvasShaderData::BLEND_MODE_MIX;

	state.current_tex = RID();

	for (uint32_t i = 0; i <= state.current_batch_index; i++) {
		//setup clip
		if (current_clip != state.canvas_instance_batches[i].clip) {
			current_clip = state.canvas_instance_batches[i].clip;
			if (current_clip) {
				glEnable(GL_SCISSOR_TEST);
				glScissor(current_clip->final_clip_rect.position.x, current_clip->final_clip_rect.position.y, current_clip->final_clip_rect.size.x, current_clip->final_clip_rect.size.y);
			} else {
				glDisable(GL_SCISSOR_TEST);
			}
		}

		GLES3::CanvasMaterialData *material_data = state.canvas_instance_batches[i].material_data;
		CanvasShaderGLES3::ShaderVariant variant = state.canvas_instance_batches[i].shader_variant;
		uint64_t specialization = 0;
		specialization |= uint64_t(state.canvas_instance_batches[i].lights_disabled);
		_bind_material(material_data, variant, specialization);

		GLES3::CanvasShaderData::BlendMode blend_mode = state.canvas_instance_batches[i].blend_mode;

		if (last_blend_mode != blend_mode) {
			if (last_blend_mode == GLES3::CanvasShaderData::BLEND_MODE_DISABLED) {
				// re-enable it
				glEnable(GL_BLEND);
			} else if (blend_mode == GLES3::CanvasShaderData::BLEND_MODE_DISABLED) {
				// disable it
				glDisable(GL_BLEND);
			}

			switch (blend_mode) {
				case GLES3::CanvasShaderData::BLEND_MODE_DISABLED: {
					// Nothing to do here.
				} break;
				case GLES3::CanvasShaderData::BLEND_MODE_LCD: {
					glBlendEquation(GL_FUNC_ADD);
					if (state.transparent_render_target) {
						glBlendFuncSeparate(GL_CONSTANT_COLOR, GL_ONE_MINUS_SRC_COLOR, GL_ONE, GL_ONE_MINUS_SRC_ALPHA);
					} else {
						glBlendFuncSeparate(GL_CONSTANT_COLOR, GL_ONE_MINUS_SRC_COLOR, GL_ZERO, GL_ONE);
					}
					Color blend_color = state.canvas_instance_batches[state.current_batch_index].blend_color;
					glBlendColor(blend_color.r, blend_color.g, blend_color.b, blend_color.a);

				} break;
				case GLES3::CanvasShaderData::BLEND_MODE_MIX: {
					glBlendEquation(GL_FUNC_ADD);
					if (state.transparent_render_target) {
						glBlendFuncSeparate(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA, GL_ONE, GL_ONE_MINUS_SRC_ALPHA);
					} else {
						glBlendFuncSeparate(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA, GL_ZERO, GL_ONE);
					}

				} break;
				case GLES3::CanvasShaderData::BLEND_MODE_ADD: {
					glBlendEquation(GL_FUNC_ADD);
					if (state.transparent_render_target) {
						glBlendFuncSeparate(GL_SRC_ALPHA, GL_ONE, GL_SRC_ALPHA, GL_ONE);
					} else {
						glBlendFuncSeparate(GL_SRC_ALPHA, GL_ONE, GL_ZERO, GL_ONE);
					}

				} break;
				case GLES3::CanvasShaderData::BLEND_MODE_SUB: {
					glBlendEquation(GL_FUNC_REVERSE_SUBTRACT);
					if (state.transparent_render_target) {
						glBlendFuncSeparate(GL_SRC_ALPHA, GL_ONE, GL_SRC_ALPHA, GL_ONE);
					} else {
						glBlendFuncSeparate(GL_SRC_ALPHA, GL_ONE, GL_ZERO, GL_ONE);
					}
				} break;
				case GLES3::CanvasShaderData::BLEND_MODE_MUL: {
					glBlendEquation(GL_FUNC_ADD);
					if (state.transparent_render_target) {
						glBlendFuncSeparate(GL_DST_COLOR, GL_ZERO, GL_DST_ALPHA, GL_ZERO);
					} else {
						glBlendFuncSeparate(GL_DST_COLOR, GL_ZERO, GL_ZERO, GL_ONE);
					}

				} break;
				case GLES3::CanvasShaderData::BLEND_MODE_PMALPHA: {
					glBlendEquation(GL_FUNC_ADD);
					if (state.transparent_render_target) {
						glBlendFuncSeparate(GL_ONE, GL_ONE_MINUS_SRC_ALPHA, GL_ONE, GL_ONE_MINUS_SRC_ALPHA);
					} else {
						glBlendFuncSeparate(GL_ONE, GL_ONE_MINUS_SRC_ALPHA, GL_ZERO, GL_ONE);
					}

				} break;
			}
			last_blend_mode = blend_mode;
		}

		_render_batch(p_lights, i);
	}

	state.current_batch_index = 0;
	state.canvas_instance_batches.clear();
	r_last_index += index;
}

void RasterizerCanvasGLES3::_record_item_commands(const Item *p_item, const Transform2D &p_canvas_transform_inverse, Item *&current_clip, GLES3::CanvasShaderData::BlendMode p_blend_mode, Light *p_lights, uint32_t &r_index, bool &r_batch_broken) {
	RenderingServer::CanvasItemTextureFilter texture_filter = p_item->texture_filter == RS::CANVAS_ITEM_TEXTURE_FILTER_DEFAULT ? state.default_filter : p_item->texture_filter;

	if (texture_filter != state.canvas_instance_batches[state.current_batch_index].filter) {
		_new_batch(r_batch_broken, r_index);

		state.canvas_instance_batches[state.current_batch_index].filter = texture_filter;
	}

	RenderingServer::CanvasItemTextureRepeat texture_repeat = p_item->texture_repeat == RS::CANVAS_ITEM_TEXTURE_REPEAT_DEFAULT ? state.default_repeat : p_item->texture_repeat;

	if (texture_repeat != state.canvas_instance_batches[state.current_batch_index].repeat) {
		_new_batch(r_batch_broken, r_index);

		state.canvas_instance_batches[state.current_batch_index].repeat = texture_repeat;
	}

	Transform2D base_transform = p_canvas_transform_inverse * p_item->final_transform;
	Transform2D draw_transform; // Used by transform command

	Color base_color = p_item->final_modulate;
	uint32_t base_flags = 0;
	Size2 texpixel_size;

	bool reclip = false;

	bool skipping = false;

	// TODO: consider making lights a per-batch property and then baking light operations in the shader for better performance.
	uint32_t lights[4] = { 0, 0, 0, 0 };

	uint16_t light_count = 0;

	{
		Light *light = p_lights;

		while (light) {
			if (light->render_index_cache >= 0 && p_item->light_mask & light->item_mask && p_item->z_final >= light->z_min && p_item->z_final <= light->z_max && p_item->global_rect_cache.intersects_transformed(light->xform_cache, light->rect_cache)) {
				uint32_t light_index = light->render_index_cache;
				lights[light_count >> 2] |= light_index << ((light_count & 3) * 8);

				light_count++;

				if (light_count == data.max_lights_per_item) {
					break;
				}
			}
			light = light->next_ptr;
		}

		base_flags |= light_count << FLAGS_LIGHT_COUNT_SHIFT;
	}

	bool lights_disabled = light_count == 0 && !state.using_directional_lights;

	if (lights_disabled != state.canvas_instance_batches[state.current_batch_index].lights_disabled) {
		_new_batch(r_batch_broken, r_index);
		state.canvas_instance_batches[state.current_batch_index].lights_disabled = lights_disabled;
	}

	const Item::Command *c = p_item->commands;
	while (c) {
		if (skipping && c->type != Item::Command::TYPE_ANIMATION_SLICE) {
			c = c->next;
			continue;
		}

		if (c->type != Item::Command::TYPE_MESH) {
			// For Meshes, this gets updated below.
			_update_transform_2d_to_mat2x3(base_transform * draw_transform, state.instance_data_array[r_index].world);
		}

		// Zero out most fields.
		for (int i = 0; i < 4; i++) {
			state.instance_data_array[r_index].modulation[i] = 0.0;
			state.instance_data_array[r_index].ninepatch_margins[i] = 0.0;
			state.instance_data_array[r_index].src_rect[i] = 0.0;
			state.instance_data_array[r_index].dst_rect[i] = 0.0;
			state.instance_data_array[r_index].lights[i] = uint32_t(0);
		}
		state.instance_data_array[r_index].color_texture_pixel_size[0] = 0.0;
		state.instance_data_array[r_index].color_texture_pixel_size[1] = 0.0;

		state.instance_data_array[r_index].pad[0] = 0.0;
		state.instance_data_array[r_index].pad[1] = 0.0;

		state.instance_data_array[r_index].lights[0] = lights[0];
		state.instance_data_array[r_index].lights[1] = lights[1];
		state.instance_data_array[r_index].lights[2] = lights[2];
		state.instance_data_array[r_index].lights[3] = lights[3];

		state.instance_data_array[r_index].flags = base_flags | (state.instance_data_array[r_index == 0 ? 0 : r_index - 1].flags & (FLAGS_DEFAULT_NORMAL_MAP_USED | FLAGS_DEFAULT_SPECULAR_MAP_USED)); //reset on each command for sanity, keep canvastexture binding config

