godot/thirdparty/meshoptimizer/meshoptimizer.h

/**
 * meshoptimizer - version 0.22
 *
 * Copyright (C) 2016-2024, by Arseny Kapoulkine (arseny.kapoulkine@gmail.com)
 * Report bugs and download new versions at https://github.com/zeux/meshoptimizer
 *
 * This library is distributed under the MIT License. See notice at the end of this file.
 */
#pragma once

#include <assert.h>
#include <stddef.h>

/* Version macro; major * 1000 + minor * 10 + patch */
#define MESHOPTIMIZER_VERSION 220 /* 0.22 */

/* If no API is defined, assume default */
#ifndef MESHOPTIMIZER_API
#define MESHOPTIMIZER_API
#endif

/* Set the calling-convention for alloc/dealloc function pointers */
#ifndef MESHOPTIMIZER_ALLOC_CALLCONV
#ifdef _MSC_VER
#define MESHOPTIMIZER_ALLOC_CALLCONV __cdecl
#else
#define MESHOPTIMIZER_ALLOC_CALLCONV
#endif
#endif

/* Experimental APIs have unstable interface and might have implementation that's not fully tested or optimized */
#ifndef MESHOPTIMIZER_EXPERIMENTAL
#define MESHOPTIMIZER_EXPERIMENTAL MESHOPTIMIZER_API
#endif

/* C interface */
#ifdef __cplusplus
extern "C"
{
#endif

/**
 * Vertex attribute stream
 * Each element takes size bytes, beginning at data, with stride controlling the spacing between successive elements (stride >= size).
 */
struct meshopt_Stream
{
	const void* data;
	size_t size;
	size_t stride;
};

/**
 * Generates a vertex remap table from the vertex buffer and an optional index buffer and returns number of unique vertices
 * As a result, all vertices that are binary equivalent map to the same (new) location, with no gaps in the resulting sequence.
 * Resulting remap table maps old vertices to new vertices and can be used in meshopt_remapVertexBuffer/meshopt_remapIndexBuffer.
 * Note that binary equivalence considers all vertex_size bytes, including padding which should be zero-initialized.
 *
 * destination must contain enough space for the resulting remap table (vertex_count elements)
 * indices can be NULL if the input is unindexed
 */
MESHOPTIMIZER_API size_t meshopt_generateVertexRemap(unsigned int* destination, const unsigned int* indices, size_t index_count, const void* vertices, size_t vertex_count, size_t vertex_size);

/**
 * Generates a vertex remap table from multiple vertex streams and an optional index buffer and returns number of unique vertices
 * As a result, all vertices that are binary equivalent map to the same (new) location, with no gaps in the resulting sequence.
 * Resulting remap table maps old vertices to new vertices and can be used in meshopt_remapVertexBuffer/meshopt_remapIndexBuffer.
 * To remap vertex buffers, you will need to call meshopt_remapVertexBuffer for each vertex stream.
 * Note that binary equivalence considers all size bytes in each stream, including padding which should be zero-initialized.
 *
 * destination must contain enough space for the resulting remap table (vertex_count elements)
 * indices can be NULL if the input is unindexed
 * stream_count must be <= 16
 */
MESHOPTIMIZER_API size_t meshopt_generateVertexRemapMulti(unsigned int* destination, const unsigned int* indices, size_t index_count, size_t vertex_count, const struct meshopt_Stream* streams, size_t stream_count);

/**
 * Generates vertex buffer from the source vertex buffer and remap table generated by meshopt_generateVertexRemap
 *
 * destination must contain enough space for the resulting vertex buffer (unique_vertex_count elements, returned by meshopt_generateVertexRemap)
 * vertex_count should be the initial vertex count and not the value returned by meshopt_generateVertexRemap
 */
MESHOPTIMIZER_API void meshopt_remapVertexBuffer(void* destination, const void* vertices, size_t vertex_count, size_t vertex_size, const unsigned int* remap);