		Color blend_color;
		if (c->type == Item::Command::TYPE_RECT) {
			const Item::CommandRect *rect = static_cast<const Item::CommandRect *>(c);
			if (rect->flags & CANVAS_RECT_LCD) {
				p_blend_mode = GLES3::CanvasShaderData::BLEND_MODE_LCD;
				blend_color = rect->modulate * base_color;
			}
		}

		if (p_blend_mode != state.canvas_instance_batches[state.current_batch_index].blend_mode || blend_color != state.canvas_instance_batches[state.current_batch_index].blend_color) {
			_new_batch(r_batch_broken, r_index);
			state.canvas_instance_batches[state.current_batch_index].blend_mode = p_blend_mode;
			state.canvas_instance_batches[state.current_batch_index].blend_color = blend_color;
		}

		switch (c->type) {
			case Item::Command::TYPE_RECT: {
				const Item::CommandRect *rect = static_cast<const Item::CommandRect *>(c);

				if (rect->flags & CANVAS_RECT_TILE && state.canvas_instance_batches[state.current_batch_index].repeat != RenderingServer::CanvasItemTextureRepeat::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED) {
					_new_batch(r_batch_broken, r_index);
					state.canvas_instance_batches[state.current_batch_index].repeat = RenderingServer::CanvasItemTextureRepeat::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED;
				}

				if (rect->texture != state.canvas_instance_batches[state.current_batch_index].tex || state.canvas_instance_batches[state.current_batch_index].command_type != Item::Command::TYPE_RECT) {
					_new_batch(r_batch_broken, r_index);
					state.canvas_instance_batches[state.current_batch_index].tex = rect->texture;
					state.canvas_instance_batches[state.current_batch_index].command_type = Item::Command::TYPE_RECT;
					state.canvas_instance_batches[state.current_batch_index].command = c;
					state.canvas_instance_batches[state.current_batch_index].shader_variant = CanvasShaderGLES3::MODE_QUAD;
				}

				_prepare_canvas_texture(rect->texture, state.canvas_instance_batches[state.current_batch_index].filter, state.canvas_instance_batches[state.current_batch_index].repeat, r_index, texpixel_size);

				Rect2 src_rect;
				Rect2 dst_rect;

				if (rect->texture != RID()) {
					src_rect = (rect->flags & CANVAS_RECT_REGION) ? Rect2(rect->source.position * texpixel_size, rect->source.size * texpixel_size) : Rect2(0, 0, 1, 1);
					dst_rect = Rect2(rect->rect.position, rect->rect.size);

					if (dst_rect.size.width < 0) {
						dst_rect.position.x += dst_rect.size.width;
						dst_rect.size.width *= -1;
					}
					if (dst_rect.size.height < 0) {
						dst_rect.position.y += dst_rect.size.height;
						dst_rect.size.height *= -1;
					}

					if (rect->flags & CANVAS_RECT_FLIP_H) {
						src_rect.size.x *= -1;
					}

					if (rect->flags & CANVAS_RECT_FLIP_V) {
						src_rect.size.y *= -1;
					}

					if (rect->flags & CANVAS_RECT_TRANSPOSE) {
						dst_rect.size.x *= -1; // Encoding in the dst_rect.z uniform
					}

					if (rect->flags & CANVAS_RECT_CLIP_UV) {
						state.instance_data_array[r_index].flags |= FLAGS_CLIP_RECT_UV;
					}

				} else {
					dst_rect = Rect2(rect->rect.position, rect->rect.size);

					if (dst_rect.size.width < 0) {
						dst_rect.position.x += dst_rect.size.width;
						dst_rect.size.width *= -1;
					}
					if (dst_rect.size.height < 0) {
						dst_rect.position.y += dst_rect.size.height;
						dst_rect.size.height *= -1;
					}

					src_rect = Rect2(0, 0, 1, 1);
				}

				if (rect->flags & CANVAS_RECT_MSDF) {
					state.instance_data_array[r_index].flags |= FLAGS_USE_MSDF;
					state.instance_data_array[r_index].msdf[0] = rect->px_range; // Pixel range.
					state.instance_data_array[r_index].msdf[1] = rect->outline; // Outline size.
					state.instance_data_array[r_index].msdf[2] = 0.f; // Reserved.
					state.instance_data_array[r_index].msdf[3] = 0.f; // Reserved.
				} else if (rect->flags & CANVAS_RECT_LCD) {
					state.instance_data_array[r_index].flags |= FLAGS_USE_LCD;
				}

				state.instance_data_array[r_index].modulation[0] = rect->modulate.r * base_color.r;
				state.instance_data_array[r_index].modulation[1] = rect->modulate.g * base_color.g;
				state.instance_data_array[r_index].modulation[2] = rect->modulate.b * base_color.b;
				state.instance_data_array[r_index].modulation[3] = rect->modulate.a * base_color.a;

				state.instance_data_array[r_index].src_rect[0] = src_rect.position.x;
				state.instance_data_array[r_index].src_rect[1] = src_rect.position.y;
				state.instance_data_array[r_index].src_rect[2] = src_rect.size.width;
				state.instance_data_array[r_index].src_rect[3] = src_rect.size.height;

				state.instance_data_array[r_index].dst_rect[0] = dst_rect.position.x;
				state.instance_data_array[r_index].dst_rect[1] = dst_rect.position.y;
				state.instance_data_array[r_index].dst_rect[2] = dst_rect.size.width;
				state.instance_data_array[r_index].dst_rect[3] = dst_rect.size.height;

				_add_to_batch(r_index, r_batch_broken);
			} break;

			case Item::Command::TYPE_NINEPATCH: {
				const Item::CommandNinePatch *np = static_cast<const Item::CommandNinePatch *>(c);

				if (np->texture != state.canvas_instance_batches[state.current_batch_index].tex || state.canvas_instance_batches[state.current_batch_index].command_type != Item::Command::TYPE_NINEPATCH) {
					_new_batch(r_batch_broken, r_index);
					state.canvas_instance_batches[state.current_batch_index].tex = np->texture;
					state.canvas_instance_batches[state.current_batch_index].command_type = Item::Command::TYPE_NINEPATCH;
					state.canvas_instance_batches[state.current_batch_index].command = c;
					state.canvas_instance_batches[state.current_batch_index].shader_variant = CanvasShaderGLES3::MODE_NINEPATCH;
				}

				_prepare_canvas_texture(np->texture, state.canvas_instance_batches[state.current_batch_index].filter, state.canvas_instance_batches[state.current_batch_index].repeat, r_index, texpixel_size);

				Rect2 src_rect;
				Rect2 dst_rect(np->rect.position.x, np->rect.position.y, np->rect.size.x, np->rect.size.y);

				if (np->texture == RID()) {
					texpixel_size = Size2(1, 1);
					src_rect = Rect2(0, 0, 1, 1);

				} else {
					if (np->source != Rect2()) {
						src_rect = Rect2(np->source.position.x * texpixel_size.width, np->source.position.y * texpixel_size.height, np->source.size.x * texpixel_size.width, np->source.size.y * texpixel_size.height);
						state.instance_data_array[r_index].color_texture_pixel_size[0] = 1.0 / np->source.size.width;
						state.instance_data_array[r_index].color_texture_pixel_size[1] = 1.0 / np->source.size.height;

					} else {
						src_rect = Rect2(0, 0, 1, 1);
					}
				}

				state.instance_data_array[r_index].modulation[0] = np->color.r * base_color.r;
				state.instance_data_array[r_index].modulation[1] = np->color.g * base_color.g;
				state.instance_data_array[r_index].modulation[2] = np->color.b * base_color.b;
				state.instance_data_array[r_index].modulation[3] = np->color.a * base_color.a;

				state.instance_data_array[r_index].src_rect[0] = src_rect.position.x;
				state.instance_data_array[r_index].src_rect[1] = src_rect.position.y;
				state.instance_data_array[r_index].src_rect[2] = src_rect.size.width;
				state.instance_data_array[r_index].src_rect[3] = src_rect.size.height;

				state.instance_data_array[r_index].dst_rect[0] = dst_rect.position.x;
				state.instance_data_array[r_index].dst_rect[1] = dst_rect.position.y;
				state.instance_data_array[r_index].dst_rect[2] = dst_rect.size.width;
				state.instance_data_array[r_index].dst_rect[3] = dst_rect.size.height;

				state.instance_data_array[r_index].flags |= int(np->axis_x) << FLAGS_NINEPATCH_H_MODE_SHIFT;
				state.instance_data_array[r_index].flags |= int(np->axis_y) << FLAGS_NINEPATCH_V_MODE_SHIFT;

				if (np->draw_center) {
					state.instance_data_array[r_index].flags |= FLAGS_NINEPACH_DRAW_CENTER;
				}

				state.instance_data_array[r_index].ninepatch_margins[0] = np->margin[SIDE_LEFT];
				state.instance_data_array[r_index].ninepatch_margins[1] = np->margin[SIDE_TOP];
				state.instance_data_array[r_index].ninepatch_margins[2] = np->margin[SIDE_RIGHT];
				state.instance_data_array[r_index].ninepatch_margins[3] = np->margin[SIDE_BOTTOM];

				_add_to_batch(r_index, r_batch_broken);

				// Restore if overridden.
				state.instance_data_array[r_index].color_texture_pixel_size[0] = texpixel_size.x;
				state.instance_data_array[r_index].color_texture_pixel_size[1] = texpixel_size.y;
			} break;

			case Item::Command::TYPE_POLYGON: {
				const Item::CommandPolygon *polygon = static_cast<const Item::CommandPolygon *>(c);