/**
 * Generate index buffer from the source index buffer and remap table generated by meshopt_generateVertexRemap
 *
 * destination must contain enough space for the resulting index buffer (index_count elements)
 * indices can be NULL if the input is unindexed
 */
MESHOPTIMIZER_API void meshopt_remapIndexBuffer(unsigned int* destination, const unsigned int* indices, size_t index_count, const unsigned int* remap);

/**
 * Generate index buffer that can be used for more efficient rendering when only a subset of the vertex attributes is necessary
 * All vertices that are binary equivalent (wrt first vertex_size bytes) map to the first vertex in the original vertex buffer.
 * This makes it possible to use the index buffer for Z pre-pass or shadowmap rendering, while using the original index buffer for regular rendering.
 * Note that binary equivalence considers all vertex_size bytes, including padding which should be zero-initialized.
 *
 * destination must contain enough space for the resulting index buffer (index_count elements)
 */
MESHOPTIMIZER_API void meshopt_generateShadowIndexBuffer(unsigned int* destination, const unsigned int* indices, size_t index_count, const void* vertices, size_t vertex_count, size_t vertex_size, size_t vertex_stride);

/**
 * Generate index buffer that can be used for more efficient rendering when only a subset of the vertex attributes is necessary
 * All vertices that are binary equivalent (wrt specified streams) map to the first vertex in the original vertex buffer.
 * This makes it possible to use the index buffer for Z pre-pass or shadowmap rendering, while using the original index buffer for regular rendering.
 * Note that binary equivalence considers all size bytes in each stream, including padding which should be zero-initialized.
 *
 * destination must contain enough space for the resulting index buffer (index_count elements)
 * stream_count must be <= 16
 */
MESHOPTIMIZER_API void meshopt_generateShadowIndexBufferMulti(unsigned int* destination, const unsigned int* indices, size_t index_count, size_t vertex_count, const struct meshopt_Stream* streams, size_t stream_count);

/**
 * Generate index buffer that can be used as a geometry shader input with triangle adjacency topology
 * Each triangle is converted into a 6-vertex patch with the following layout:
 * - 0, 2, 4: original triangle vertices
 * - 1, 3, 5: vertices adjacent to edges 02, 24 and 40
 * The resulting patch can be rendered with geometry shaders using e.g. VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST_WITH_ADJACENCY.
 * This can be used to implement algorithms like silhouette detection/expansion and other forms of GS-driven rendering.
 *
 * destination must contain enough space for the resulting index buffer (index_count*2 elements)
 * vertex_positions should have float3 position in the first 12 bytes of each vertex
 */
MESHOPTIMIZER_API void meshopt_generateAdjacencyIndexBuffer(unsigned int* destination, const unsigned int* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride);

/**
 * Generate index buffer that can be used for PN-AEN tessellation with crack-free displacement
 * Each triangle is converted into a 12-vertex patch with the following layout:
 * - 0, 1, 2: original triangle vertices
 * - 3, 4: opposing edge for edge 0, 1
 * - 5, 6: opposing edge for edge 1, 2
 * - 7, 8: opposing edge for edge 2, 0
 * - 9, 10, 11: dominant vertices for corners 0, 1, 2
 * The resulting patch can be rendered with hardware tessellation using PN-AEN and displacement mapping.
 * See "Tessellation on Any Budget" (John McDonald, GDC 2011) for implementation details.
 *
 * destination must contain enough space for the resulting index buffer (index_count*4 elements)
 * vertex_positions should have float3 position in the first 12 bytes of each vertex
 */
MESHOPTIMIZER_API void meshopt_generateTessellationIndexBuffer(unsigned int* destination, const unsigned int* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride);

/**
 * Experimental: Generate index buffer that can be used for visibility buffer rendering and returns the size of the reorder table
 * Each triangle's provoking vertex index is equal to primitive id; this allows passing it to the fragment shader using nointerpolate attribute.
 * This is important for performance on hardware where primitive id can't be accessed efficiently in fragment shader.
 * The reorder table stores the original vertex id for each vertex in the new index buffer, and should be used in the vertex shader to load vertex data.
 * The provoking vertex is assumed to be the first vertex in the triangle; if this is not the case (OpenGL), rotate each triangle (abc -> bca) before rendering.
 * For maximum efficiency the input index buffer should be optimized for vertex cache first.
 *
 * destination must contain enough space for the resulting index buffer (index_count elements)
 * reorder must contain enough space for the worst case reorder table (vertex_count + index_count/3 elements)
 */
MESHOPTIMIZER_EXPERIMENTAL size_t meshopt_generateProvokingIndexBuffer(unsigned int* destination, unsigned int* reorder, const unsigned int* indices, size_t index_count, size_t vertex_count);

/**
 * Vertex transform cache optimizer
 * Reorders indices to reduce the number of GPU vertex shader invocations
 * If index buffer contains multiple ranges for multiple draw calls, this functions needs to be called on each range individually.
 *
 * destination must contain enough space for the resulting index buffer (index_count elements)
 */
MESHOPTIMIZER_API void meshopt_optimizeVertexCache(unsigned int* destination, const unsigned int* indices, size_t index_count, size_t vertex_count);

/**
 * Vertex transform cache optimizer for strip-like caches
 * Produces inferior results to meshopt_optimizeVertexCache from the GPU vertex cache perspective
 * However, the resulting index order is more optimal if the goal is to reduce the triangle strip length or improve compression efficiency
 *
 * destination must contain enough space for the resulting index buffer (index_count elements)
 */
MESHOPTIMIZER_API void meshopt_optimizeVertexCacheStrip(unsigned int* destination, const unsigned int* indices, size_t index_count, size_t vertex_count);

/**
 * Vertex transform cache optimizer for FIFO caches
 * Reorders indices to reduce the number of GPU vertex shader invocations
 * Generally takes ~3x less time to optimize meshes but produces inferior results compared to meshopt_optimizeVertexCache
 * If index buffer contains multiple ranges for multiple draw calls, this functions needs to be called on each range individually.
 *
 * destination must contain enough space for the resulting index buffer (index_count elements)
 * cache_size should be less than the actual GPU cache size to avoid cache thrashing
 */
MESHOPTIMIZER_API void meshopt_optimizeVertexCacheFifo(unsigned int* destination, const unsigned int* indices, size_t index_count, size_t vertex_count, unsigned int cache_size);

/**
 * Overdraw optimizer
 * Reorders indices to reduce the number of GPU vertex shader invocations and the pixel overdraw
 * If index buffer contains multiple ranges for multiple draw calls, this functions needs to be called on each range individually.
 *
 * destination must contain enough space for the resulting index buffer (index_count elements)
 * indices must contain index data that is the result of meshopt_optimizeVertexCache (*not* the original mesh indices!)
 * vertex_positions should have float3 position in the first 12 bytes of each vertex
 * threshold indicates how much the overdraw optimizer can degrade vertex cache efficiency (1.05 = up to 5%) to reduce overdraw more efficiently
 */
MESHOPTIMIZER_API void meshopt_optimizeOverdraw(unsigned int* destination, const unsigned int* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, float threshold);

/**
 * Vertex fetch cache optimizer
 * Reorders vertices and changes indices to reduce the amount of GPU memory fetches during vertex processing
 * Returns the number of unique vertices, which is the same as input vertex count unless some vertices are unused
 * This functions works for a single vertex stream; for multiple vertex streams, use meshopt_optimizeVertexFetchRemap + meshopt_remapVertexBuffer for each stream.
 *
 * destination must contain enough space for the resulting vertex buffer (vertex_count elements)
 * indices is used both as an input and as an output index buffer
 */
MESHOPTIMIZER_API size_t meshopt_optimizeVertexFetch(void* destination, unsigned int* indices, size_t index_count, const void* vertices, size_t vertex_count, size_t vertex_size);