				// Polygon's can't be batched, so always create a new batch
				_new_batch(r_batch_broken, r_index);

				state.canvas_instance_batches[state.current_batch_index].tex = polygon->texture;
				state.canvas_instance_batches[state.current_batch_index].command_type = Item::Command::TYPE_POLYGON;
				state.canvas_instance_batches[state.current_batch_index].command = c;
				state.canvas_instance_batches[state.current_batch_index].shader_variant = CanvasShaderGLES3::MODE_ATTRIBUTES;

				_prepare_canvas_texture(polygon->texture, state.canvas_instance_batches[state.current_batch_index].filter, state.canvas_instance_batches[state.current_batch_index].repeat, r_index, texpixel_size);

				state.instance_data_array[r_index].modulation[0] = base_color.r;
				state.instance_data_array[r_index].modulation[1] = base_color.g;
				state.instance_data_array[r_index].modulation[2] = base_color.b;
				state.instance_data_array[r_index].modulation[3] = base_color.a;

				for (int j = 0; j < 4; j++) {
					state.instance_data_array[r_index].src_rect[j] = 0;
					state.instance_data_array[r_index].dst_rect[j] = 0;
					state.instance_data_array[r_index].ninepatch_margins[j] = 0;
				}

				_add_to_batch(r_index, r_batch_broken);
			} break;

			case Item::Command::TYPE_PRIMITIVE: {
				const Item::CommandPrimitive *primitive = static_cast<const Item::CommandPrimitive *>(c);

				if (primitive->point_count != state.canvas_instance_batches[state.current_batch_index].primitive_points || state.canvas_instance_batches[state.current_batch_index].command_type != Item::Command::TYPE_PRIMITIVE) {
					_new_batch(r_batch_broken, r_index);
					state.canvas_instance_batches[state.current_batch_index].tex = RID();
					state.canvas_instance_batches[state.current_batch_index].primitive_points = primitive->point_count;
					state.canvas_instance_batches[state.current_batch_index].command_type = Item::Command::TYPE_PRIMITIVE;
					state.canvas_instance_batches[state.current_batch_index].command = c;
					state.canvas_instance_batches[state.current_batch_index].shader_variant = CanvasShaderGLES3::MODE_PRIMITIVE;
				}

				for (uint32_t j = 0; j < MIN(3u, primitive->point_count); j++) {
					state.instance_data_array[r_index].points[j * 2 + 0] = primitive->points[j].x;
					state.instance_data_array[r_index].points[j * 2 + 1] = primitive->points[j].y;
					state.instance_data_array[r_index].uvs[j * 2 + 0] = primitive->uvs[j].x;
					state.instance_data_array[r_index].uvs[j * 2 + 1] = primitive->uvs[j].y;
					Color col = primitive->colors[j] * base_color;
					state.instance_data_array[r_index].colors[j * 2 + 0] = (uint32_t(Math::make_half_float(col.g)) << 16) | Math::make_half_float(col.r);
					state.instance_data_array[r_index].colors[j * 2 + 1] = (uint32_t(Math::make_half_float(col.a)) << 16) | Math::make_half_float(col.b);
				}

				_add_to_batch(r_index, r_batch_broken);

				if (primitive->point_count == 4) {
					// Reset base data.
					_update_transform_2d_to_mat2x3(base_transform * draw_transform, state.instance_data_array[r_index].world);
					state.instance_data_array[r_index].color_texture_pixel_size[0] = 0.0;
					state.instance_data_array[r_index].color_texture_pixel_size[1] = 0.0;

					state.instance_data_array[r_index].flags = base_flags | (state.instance_data_array[r_index - 1].flags & (FLAGS_DEFAULT_NORMAL_MAP_USED | FLAGS_DEFAULT_SPECULAR_MAP_USED)); //reset on each command for sanity, keep canvastexture binding config

					for (uint32_t j = 0; j < 3; j++) {
						int offset = j == 0 ? 0 : 1;
						// Second triangle in the quad. Uses vertices 0, 2, 3.
						state.instance_data_array[r_index].points[j * 2 + 0] = primitive->points[j + offset].x;
						state.instance_data_array[r_index].points[j * 2 + 1] = primitive->points[j + offset].y;
						state.instance_data_array[r_index].uvs[j * 2 + 0] = primitive->uvs[j + offset].x;
						state.instance_data_array[r_index].uvs[j * 2 + 1] = primitive->uvs[j + offset].y;
						Color col = primitive->colors[j + offset] * base_color;
						state.instance_data_array[r_index].colors[j * 2 + 0] = (uint32_t(Math::make_half_float(col.g)) << 16) | Math::make_half_float(col.r);
						state.instance_data_array[r_index].colors[j * 2 + 1] = (uint32_t(Math::make_half_float(col.a)) << 16) | Math::make_half_float(col.b);
					}

					_add_to_batch(r_index, r_batch_broken);
				}
			} break;

			case Item::Command::TYPE_MESH:
			case Item::Command::TYPE_MULTIMESH:
			case Item::Command::TYPE_PARTICLES: {
				// Mesh's can't be batched, so always create a new batch
				_new_batch(r_batch_broken, r_index);

				Color modulate(1, 1, 1, 1);
				state.canvas_instance_batches[state.current_batch_index].shader_variant = CanvasShaderGLES3::MODE_ATTRIBUTES;
				if (c->type == Item::Command::TYPE_MESH) {
					const Item::CommandMesh *m = static_cast<const Item::CommandMesh *>(c);
					state.canvas_instance_batches[state.current_batch_index].tex = m->texture;
					_update_transform_2d_to_mat2x3(base_transform * draw_transform * m->transform, state.instance_data_array[r_index].world);
					modulate = m->modulate;
				} else if (c->type == Item::Command::TYPE_MULTIMESH) {
					const Item::CommandMultiMesh *mm = static_cast<const Item::CommandMultiMesh *>(c);
					state.canvas_instance_batches[state.current_batch_index].tex = mm->texture;
					uint32_t instance_count = GLES3::MeshStorage::get_singleton()->multimesh_get_instances_to_draw(mm->multimesh);
					if (instance_count > 1) {
						state.canvas_instance_batches[state.current_batch_index].shader_variant = CanvasShaderGLES3::MODE_INSTANCED;
					}
				} else if (c->type == Item::Command::TYPE_PARTICLES) {
					WARN_PRINT_ONCE("Particles not supported yet, sorry :(");
				}

				state.canvas_instance_batches[state.current_batch_index].command = c;
				state.canvas_instance_batches[state.current_batch_index].command_type = c->type;

				_prepare_canvas_texture(state.canvas_instance_batches[state.current_batch_index].tex, state.canvas_instance_batches[state.current_batch_index].filter, state.canvas_instance_batches[state.current_batch_index].repeat, r_index, texpixel_size);

				state.instance_data_array[r_index].modulation[0] = base_color.r * modulate.r;
				state.instance_data_array[r_index].modulation[1] = base_color.g * modulate.g;
				state.instance_data_array[r_index].modulation[2] = base_color.b * modulate.b;
				state.instance_data_array[r_index].modulation[3] = base_color.a * modulate.a;

				for (int j = 0; j < 4; j++) {
					state.instance_data_array[r_index].src_rect[j] = 0;
					state.instance_data_array[r_index].dst_rect[j] = 0;
					state.instance_data_array[r_index].ninepatch_margins[j] = 0;
				}
				_add_to_batch(r_index, r_batch_broken);
			} break;

			case Item::Command::TYPE_TRANSFORM: {
				const Item::CommandTransform *transform = static_cast<const Item::CommandTransform *>(c);
				draw_transform = transform->xform;
			} break;

			case Item::Command::TYPE_CLIP_IGNORE: {
				const Item::CommandClipIgnore *ci = static_cast<const Item::CommandClipIgnore *>(c);
				if (current_clip) {
					if (ci->ignore != reclip) {
						_new_batch(r_batch_broken, r_index);
						if (ci->ignore) {
							state.canvas_instance_batches[state.current_batch_index].clip = nullptr;
							reclip = true;
						} else {
							state.canvas_instance_batches[state.current_batch_index].clip = current_clip;
							reclip = false;
						}
					}
				}
			} break;

			case Item::Command::TYPE_ANIMATION_SLICE: {
				const Item::CommandAnimationSlice *as = static_cast<const Item::CommandAnimationSlice *>(c);
				double current_time = RSG::rasterizer->get_total_time();
				double local_time = Math::fposmod(current_time - as->offset, as->animation_length);
				skipping = !(local_time >= as->slice_begin && local_time < as->slice_end);

				RenderingServerDefault::redraw_request(); // animation visible means redraw request
			} break;
		}

		c = c->next;
		r_batch_broken = false;
	}

	if (current_clip && reclip) {
		//will make it re-enable clipping if needed afterwards
		current_clip = nullptr;
	}
}

void RasterizerCanvasGLES3::_render_batch(Light *p_lights, uint32_t p_index) {
	ERR_FAIL_COND(!state.canvas_instance_batches[state.current_batch_index].command);

	// Used by Polygon and Mesh.
	static const GLenum prim[5] = { GL_POINTS, GL_LINES, GL_LINE_STRIP, GL_TRIANGLES, GL_TRIANGLE_STRIP };

	_bind_canvas_texture(state.canvas_instance_batches[p_index].tex, state.canvas_instance_batches[p_index].filter, state.canvas_instance_batches[p_index].repeat);