/**
 * Vertex fetch cache optimizer
 * Generates vertex remap to reduce the amount of GPU memory fetches during vertex processing
 * Returns the number of unique vertices, which is the same as input vertex count unless some vertices are unused
 * The resulting remap table should be used to reorder vertex/index buffers using meshopt_remapVertexBuffer/meshopt_remapIndexBuffer
 *
 * destination must contain enough space for the resulting remap table (vertex_count elements)
 */
MESHOPTIMIZER_API size_t meshopt_optimizeVertexFetchRemap(unsigned int* destination, const unsigned int* indices, size_t index_count, size_t vertex_count);

/**
 * Index buffer encoder
 * Encodes index data into an array of bytes that is generally much smaller (<1.5 bytes/triangle) and compresses better (<1 bytes/triangle) compared to original.
 * Input index buffer must represent a triangle list.
 * Returns encoded data size on success, 0 on error; the only error condition is if buffer doesn't have enough space
 * For maximum efficiency the index buffer being encoded has to be optimized for vertex cache and vertex fetch first.
 *
 * buffer must contain enough space for the encoded index buffer (use meshopt_encodeIndexBufferBound to compute worst case size)
 */
MESHOPTIMIZER_API size_t meshopt_encodeIndexBuffer(unsigned char* buffer, size_t buffer_size, const unsigned int* indices, size_t index_count);
MESHOPTIMIZER_API size_t meshopt_encodeIndexBufferBound(size_t index_count, size_t vertex_count);

/**
 * Set index encoder format version
 * version must specify the data format version to encode; valid values are 0 (decodable by all library versions) and 1 (decodable by 0.14+)
 */
MESHOPTIMIZER_API void meshopt_encodeIndexVersion(int version);

/**
 * Index buffer decoder
 * Decodes index data from an array of bytes generated by meshopt_encodeIndexBuffer
 * Returns 0 if decoding was successful, and an error code otherwise
 * The decoder is safe to use for untrusted input, but it may produce garbage data (e.g. out of range indices).
 *
 * destination must contain enough space for the resulting index buffer (index_count elements)
 */
MESHOPTIMIZER_API int meshopt_decodeIndexBuffer(void* destination, size_t index_count, size_t index_size, const unsigned char* buffer, size_t buffer_size);

/**
 * Index sequence encoder
 * Encodes index sequence into an array of bytes that is generally smaller and compresses better compared to original.
 * Input index sequence can represent arbitrary topology; for triangle lists meshopt_encodeIndexBuffer is likely to be better.
 * Returns encoded data size on success, 0 on error; the only error condition is if buffer doesn't have enough space
 *
 * buffer must contain enough space for the encoded index sequence (use meshopt_encodeIndexSequenceBound to compute worst case size)
 */
MESHOPTIMIZER_API size_t meshopt_encodeIndexSequence(unsigned char* buffer, size_t buffer_size, const unsigned int* indices, size_t index_count);
MESHOPTIMIZER_API size_t meshopt_encodeIndexSequenceBound(size_t index_count, size_t vertex_count);

/**
 * Index sequence decoder
 * Decodes index data from an array of bytes generated by meshopt_encodeIndexSequence
 * Returns 0 if decoding was successful, and an error code otherwise
 * The decoder is safe to use for untrusted input, but it may produce garbage data (e.g. out of range indices).
 *
 * destination must contain enough space for the resulting index sequence (index_count elements)
 */
MESHOPTIMIZER_API int meshopt_decodeIndexSequence(void* destination, size_t index_count, size_t index_size, const unsigned char* buffer, size_t buffer_size);

/**
 * Vertex buffer encoder
 * Encodes vertex data into an array of bytes that is generally smaller and compresses better compared to original.
 * Returns encoded data size on success, 0 on error; the only error condition is if buffer doesn't have enough space
 * This function works for a single vertex stream; for multiple vertex streams, call meshopt_encodeVertexBuffer for each stream.
 * Note that all vertex_size bytes of each vertex are encoded verbatim, including padding which should be zero-initialized.
 * For maximum efficiency the vertex buffer being encoded has to be quantized and optimized for locality of reference (cache/fetch) first.
 *
 * buffer must contain enough space for the encoded vertex buffer (use meshopt_encodeVertexBufferBound to compute worst case size)
 */
MESHOPTIMIZER_API size_t meshopt_encodeVertexBuffer(unsigned char* buffer, size_t buffer_size, const void* vertices, size_t vertex_count, size_t vertex_size);
MESHOPTIMIZER_API size_t meshopt_encodeVertexBufferBound(size_t vertex_count, size_t vertex_size);