	// Bind the region of the UBO used by this batch.
	// If region exceeds the boundary of the UBO, just ignore.
	uint32_t range_bytes = data.max_instances_per_batch * sizeof(InstanceData);
	if (state.canvas_instance_batches[p_index].start >= (data.max_instances_per_ubo - 1) * sizeof(InstanceData)) {
		return;
	} else if (state.canvas_instance_batches[p_index].start >= (data.max_instances_per_ubo - data.max_instances_per_batch) * sizeof(InstanceData)) {
		// If we have less than a full batch at the end, we can just draw it anyway.
		// OpenGL will complain about the UBO being smaller than expected, but it should render fine.
		range_bytes = (data.max_instances_per_ubo - 1) * sizeof(InstanceData) - state.canvas_instance_batches[p_index].start;
	}

	uint32_t range_start = state.canvas_instance_batches[p_index].start;
	glBindBufferRange(GL_UNIFORM_BUFFER, INSTANCE_UNIFORM_LOCATION, state.canvas_instance_data_buffers[state.current_buffer].ubo, range_start, range_bytes);

	switch (state.canvas_instance_batches[p_index].command_type) {
		case Item::Command::TYPE_RECT:
		case Item::Command::TYPE_NINEPATCH: {
			glBindVertexArray(data.indexed_quad_array);
			glDrawElements(GL_TRIANGLES, state.canvas_instance_batches[p_index].instance_count * 6, GL_UNSIGNED_INT, 0);
			glBindBuffer(GL_UNIFORM_BUFFER, 0);
			glBindVertexArray(0);

		} break;

		case Item::Command::TYPE_POLYGON: {
			const Item::CommandPolygon *polygon = static_cast<const Item::CommandPolygon *>(state.canvas_instance_batches[p_index].command);

			PolygonBuffers *pb = polygon_buffers.polygons.getptr(polygon->polygon.polygon_id);
			ERR_FAIL_COND(!pb);

			glBindVertexArray(pb->vertex_array);

			if (pb->color_disabled && pb->color != Color(1.0, 1.0, 1.0, 1.0)) {
				glVertexAttrib4f(RS::ARRAY_COLOR, pb->color.r, pb->color.g, pb->color.b, pb->color.a);
			}

			if (pb->index_buffer != 0) {
				glDrawElements(prim[polygon->primitive], pb->count, GL_UNSIGNED_INT, nullptr);
			} else {
				glDrawArrays(prim[polygon->primitive], 0, pb->count);
			}
			glBindVertexArray(0);
			glBindBuffer(GL_UNIFORM_BUFFER, 0);

			if (pb->color_disabled && pb->color != Color(1.0, 1.0, 1.0, 1.0)) {
				// Reset so this doesn't pollute other draw calls.
				glVertexAttrib4f(RS::ARRAY_COLOR, 1.0, 1.0, 1.0, 1.0);
			}
		} break;

		case Item::Command::TYPE_PRIMITIVE: {
			glBindVertexArray(data.canvas_quad_array);
			const GLenum primitive[5] = { GL_POINTS, GL_POINTS, GL_LINES, GL_TRIANGLES, GL_TRIANGLES };
			int instance_count = state.canvas_instance_batches[p_index].instance_count;
			if (instance_count > 1) {
				glDrawArraysInstanced(primitive[state.canvas_instance_batches[p_index].primitive_points], 0, state.canvas_instance_batches[p_index].primitive_points, instance_count);
			} else {
				glDrawArrays(primitive[state.canvas_instance_batches[p_index].primitive_points], 0, state.canvas_instance_batches[p_index].primitive_points);
			}
			glBindBuffer(GL_UNIFORM_BUFFER, 0);

		} break;

		case Item::Command::TYPE_MESH:
		case Item::Command::TYPE_MULTIMESH:
		case Item::Command::TYPE_PARTICLES: {
			GLES3::MeshStorage *mesh_storage = GLES3::MeshStorage::get_singleton();
			RID mesh;
			RID mesh_instance;
			RID texture;
			uint32_t instance_count = 1;
			GLuint multimesh_buffer = 0;
			uint32_t multimesh_stride = 0;
			uint32_t multimesh_color_offset = 0;
			bool multimesh_uses_color = false;
			bool multimesh_uses_custom_data = false;

			if (state.canvas_instance_batches[p_index].command_type == Item::Command::TYPE_MESH) {
				const Item::CommandMesh *m = static_cast<const Item::CommandMesh *>(state.canvas_instance_batches[p_index].command);
				mesh = m->mesh;
				mesh_instance = m->mesh_instance;
			} else if (state.canvas_instance_batches[p_index].command_type == Item::Command::TYPE_MULTIMESH) {
				const Item::CommandMultiMesh *mm = static_cast<const Item::CommandMultiMesh *>(state.canvas_instance_batches[p_index].command);
				RID multimesh = mm->multimesh;
				mesh = mesh_storage->multimesh_get_mesh(multimesh);

				if (mesh_storage->multimesh_get_transform_format(multimesh) != RS::MULTIMESH_TRANSFORM_2D) {
					break;
				}

				instance_count = mesh_storage->multimesh_get_instances_to_draw(multimesh);

				if (instance_count == 0) {
					break;
				}

				multimesh_buffer = mesh_storage->multimesh_get_gl_buffer(multimesh);
				multimesh_stride = mesh_storage->multimesh_get_stride(multimesh);
				multimesh_color_offset = mesh_storage->multimesh_get_color_offset(multimesh);
				multimesh_uses_color = mesh_storage->multimesh_uses_colors(multimesh);
				multimesh_uses_custom_data = mesh_storage->multimesh_uses_custom_data(multimesh);
			} else if (state.canvas_instance_batches[p_index].command_type == Item::Command::TYPE_PARTICLES) {
				// Do nothing for now.
			}

			ERR_FAIL_COND(mesh.is_null());

			uint32_t surf_count = mesh_storage->mesh_get_surface_count(mesh);

			for (uint32_t j = 0; j < surf_count; j++) {
				void *surface = mesh_storage->mesh_get_surface(mesh, j);

				RS::PrimitiveType primitive = mesh_storage->mesh_surface_get_primitive(surface);
				ERR_CONTINUE(primitive < 0 || primitive >= RS::PRIMITIVE_MAX);

				GLuint vertex_array_gl = 0;
				GLuint index_array_gl = 0;

				uint32_t input_mask = 0; // 2D meshes always use the same vertex format
				if (mesh_instance.is_valid()) {
					mesh_storage->mesh_instance_surface_get_vertex_arrays_and_format(mesh_instance, j, input_mask, vertex_array_gl);
				} else {
					mesh_storage->mesh_surface_get_vertex_arrays_and_format(surface, input_mask, vertex_array_gl);
				}

				index_array_gl = mesh_storage->mesh_surface_get_index_buffer(surface, 0);
				bool use_index_buffer = false;
				glBindVertexArray(vertex_array_gl);
				if (index_array_gl != 0) {
					glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, index_array_gl);
					use_index_buffer = true;
				}

				if (instance_count > 1) {
					// Bind instance buffers.
					glBindBuffer(GL_ARRAY_BUFFER, multimesh_buffer);
					glEnableVertexAttribArray(1);
					glVertexAttribPointer(1, 4, GL_FLOAT, GL_FALSE, multimesh_stride * sizeof(float), CAST_INT_TO_UCHAR_PTR(0));
					glVertexAttribDivisor(1, 1);
					glEnableVertexAttribArray(2);
					glVertexAttribPointer(2, 4, GL_FLOAT, GL_FALSE, multimesh_stride * sizeof(float), CAST_INT_TO_UCHAR_PTR(4 * 4));
					glVertexAttribDivisor(2, 1);

					if (multimesh_uses_color || multimesh_uses_custom_data) {
						glEnableVertexAttribArray(5);
						glVertexAttribIPointer(5, 4, GL_UNSIGNED_INT, multimesh_stride * sizeof(float), CAST_INT_TO_UCHAR_PTR(multimesh_color_offset * sizeof(float)));
						glVertexAttribDivisor(5, 1);
					}
				}

				GLenum primitive_gl = prim[int(primitive)];
				if (instance_count == 1) {
					if (use_index_buffer) {
						glDrawElements(primitive_gl, mesh_storage->mesh_surface_get_vertices_drawn_count(surface), mesh_storage->mesh_surface_get_index_type(surface), 0);
					} else {
						glDrawArrays(primitive_gl, 0, mesh_storage->mesh_surface_get_vertices_drawn_count(surface));
					}
				} else if (instance_count > 1) {
					if (use_index_buffer) {
						glDrawElementsInstanced(primitive_gl, mesh_storage->mesh_surface_get_vertices_drawn_count(surface), mesh_storage->mesh_surface_get_index_type(surface), 0, instance_count);
					} else {
						glDrawArraysInstanced(primitive_gl, 0, mesh_storage->mesh_surface_get_vertices_drawn_count(surface), instance_count);
					}
				}

				glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0);
				if (instance_count > 1) {
					glDisableVertexAttribArray(5);
					glDisableVertexAttribArray(6);
					glDisableVertexAttribArray(7);
					glDisableVertexAttribArray(8);
				}
			}

		} break;
		case Item::Command::TYPE_TRANSFORM:
		case Item::Command::TYPE_CLIP_IGNORE:
		case Item::Command::TYPE_ANIMATION_SLICE: {
			// Can ignore these as they only impact batch creation.
		} break;
	}
}

void RasterizerCanvasGLES3::_add_to_batch(uint32_t &r_index, bool &r_batch_broken) {
	if (r_index >= data.max_instances_per_ubo - 1) {
		WARN_PRINT_ONCE("Trying to draw too many items. Please increase maximum number of items in the project settings 'rendering/gl_compatibility/item_buffer_size'");
		return;
	}

	if (state.canvas_instance_batches[state.current_batch_index].instance_count >= data.max_instances_per_batch) {
		_new_batch(r_batch_broken, r_index);
	}

	state.canvas_instance_batches[state.current_batch_index].instance_count++;
	r_index++;
}

void RasterizerCanvasGLES3::_new_batch(bool &r_batch_broken, uint32_t &r_index) {
	if (state.canvas_instance_batches.size() == 0) {
		state.canvas_instance_batches.push_back(Batch());
		return;
	}

	if (r_batch_broken || state.canvas_instance_batches[state.current_batch_index].instance_count == 0) {
		return;
	}

	r_batch_broken = true;

	// Copy the properties of the current batch, we will manually update the things that changed.
	Batch new_batch = state.canvas_instance_batches[state.current_batch_index];
	new_batch.instance_count = 0;
	new_batch.start = state.canvas_instance_batches[state.current_batch_index].start + state.canvas_instance_batches[state.current_batch_index].instance_count * sizeof(InstanceData);

	state.current_batch_index++;
	state.canvas_instance_batches.push_back(new_batch);
	_align_instance_data_buffer(r_index);
}

void RasterizerCanvasGLES3::_bind_material(GLES3::CanvasMaterialData *p_material_data, CanvasShaderGLES3::ShaderVariant p_variant, uint64_t p_specialization) {
	if (p_material_data) {
		if (p_material_data->shader_data->version.is_valid() && p_material_data->shader_data->valid) {
			// Bind uniform buffer and textures
			p_material_data->bind_uniforms();
			GLES3::MaterialStorage::get_singleton()->shaders.canvas_shader.version_bind_shader(p_material_data->shader_data->version, p_variant, p_specialization);
		} else {
			GLES3::MaterialStorage::get_singleton()->shaders.canvas_shader.version_bind_shader(data.canvas_shader_default_version, p_variant, p_specialization);
		}
	} else {
		GLES3::MaterialStorage::get_singleton()->shaders.canvas_shader.version_bind_shader(data.canvas_shader_default_version, p_variant, p_specialization);
	}
}

RID RasterizerCanvasGLES3::light_create() {
	CanvasLight canvas_light;
	return canvas_light_owner.make_rid(canvas_light);
}

void RasterizerCanvasGLES3::light_set_texture(RID p_rid, RID p_texture) {
	GLES3::TextureStorage *texture_storage = GLES3::TextureStorage::get_singleton();

	CanvasLight *cl = canvas_light_owner.get_or_null(p_rid);
	ERR_FAIL_COND(!cl);
	if (cl->texture == p_texture) {
		return;
	}
	if (cl->texture.is_valid()) {
		texture_storage->texture_remove_from_texture_atlas(cl->texture);
	}
	cl->texture = p_texture;

	if (cl->texture.is_valid()) {
		texture_storage->texture_add_to_texture_atlas(cl->texture);
	}
}

void RasterizerCanvasGLES3::light_set_use_shadow(RID p_rid, bool p_enable) {
}

void RasterizerCanvasGLES3::light_update_shadow(RID p_rid, int p_shadow_index, const Transform2D &p_light_xform, int p_light_mask, float p_near, float p_far, LightOccluderInstance *p_occluders) {
}

void RasterizerCanvasGLES3::light_update_directional_shadow(RID p_rid, int p_shadow_index, const Transform2D &p_light_xform, int p_light_mask, float p_cull_distance, const Rect2 &p_clip_rect, LightOccluderInstance *p_occluders) {
}

void RasterizerCanvasGLES3::render_sdf(RID p_render_target, LightOccluderInstance *p_occluders) {
}

RID RasterizerCanvasGLES3::occluder_polygon_create() {
	return RID();
}

void RasterizerCanvasGLES3::occluder_polygon_set_shape(RID p_occluder, const Vector<Vector2> &p_points, bool p_closed) {
}

void RasterizerCanvasGLES3::occluder_polygon_set_cull_mode(RID p_occluder, RS::CanvasOccluderPolygonCullMode p_mode) {
}

void RasterizerCanvasGLES3::set_shadow_texture_size(int p_size) {
}

bool RasterizerCanvasGLES3::free(RID p_rid) {
	if (canvas_light_owner.owns(p_rid)) {
		CanvasLight *cl = canvas_light_owner.get_or_null(p_rid);
		ERR_FAIL_COND_V(!cl, false);
		canvas_light_owner.free(p_rid);
	} else {
		return false;
	}

	return true;
}

void RasterizerCanvasGLES3::update() {
}

void RasterizerCanvasGLES3::canvas_begin(RID p_to_render_target, bool p_to_backbuffer) {
	GLES3::TextureStorage *texture_storage = GLES3::TextureStorage::get_singleton();
	GLES3::Config *config = GLES3::Config::get_singleton();

	GLES3::RenderTarget *render_target = texture_storage->get_render_target(p_to_render_target);

	if (p_to_backbuffer) {
		glBindFramebuffer(GL_FRAMEBUFFER, render_target->backbuffer_fbo);
		glActiveTexture(GL_TEXTURE0 + config->max_texture_image_units - 4);
		GLES3::Texture *tex = texture_storage->get_texture(texture_storage->texture_gl_get_default(GLES3::DEFAULT_GL_TEXTURE_WHITE));
		glBindTexture(GL_TEXTURE_2D, tex->tex_id);
	} else {
		glBindFramebuffer(GL_FRAMEBUFFER, render_target->fbo);
		glActiveTexture(GL_TEXTURE0 + config->max_texture_image_units - 4);
		glBindTexture(GL_TEXTURE_2D, render_target->backbuffer);
	}

	if (render_target->is_transparent || p_to_backbuffer) {
		state.transparent_render_target = true;
		glBlendFuncSeparate(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA, GL_ONE, GL_ONE_MINUS_SRC_ALPHA);
	} else {
		state.transparent_render_target = false;
		glBlendFuncSeparate(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA, GL_ZERO, GL_ONE);
	}

	if (render_target && render_target->clear_requested) {
		const Color &col = render_target->clear_color;
		glClearColor(col.r, col.g, col.b, col.a);

		glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT);
		render_target->clear_requested = false;
	}

	glActiveTexture(GL_TEXTURE0);
	GLES3::Texture *tex = texture_storage->get_texture(texture_storage->texture_gl_get_default(GLES3::DEFAULT_GL_TEXTURE_WHITE));
	glBindTexture(GL_TEXTURE_2D, tex->tex_id);
}

void RasterizerCanvasGLES3::_bind_canvas_texture(RID p_texture, RS::CanvasItemTextureFilter p_base_filter, RS::CanvasItemTextureRepeat p_base_repeat) {
	GLES3::TextureStorage *texture_storage = GLES3::TextureStorage::get_singleton();
	GLES3::Config *config = GLES3::Config::get_singleton();

	if (p_texture == RID()) {
		p_texture = default_canvas_texture;
	}

	if (state.current_tex == p_texture && state.current_filter_mode == p_base_filter && state.current_repeat_mode == p_base_repeat) {
		return;
	}

	state.current_tex = p_texture;
	state.current_filter_mode = p_base_filter;
	state.current_repeat_mode = p_base_repeat;

	GLES3::CanvasTexture *ct = nullptr;

	GLES3::Texture *t = texture_storage->get_texture(p_texture);

	if (t) {
		ERR_FAIL_COND(!t->canvas_texture);
		ct = t->canvas_texture;
	} else {
		ct = texture_storage->get_canvas_texture(p_texture);
	}

	if (!ct) {
		// Invalid Texture RID.
		_bind_canvas_texture(default_canvas_texture, p_base_filter, p_base_repeat);
		return;
	}

	RS::CanvasItemTextureFilter filter = ct->texture_filter != RS::CANVAS_ITEM_TEXTURE_FILTER_DEFAULT ? ct->texture_filter : p_base_filter;
	ERR_FAIL_COND(filter == RS::CANVAS_ITEM_TEXTURE_FILTER_DEFAULT);

	RS::CanvasItemTextureRepeat repeat = ct->texture_repeat != RS::CANVAS_ITEM_TEXTURE_REPEAT_DEFAULT ? ct->texture_repeat : p_base_repeat;
	ERR_FAIL_COND(repeat == RS::CANVAS_ITEM_TEXTURE_REPEAT_DEFAULT);

	GLES3::Texture *texture = texture_storage->get_texture(ct->diffuse);

	if (!texture) {
		GLES3::Texture *tex = texture_storage->get_texture(texture_storage->texture_gl_get_default(GLES3::DEFAULT_GL_TEXTURE_WHITE));
		glActiveTexture(GL_TEXTURE0);
		glBindTexture(GL_TEXTURE_2D, tex->tex_id);
	} else {
		glActiveTexture(GL_TEXTURE0);
		glBindTexture(GL_TEXTURE_2D, texture->tex_id);
		texture->gl_set_filter(filter);
		texture->gl_set_repeat(repeat);
	}