/**
 * Set vertex encoder format version
 * version must specify the data format version to encode; valid values are 0 (decodable by all library versions)
 */
MESHOPTIMIZER_API void meshopt_encodeVertexVersion(int version);

/**
 * Vertex buffer decoder
 * Decodes vertex data from an array of bytes generated by meshopt_encodeVertexBuffer
 * Returns 0 if decoding was successful, and an error code otherwise
 * The decoder is safe to use for untrusted input, but it may produce garbage data.
 *
 * destination must contain enough space for the resulting vertex buffer (vertex_count * vertex_size bytes)
 */
MESHOPTIMIZER_API int meshopt_decodeVertexBuffer(void* destination, size_t vertex_count, size_t vertex_size, const unsigned char* buffer, size_t buffer_size);

/**
 * Vertex buffer filters
 * These functions can be used to filter output of meshopt_decodeVertexBuffer in-place.
 *
 * meshopt_decodeFilterOct decodes octahedral encoding of a unit vector with K-bit (K <= 16) signed X/Y as an input; Z must store 1.0f.
 * Each component is stored as an 8-bit or 16-bit normalized integer; stride must be equal to 4 or 8. W is preserved as is.
 *
 * meshopt_decodeFilterQuat decodes 3-component quaternion encoding with K-bit (4 <= K <= 16) component encoding and a 2-bit component index indicating which component to reconstruct.
 * Each component is stored as an 16-bit integer; stride must be equal to 8.
 *
 * meshopt_decodeFilterExp decodes exponential encoding of floating-point data with 8-bit exponent and 24-bit integer mantissa as 2^E*M.
 * Each 32-bit component is decoded in isolation; stride must be divisible by 4.
 */
MESHOPTIMIZER_API void meshopt_decodeFilterOct(void* buffer, size_t count, size_t stride);
MESHOPTIMIZER_API void meshopt_decodeFilterQuat(void* buffer, size_t count, size_t stride);
MESHOPTIMIZER_API void meshopt_decodeFilterExp(void* buffer, size_t count, size_t stride);

/**
 * Vertex buffer filter encoders
 * These functions can be used to encode data in a format that meshopt_decodeFilter can decode
 *
 * meshopt_encodeFilterOct encodes unit vectors with K-bit (K <= 16) signed X/Y as an output.
 * Each component is stored as an 8-bit or 16-bit normalized integer; stride must be equal to 4 or 8. W is preserved as is.
 * Input data must contain 4 floats for every vector (count*4 total).
 *
 * meshopt_encodeFilterQuat encodes unit quaternions with K-bit (4 <= K <= 16) component encoding.
 * Each component is stored as an 16-bit integer; stride must be equal to 8.
 * Input data must contain 4 floats for every quaternion (count*4 total).
 *
 * meshopt_encodeFilterExp encodes arbitrary (finite) floating-point data with 8-bit exponent and K-bit integer mantissa (1 <= K <= 24).
 * Exponent can be shared between all components of a given vector as defined by stride or all values of a given component; stride must be divisible by 4.
 * Input data must contain stride/4 floats for every vector (count*stride/4 total).
 */
enum meshopt_EncodeExpMode
{
	/* When encoding exponents, use separate values for each component (maximum quality) */
	meshopt_EncodeExpSeparate,
	/* When encoding exponents, use shared value for all components of each vector (better compression) */
	meshopt_EncodeExpSharedVector,
	/* When encoding exponents, use shared value for each component of all vectors (best compression) */
	meshopt_EncodeExpSharedComponent,
	/* Experimental: When encoding exponents, use separate values for each component, but clamp to 0 (good quality if very small values are not important) */
	meshopt_EncodeExpClamped,
};