	GLES3::Texture *normal_map = texture_storage->get_texture(ct->normal_map);

	if (!normal_map) {
		glActiveTexture(GL_TEXTURE0 + config->max_texture_image_units - 6);
		GLES3::Texture *tex = texture_storage->get_texture(texture_storage->texture_gl_get_default(GLES3::DEFAULT_GL_TEXTURE_NORMAL));
		glBindTexture(GL_TEXTURE_2D, tex->tex_id);
	} else {
		glActiveTexture(GL_TEXTURE0 + config->max_texture_image_units - 6);
		glBindTexture(GL_TEXTURE_2D, normal_map->tex_id);
		normal_map->gl_set_filter(filter);
		normal_map->gl_set_repeat(repeat);
	}

	GLES3::Texture *specular_map = texture_storage->get_texture(ct->specular);

	if (!specular_map) {
		glActiveTexture(GL_TEXTURE0 + config->max_texture_image_units - 7);
		GLES3::Texture *tex = texture_storage->get_texture(texture_storage->texture_gl_get_default(GLES3::DEFAULT_GL_TEXTURE_WHITE));
		glBindTexture(GL_TEXTURE_2D, tex->tex_id);
	} else {
		glActiveTexture(GL_TEXTURE0 + config->max_texture_image_units - 7);
		glBindTexture(GL_TEXTURE_2D, specular_map->tex_id);
		specular_map->gl_set_filter(filter);
		specular_map->gl_set_repeat(repeat);
	}
}

void RasterizerCanvasGLES3::_prepare_canvas_texture(RID p_texture, RS::CanvasItemTextureFilter p_base_filter, RS::CanvasItemTextureRepeat p_base_repeat, uint32_t &r_index, Size2 &r_texpixel_size) {
	GLES3::TextureStorage *texture_storage = GLES3::TextureStorage::get_singleton();

	if (p_texture == RID()) {
		p_texture = default_canvas_texture;
	}

	GLES3::CanvasTexture *ct = nullptr;

	GLES3::Texture *t = texture_storage->get_texture(p_texture);

	if (t) {
		//regular texture
		if (!t->canvas_texture) {
			t->canvas_texture = memnew(GLES3::CanvasTexture);
			t->canvas_texture->diffuse = p_texture;
		}

		ct = t->canvas_texture;
	} else {
		ct = texture_storage->get_canvas_texture(p_texture);
	}

	if (!ct) {
		// Invalid Texture RID.
		_prepare_canvas_texture(default_canvas_texture, p_base_filter, p_base_repeat, r_index, r_texpixel_size);
		return;
	}

	GLES3::Texture *texture = texture_storage->get_texture(ct->diffuse);
	Size2i size_cache;
	if (!texture) {
		ct->diffuse = texture_storage->texture_gl_get_default(GLES3::DEFAULT_GL_TEXTURE_WHITE);
		GLES3::Texture *tex = texture_storage->get_texture(ct->diffuse);
		size_cache = Size2i(tex->width, tex->height);
	} else {
		size_cache = Size2i(texture->width, texture->height);
	}

	GLES3::Texture *normal_map = texture_storage->get_texture(ct->normal_map);

	if (ct->specular_color.a < 0.999) {
		state.instance_data_array[r_index].flags |= FLAGS_DEFAULT_SPECULAR_MAP_USED;
	} else {
		state.instance_data_array[r_index].flags &= ~FLAGS_DEFAULT_SPECULAR_MAP_USED;
	}

	if (normal_map) {
		state.instance_data_array[r_index].flags |= FLAGS_DEFAULT_NORMAL_MAP_USED;
	} else {
		state.instance_data_array[r_index].flags &= ~FLAGS_DEFAULT_NORMAL_MAP_USED;
	}

	state.instance_data_array[r_index].specular_shininess = uint32_t(CLAMP(ct->specular_color.a * 255.0, 0, 255)) << 24;
	state.instance_data_array[r_index].specular_shininess |= uint32_t(CLAMP(ct->specular_color.b * 255.0, 0, 255)) << 16;
	state.instance_data_array[r_index].specular_shininess |= uint32_t(CLAMP(ct->specular_color.g * 255.0, 0, 255)) << 8;
	state.instance_data_array[r_index].specular_shininess |= uint32_t(CLAMP(ct->specular_color.r * 255.0, 0, 255));

	r_texpixel_size.x = 1.0 / float(size_cache.x);
	r_texpixel_size.y = 1.0 / float(size_cache.y);

	state.instance_data_array[r_index].color_texture_pixel_size[0] = r_texpixel_size.x;
	state.instance_data_array[r_index].color_texture_pixel_size[1] = r_texpixel_size.y;
}

void RasterizerCanvasGLES3::reset_canvas() {
	glDisable(GL_CULL_FACE);
	glDisable(GL_DEPTH_TEST);
	glDisable(GL_SCISSOR_TEST);
	glEnable(GL_BLEND);
	glBlendFuncSeparate(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA, GL_ZERO, GL_ONE);

	glBindBuffer(GL_ARRAY_BUFFER, 0);
	glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0);
}

void RasterizerCanvasGLES3::canvas_debug_viewport_shadows(Light *p_lights_with_shadow) {
}

void RasterizerCanvasGLES3::canvas_light_shadow_buffer_update(RID p_buffer, const Transform2D &p_light_xform, int p_light_mask, float p_near, float p_far, LightOccluderInstance *p_occluders, Projection *p_xform_cache) {
}

void RasterizerCanvasGLES3::draw_lens_distortion_rect(const Rect2 &p_rect, float p_k1, float p_k2, const Vector2 &p_eye_center, float p_oversample) {
}

RendererCanvasRender::PolygonID RasterizerCanvasGLES3::request_polygon(const Vector<int> &p_indices, const Vector<Point2> &p_points, const Vector<Color> &p_colors, const Vector<Point2> &p_uvs, const Vector<int> &p_bones, const Vector<float> &p_weights) {
	// We interleave the vertex data into one big VBO to improve cache coherence
	uint32_t vertex_count = p_points.size();
	uint32_t stride = 2;
	if ((uint32_t)p_colors.size() == vertex_count) {
		stride += 4;
	}
	if ((uint32_t)p_uvs.size() == vertex_count) {
		stride += 2;
	}
	if ((uint32_t)p_bones.size() == vertex_count * 4 && (uint32_t)p_weights.size() == vertex_count * 4) {
		stride += 4;
	}

	PolygonBuffers pb;
	glGenBuffers(1, &pb.vertex_buffer);
	glGenVertexArrays(1, &pb.vertex_array);
	glBindVertexArray(pb.vertex_array);
	pb.count = vertex_count;
	pb.index_buffer = 0;

	uint32_t buffer_size = stride * p_points.size();

	Vector<uint8_t> polygon_buffer;
	polygon_buffer.resize(buffer_size * sizeof(float));
	{
		glBindBuffer(GL_ARRAY_BUFFER, pb.vertex_buffer);
		glBufferData(GL_ARRAY_BUFFER, stride * vertex_count * sizeof(float), nullptr, GL_STATIC_DRAW); // TODO may not be necessary
		uint8_t *r = polygon_buffer.ptrw();
		float *fptr = reinterpret_cast<float *>(r);
		uint32_t *uptr = (uint32_t *)r;
		uint32_t base_offset = 0;
		{
			// Always uses vertex positions
			glEnableVertexAttribArray(RS::ARRAY_VERTEX);
			glVertexAttribPointer(RS::ARRAY_VERTEX, 2, GL_FLOAT, GL_FALSE, stride * sizeof(float), nullptr);
			const Vector2 *points_ptr = p_points.ptr();

			for (uint32_t i = 0; i < vertex_count; i++) {
				fptr[base_offset + i * stride + 0] = points_ptr[i].x;
				fptr[base_offset + i * stride + 1] = points_ptr[i].y;
			}

			base_offset += 2;
		}

		// Next add colors
		if ((uint32_t)p_colors.size() == vertex_count) {
			glEnableVertexAttribArray(RS::ARRAY_COLOR);
			glVertexAttribPointer(RS::ARRAY_COLOR, 4, GL_FLOAT, GL_FALSE, stride * sizeof(float), CAST_INT_TO_UCHAR_PTR(base_offset * sizeof(float)));

			const Color *color_ptr = p_colors.ptr();

			for (uint32_t i = 0; i < vertex_count; i++) {
				fptr[base_offset + i * stride + 0] = color_ptr[i].r;
				fptr[base_offset + i * stride + 1] = color_ptr[i].g;
				fptr[base_offset + i * stride + 2] = color_ptr[i].b;
				fptr[base_offset + i * stride + 3] = color_ptr[i].a;
			}
			base_offset += 4;
		} else {
			glDisableVertexAttribArray(RS::ARRAY_COLOR);
			pb.color_disabled = true;
			pb.color = p_colors.size() == 1 ? p_colors[0] : Color(1.0, 1.0, 1.0, 1.0);
		}

		if ((uint32_t)p_uvs.size() == vertex_count) {
			glEnableVertexAttribArray(RS::ARRAY_TEX_UV);
			glVertexAttribPointer(RS::ARRAY_TEX_UV, 2, GL_FLOAT, GL_FALSE, stride * sizeof(float), CAST_INT_TO_UCHAR_PTR(base_offset * sizeof(float)));

			const Vector2 *uv_ptr = p_uvs.ptr();

			for (uint32_t i = 0; i < vertex_count; i++) {
				fptr[base_offset + i * stride + 0] = uv_ptr[i].x;
				fptr[base_offset + i * stride + 1] = uv_ptr[i].y;
			}