MESHOPTIMIZER_API void meshopt_encodeFilterOct(void* destination, size_t count, size_t stride, int bits, const float* data);
MESHOPTIMIZER_API void meshopt_encodeFilterQuat(void* destination, size_t count, size_t stride, int bits, const float* data);
MESHOPTIMIZER_API void meshopt_encodeFilterExp(void* destination, size_t count, size_t stride, int bits, const float* data, enum meshopt_EncodeExpMode mode);

/**
 * Simplification options
 */
enum
{
	/* Do not move vertices that are located on the topological border (vertices on triangle edges that don't have a paired triangle). Useful for simplifying portions of the larger mesh. */
	meshopt_SimplifyLockBorder = 1 << 0,
	/* Improve simplification performance assuming input indices are a sparse subset of the mesh. Note that error becomes relative to subset extents. */
	meshopt_SimplifySparse = 1 << 1,
	/* Treat error limit and resulting error as absolute instead of relative to mesh extents. */
	meshopt_SimplifyErrorAbsolute = 1 << 2,
	/* Experimental: remove disconnected parts of the mesh during simplification incrementally, regardless of the topological restrictions inside components. */
	meshopt_SimplifyPrune = 1 << 3,
};

/**
 * Mesh simplifier
 * Reduces the number of triangles in the mesh, attempting to preserve mesh appearance as much as possible
 * The algorithm tries to preserve mesh topology and can stop short of the target goal based on topology constraints or target error.
 * If not all attributes from the input mesh are required, it's recommended to reindex the mesh without them prior to simplification.
 * Returns the number of indices after simplification, with destination containing new index data
 * The resulting index buffer references vertices from the original vertex buffer.
 * If the original vertex data isn't required, creating a compact vertex buffer using meshopt_optimizeVertexFetch is recommended.
 *
 * destination must contain enough space for the target index buffer, worst case is index_count elements (*not* target_index_count)!
 * vertex_positions should have float3 position in the first 12 bytes of each vertex
 * target_error represents the error relative to mesh extents that can be tolerated, e.g. 0.01 = 1% deformation; value range [0..1]
 * options must be a bitmask composed of meshopt_SimplifyX options; 0 is a safe default
 * result_error can be NULL; when it's not NULL, it will contain the resulting (relative) error after simplification
 */
MESHOPTIMIZER_API size_t meshopt_simplify(unsigned int* destination, const unsigned int* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, size_t target_index_count, float target_error, unsigned int options, float* result_error);

/**
 * Experimental: Mesh simplifier with attribute metric
 * The algorithm enhances meshopt_simplify by incorporating attribute values into the error metric used to prioritize simplification order; see meshopt_simplify documentation for details.
 * Note that the number of attributes affects memory requirements and running time; this algorithm requires ~1.5x more memory and time compared to meshopt_simplify when using 4 scalar attributes.
 *
 * vertex_attributes should have attribute_count floats for each vertex
 * attribute_weights should have attribute_count floats in total; the weights determine relative priority of attributes between each other and wrt position
 * attribute_count must be <= 32
 * vertex_lock can be NULL; when it's not NULL, it should have a value for each vertex; 1 denotes vertices that can't be moved
 */
MESHOPTIMIZER_EXPERIMENTAL size_t meshopt_simplifyWithAttributes(unsigned int* destination, const unsigned int* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, const float* vertex_attributes, size_t vertex_attributes_stride, const float* attribute_weights, size_t attribute_count, const unsigned char* vertex_lock, size_t target_index_count, float target_error, unsigned int options, float* result_error);