			base_offset += 2;
		} else {
			glDisableVertexAttribArray(RS::ARRAY_TEX_UV);
		}

		if ((uint32_t)p_indices.size() == vertex_count * 4 && (uint32_t)p_weights.size() == vertex_count * 4) {
			glEnableVertexAttribArray(RS::ARRAY_BONES);
			glVertexAttribPointer(RS::ARRAY_BONES, 4, GL_UNSIGNED_INT, GL_FALSE, stride * sizeof(float), CAST_INT_TO_UCHAR_PTR(base_offset * sizeof(float)));

			const int *bone_ptr = p_bones.ptr();

			for (uint32_t i = 0; i < vertex_count; i++) {
				uint16_t *bone16w = (uint16_t *)&uptr[base_offset + i * stride];

				bone16w[0] = bone_ptr[i * 4 + 0];
				bone16w[1] = bone_ptr[i * 4 + 1];
				bone16w[2] = bone_ptr[i * 4 + 2];
				bone16w[3] = bone_ptr[i * 4 + 3];
			}

			base_offset += 2;
		} else {
			glDisableVertexAttribArray(RS::ARRAY_BONES);
		}

		if ((uint32_t)p_weights.size() == vertex_count * 4) {
			glEnableVertexAttribArray(RS::ARRAY_WEIGHTS);
			glVertexAttribPointer(RS::ARRAY_WEIGHTS, 4, GL_FLOAT, GL_FALSE, stride * sizeof(float), CAST_INT_TO_UCHAR_PTR(base_offset * sizeof(float)));

			const float *weight_ptr = p_weights.ptr();

			for (uint32_t i = 0; i < vertex_count; i++) {
				uint16_t *weight16w = (uint16_t *)&uptr[base_offset + i * stride];

				weight16w[0] = CLAMP(weight_ptr[i * 4 + 0] * 65535, 0, 65535);
				weight16w[1] = CLAMP(weight_ptr[i * 4 + 1] * 65535, 0, 65535);
				weight16w[2] = CLAMP(weight_ptr[i * 4 + 2] * 65535, 0, 65535);
				weight16w[3] = CLAMP(weight_ptr[i * 4 + 3] * 65535, 0, 65535);
			}

			base_offset += 2;
		} else {
			glDisableVertexAttribArray(RS::ARRAY_WEIGHTS);
		}

		ERR_FAIL_COND_V(base_offset != stride, 0);
		glBufferData(GL_ARRAY_BUFFER, vertex_count * stride * sizeof(float), polygon_buffer.ptr(), GL_STATIC_DRAW);
	}

	if (p_indices.size()) {
		//create indices, as indices were requested
		Vector<uint8_t> index_buffer;
		index_buffer.resize(p_indices.size() * sizeof(int32_t));
		{
			uint8_t *w = index_buffer.ptrw();
			memcpy(w, p_indices.ptr(), sizeof(int32_t) * p_indices.size());
		}
		glGenBuffers(1, &pb.index_buffer);
		glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, pb.index_buffer);
		glBufferData(GL_ELEMENT_ARRAY_BUFFER, p_indices.size() * 4, nullptr, GL_STATIC_DRAW); // TODO may not be necessary
		glBufferData(GL_ELEMENT_ARRAY_BUFFER, p_indices.size() * 4, index_buffer.ptr(), GL_STATIC_DRAW);
		pb.count = p_indices.size();
	}

	glBindVertexArray(0);
	glBindBuffer(GL_ARRAY_BUFFER, 0);
	glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0);

	PolygonID id = polygon_buffers.last_id++;

	polygon_buffers.polygons[id] = pb;

	return id;
}
void RasterizerCanvasGLES3::free_polygon(PolygonID p_polygon) {
	PolygonBuffers *pb_ptr = polygon_buffers.polygons.getptr(p_polygon);
	ERR_FAIL_COND(!pb_ptr);

	PolygonBuffers &pb = *pb_ptr;

	if (pb.index_buffer != 0) {
		glDeleteBuffers(1, &pb.index_buffer);
	}

	glDeleteVertexArrays(1, &pb.vertex_array);
	glDeleteBuffers(1, &pb.vertex_buffer);

	polygon_buffers.polygons.erase(p_polygon);
}

// Creates a new uniform buffer and uses it right away
// This expands the instance buffer continually
// In theory allocations can reach as high as number of windows * 3 frames
// because OpenGL can start rendering subsequent frames before finishing the current one
void RasterizerCanvasGLES3::_allocate_instance_data_buffer() {
	GLuint new_buffers[3];
	glGenBuffers(3, new_buffers);
	// Batch UBO.
	glBindBuffer(GL_UNIFORM_BUFFER, new_buffers[0]);
	glBufferData(GL_UNIFORM_BUFFER, data.max_instance_buffer_size, nullptr, GL_STREAM_DRAW);
	// Light uniform buffer.
	glBindBuffer(GL_UNIFORM_BUFFER, new_buffers[1]);
	glBufferData(GL_UNIFORM_BUFFER, sizeof(LightUniform) * data.max_lights_per_render, nullptr, GL_STREAM_DRAW);
	// State buffer.
	glBindBuffer(GL_UNIFORM_BUFFER, new_buffers[2]);
	glBufferData(GL_UNIFORM_BUFFER, sizeof(StateBuffer), nullptr, GL_STREAM_DRAW);

	state.current_buffer = (state.current_buffer + 1);
	DataBuffer db;
	db.ubo = new_buffers[0];
	db.light_ubo = new_buffers[1];
	db.state_ubo = new_buffers[2];
	db.last_frame_used = RSG::rasterizer->get_frame_number();
	state.canvas_instance_data_buffers.insert(state.current_buffer, db);
	state.current_buffer = state.current_buffer % state.canvas_instance_data_buffers.size();
	glBindBuffer(GL_UNIFORM_BUFFER, 0);
}

// Batch start positions need to be aligned to the device's GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT.
// This needs to be called anytime a new batch is created.
void RasterizerCanvasGLES3::_align_instance_data_buffer(uint32_t &r_index) {
	if (GLES3::Config::get_singleton()->uniform_buffer_offset_alignment > int(sizeof(InstanceData))) {
		uint32_t offset = state.canvas_instance_batches[state.current_batch_index].start % GLES3::Config::get_singleton()->uniform_buffer_offset_alignment;
		if (offset > 0) {
			// uniform_buffer_offset_alignment can be 4, 16, 32, or 256. Our instance batches are 128 bytes.
			// Accordingly, this branch is only triggered if we are 128 bytes off.
			uint32_t offset_bytes = GLES3::Config::get_singleton()->uniform_buffer_offset_alignment - offset;
			state.canvas_instance_batches[state.current_batch_index].start += offset_bytes;
			// Offset the instance array so it stays in sync with batch start points.
			// This creates gaps in the instance buffer with wasted space, but we can't help it.
			r_index += offset_bytes / sizeof(InstanceData);
			if (r_index > 0) {
				// In this case we need to copy over the basic data.
				state.instance_data_array[r_index] = state.instance_data_array[r_index - 1];
			}
		}
	}
}

void RasterizerCanvasGLES3::set_time(double p_time) {
	state.time = p_time;
}

RasterizerCanvasGLES3 *RasterizerCanvasGLES3::singleton = nullptr;

RasterizerCanvasGLES3 *RasterizerCanvasGLES3::get_singleton() {
	return singleton;
}

RasterizerCanvasGLES3::RasterizerCanvasGLES3() {
	singleton = this;
	GLES3::TextureStorage *texture_storage = GLES3::TextureStorage::get_singleton();
	GLES3::MaterialStorage *material_storage = GLES3::MaterialStorage::get_singleton();
	GLES3::Config *config = GLES3::Config::get_singleton();

	polygon_buffers.last_id = 1;
	// quad buffer
	{
		glGenBuffers(1, &data.canvas_quad_vertices);
		glBindBuffer(GL_ARRAY_BUFFER, data.canvas_quad_vertices);

		const float qv[8] = {
			0, 0,
			0, 1,
			1, 1,
			1, 0
		};

		glBufferData(GL_ARRAY_BUFFER, sizeof(float) * 8, qv, GL_STATIC_DRAW);

		glBindBuffer(GL_ARRAY_BUFFER, 0);

		glGenVertexArrays(1, &data.canvas_quad_array);
		glBindVertexArray(data.canvas_quad_array);
		glBindBuffer(GL_ARRAY_BUFFER, data.canvas_quad_vertices);
		glVertexAttribPointer(0, 2, GL_FLOAT, GL_FALSE, sizeof(float) * 2, nullptr);
		glEnableVertexAttribArray(0);
		glBindVertexArray(0);
		glBindBuffer(GL_ARRAY_BUFFER, 0); //unbind
	}

	{
		//particle quad buffers

		glGenBuffers(1, &data.particle_quad_vertices);
		glBindBuffer(GL_ARRAY_BUFFER, data.particle_quad_vertices);
		{
			//quad of size 1, with pivot on the center for particles, then regular UVS. Color is general plus fetched from particle
			const float qv[16] = {
				-0.5, -0.5,
				0.0, 0.0,
				-0.5, 0.5,
				0.0, 1.0,
				0.5, 0.5,
				1.0, 1.0,
				0.5, -0.5,
				1.0, 0.0
			};

			glBufferData(GL_ARRAY_BUFFER, sizeof(float) * 16, qv, GL_STATIC_DRAW);
		}

		glBindBuffer(GL_ARRAY_BUFFER, 0); //unbind

		glGenVertexArrays(1, &data.particle_quad_array);
		glBindVertexArray(data.particle_quad_array);
		glBindBuffer(GL_ARRAY_BUFFER, data.particle_quad_vertices);
		glEnableVertexAttribArray(RS::ARRAY_VERTEX);
		glVertexAttribPointer(RS::ARRAY_VERTEX, 2, GL_FLOAT, GL_FALSE, sizeof(float) * 4, nullptr);
		glEnableVertexAttribArray(RS::ARRAY_TEX_UV);
		glVertexAttribPointer(RS::ARRAY_TEX_UV, 2, GL_FLOAT, GL_FALSE, sizeof(float) * 4, CAST_INT_TO_UCHAR_PTR(8));
		glBindVertexArray(0);
		glBindBuffer(GL_ARRAY_BUFFER, 0); //unbind
	}

	// ninepatch buffers
	{
		// array buffer
		glGenBuffers(1, &data.ninepatch_vertices);
		glBindBuffer(GL_ARRAY_BUFFER, data.ninepatch_vertices);

		glBufferData(GL_ARRAY_BUFFER, sizeof(float) * (16 + 16) * 2, nullptr, GL_DYNAMIC_DRAW);

		glBindBuffer(GL_ARRAY_BUFFER, 0);

		// element buffer
		glGenBuffers(1, &data.ninepatch_elements);
		glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, data.ninepatch_elements);

#define _EIDX(y, x) (y * 4 + x)
		uint8_t elems[3 * 2 * 9] = {
			// first row

			_EIDX(0, 0), _EIDX(0, 1), _EIDX(1, 1),
			_EIDX(1, 1), _EIDX(1, 0), _EIDX(0, 0),

			_EIDX(0, 1), _EIDX(0, 2), _EIDX(1, 2),
			_EIDX(1, 2), _EIDX(1, 1), _EIDX(0, 1),

			_EIDX(0, 2), _EIDX(0, 3), _EIDX(1, 3),
			_EIDX(1, 3), _EIDX(1, 2), _EIDX(0, 2),

			// second row

			_EIDX(1, 0), _EIDX(1, 1), _EIDX(2, 1),
			_EIDX(2, 1), _EIDX(2, 0), _EIDX(1, 0),

			// the center one would be here, but we'll put it at the end
			// so it's easier to disable the center and be able to use
			// one draw call for both

			_EIDX(1, 2), _EIDX(1, 3), _EIDX(2, 3),
			_EIDX(2, 3), _EIDX(2, 2), _EIDX(1, 2),

			// third row

			_EIDX(2, 0), _EIDX(2, 1), _EIDX(3, 1),
			_EIDX(3, 1), _EIDX(3, 0), _EIDX(2, 0),

			_EIDX(2, 1), _EIDX(2, 2), _EIDX(3, 2),
			_EIDX(3, 2), _EIDX(3, 1), _EIDX(2, 1),

			_EIDX(2, 2), _EIDX(2, 3), _EIDX(3, 3),
			_EIDX(3, 3), _EIDX(3, 2), _EIDX(2, 2),

			// center field

			_EIDX(1, 1), _EIDX(1, 2), _EIDX(2, 2),
			_EIDX(2, 2), _EIDX(2, 1), _EIDX(1, 1)
		};
#undef _EIDX

		glBufferData(GL_ELEMENT_ARRAY_BUFFER, sizeof(elems), elems, GL_STATIC_DRAW);

		glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0);
	}

	int uniform_max_size = config->max_uniform_buffer_size;
	if (uniform_max_size < 65536) {
		data.max_lights_per_render = 64;
		data.max_instances_per_batch = 128;
	} else {
		data.max_lights_per_render = 256;
		data.max_instances_per_batch = 512;
	}

	// Reserve 3 Uniform Buffers for instance data Frame N, N+1 and N+2
	data.max_instances_per_ubo = MAX(data.max_instances_per_batch, uint32_t(GLOBAL_GET("rendering/gl_compatibility/item_buffer_size")));
	data.max_instance_buffer_size = data.max_instances_per_ubo * sizeof(InstanceData); // 16,384 instances * 128 bytes = 2,097,152 bytes = 2,048 kb
	state.canvas_instance_data_buffers.resize(3);
	state.canvas_instance_batches.reserve(200);

	for (int i = 0; i < 3; i++) {
		GLuint new_buffers[3];
		glGenBuffers(3, new_buffers);
		// Batch UBO.
		glBindBuffer(GL_UNIFORM_BUFFER, new_buffers[0]);
		glBufferData(GL_UNIFORM_BUFFER, data.max_instance_buffer_size, nullptr, GL_STREAM_DRAW);
		// Light uniform buffer.
		glBindBuffer(GL_UNIFORM_BUFFER, new_buffers[1]);
		glBufferData(GL_UNIFORM_BUFFER, sizeof(LightUniform) * data.max_lights_per_render, nullptr, GL_STREAM_DRAW);
		// State buffer.
		glBindBuffer(GL_UNIFORM_BUFFER, new_buffers[2]);
		glBufferData(GL_UNIFORM_BUFFER, sizeof(StateBuffer), nullptr, GL_STREAM_DRAW);
		DataBuffer db;
		db.ubo = new_buffers[0];
		db.light_ubo = new_buffers[1];
		db.state_ubo = new_buffers[2];
		db.last_frame_used = 0;
		db.fence = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
		state.canvas_instance_data_buffers[i] = db;
	}
	glBindBuffer(GL_UNIFORM_BUFFER, 0);

	state.instance_data_array = memnew_arr(InstanceData, data.max_instances_per_ubo);
	state.light_uniforms = memnew_arr(LightUniform, data.max_lights_per_render);

	{
		const uint32_t no_of_instances = data.max_instances_per_batch;

		glGenVertexArrays(1, &data.indexed_quad_array);
		glBindVertexArray(data.indexed_quad_array);
		glBindBuffer(GL_ARRAY_BUFFER, data.canvas_quad_vertices);

		const uint32_t num_indices = 6;
		const uint32_t quad_indices[num_indices] = { 0, 2, 1, 3, 2, 0 };

		const uint32_t total_indices = no_of_instances * num_indices;
		uint32_t *indices = new uint32_t[total_indices];
		for (uint32_t i = 0; i < total_indices; i++) {
			uint32_t quad = i / num_indices;
			uint32_t quad_local = i % num_indices;
			indices[i] = quad_indices[quad_local] + quad * num_indices;
		}

		glGenBuffers(1, &data.indexed_quad_buffer);
		glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, data.indexed_quad_buffer);
		glBufferData(GL_ELEMENT_ARRAY_BUFFER, sizeof(uint32_t) * total_indices, indices, GL_STATIC_DRAW);
		glBindVertexArray(0);
		delete[] indices;
	}

	String global_defines;
	global_defines += "#define MAX_GLOBAL_SHADER_UNIFORMS 256\n"; // TODO: this is arbitrary for now
	global_defines += "#define MAX_LIGHTS " + itos(data.max_lights_per_render) + "\n";
	global_defines += "#define MAX_DRAW_DATA_INSTANCES " + itos(data.max_instances_per_batch) + "\n";

	GLES3::MaterialStorage::get_singleton()->shaders.canvas_shader.initialize(global_defines);
	data.canvas_shader_default_version = GLES3::MaterialStorage::get_singleton()->shaders.canvas_shader.version_create();
	GLES3::MaterialStorage::get_singleton()->shaders.canvas_shader.version_bind_shader(data.canvas_shader_default_version, CanvasShaderGLES3::MODE_QUAD);

	{
		default_canvas_group_shader = material_storage->shader_allocate();
		material_storage->shader_initialize(default_canvas_group_shader);

		material_storage->shader_set_code(default_canvas_group_shader, R"(
// Default CanvasGroup shader.

shader_type canvas_item;

void fragment() {
	vec4 c = textureLod(SCREEN_TEXTURE, SCREEN_UV, 0.0);

	if (c.a > 0.0001) {
		c.rgb /= c.a;
	}

	COLOR *= c;
}
)");
		default_canvas_group_material = material_storage->material_allocate();
		material_storage->material_initialize(default_canvas_group_material);

		material_storage->material_set_shader(default_canvas_group_material, default_canvas_group_shader);
	}

	default_canvas_texture = texture_storage->canvas_texture_allocate();
	texture_storage->canvas_texture_initialize(default_canvas_texture);

	state.time = 0.0;
}

RasterizerCanvasGLES3::~RasterizerCanvasGLES3() {
	GLES3::MaterialStorage *material_storage = GLES3::MaterialStorage::get_singleton();

	material_storage->shaders.canvas_shader.version_free(data.canvas_shader_default_version);
	material_storage->material_free(default_canvas_group_material);
	material_storage->shader_free(default_canvas_group_shader);
	singleton = nullptr;

	glDeleteBuffers(1, &data.canvas_quad_vertices);
	glDeleteVertexArrays(1, &data.canvas_quad_array);

	glDeleteBuffers(1, &data.canvas_quad_vertices);
	glDeleteVertexArrays(1, &data.canvas_quad_array);

	GLES3::TextureStorage::get_singleton()->canvas_texture_free(default_canvas_texture);
	memdelete_arr(state.instance_data_array);
	memdelete_arr(state.light_uniforms);
}

#endif // GLES3_ENABLED