/**
 * Experimental: Mesh simplifier (sloppy)
 * Reduces the number of triangles in the mesh, sacrificing mesh appearance for simplification performance
 * The algorithm doesn't preserve mesh topology but can stop short of the target goal based on target error.
 * Returns the number of indices after simplification, with destination containing new index data
 * The resulting index buffer references vertices from the original vertex buffer.
 * If the original vertex data isn't required, creating a compact vertex buffer using meshopt_optimizeVertexFetch is recommended.
 *
 * destination must contain enough space for the target index buffer, worst case is index_count elements (*not* target_index_count)!
 * vertex_positions should have float3 position in the first 12 bytes of each vertex
 * target_error represents the error relative to mesh extents that can be tolerated, e.g. 0.01 = 1% deformation; value range [0..1]
 * result_error can be NULL; when it's not NULL, it will contain the resulting (relative) error after simplification
 */
MESHOPTIMIZER_EXPERIMENTAL size_t meshopt_simplifySloppy(unsigned int* destination, const unsigned int* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, size_t target_index_count, float target_error, float* result_error);

/**
 * Experimental: Point cloud simplifier
 * Reduces the number of points in the cloud to reach the given target
 * Returns the number of points after simplification, with destination containing new index data
 * The resulting index buffer references vertices from the original vertex buffer.
 * If the original vertex data isn't required, creating a compact vertex buffer using meshopt_optimizeVertexFetch is recommended.
 *
 * destination must contain enough space for the target index buffer (target_vertex_count elements)
 * vertex_positions should have float3 position in the first 12 bytes of each vertex
 * vertex_colors should can be NULL; when it's not NULL, it should have float3 color in the first 12 bytes of each vertex
 * color_weight determines relative priority of color wrt position; 1.0 is a safe default
 */
MESHOPTIMIZER_EXPERIMENTAL size_t meshopt_simplifyPoints(unsigned int* destination, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, const float* vertex_colors, size_t vertex_colors_stride, float color_weight, size_t target_vertex_count);

/**
 * Returns the error scaling factor used by the simplifier to convert between absolute and relative extents
 *
 * Absolute error must be *divided* by the scaling factor before passing it to meshopt_simplify as target_error
 * Relative error returned by meshopt_simplify via result_error must be *multiplied* by the scaling factor to get absolute error.
 */
MESHOPTIMIZER_API float meshopt_simplifyScale(const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride);

/**
 * Mesh stripifier
 * Converts a previously vertex cache optimized triangle list to triangle strip, stitching strips using restart index or degenerate triangles
 * Returns the number of indices in the resulting strip, with destination containing new index data
 * For maximum efficiency the index buffer being converted has to be optimized for vertex cache first.
 * Using restart indices can result in ~10% smaller index buffers, but on some GPUs restart indices may result in decreased performance.
 *
 * destination must contain enough space for the target index buffer, worst case can be computed with meshopt_stripifyBound
 * restart_index should be 0xffff or 0xffffffff depending on index size, or 0 to use degenerate triangles
 */
MESHOPTIMIZER_API size_t meshopt_stripify(unsigned int* destination, const unsigned int* indices, size_t index_count, size_t vertex_count, unsigned int restart_index);
MESHOPTIMIZER_API size_t meshopt_stripifyBound(size_t index_count);

/**
 * Mesh unstripifier
 * Converts a triangle strip to a triangle list
 * Returns the number of indices in the resulting list, with destination containing new index data
 *
 * destination must contain enough space for the target index buffer, worst case can be computed with meshopt_unstripifyBound
 */
MESHOPTIMIZER_API size_t meshopt_unstripify(unsigned int* destination, const unsigned int* indices, size_t index_count, unsigned int restart_index);
MESHOPTIMIZER_API size_t meshopt_unstripifyBound(size_t index_count);

struct meshopt_VertexCacheStatistics
{
	unsigned int vertices_transformed;
	unsigned int warps_executed;
	float acmr; /* transformed vertices / triangle count; best case 0.5, worst case 3.0, optimum depends on topology */
	float atvr; /* transformed vertices / vertex count; best case 1.0, worst case 6.0, optimum is 1.0 (each vertex is transformed once) */
};

/**
 * Vertex transform cache analyzer
 * Returns cache hit statistics using a simplified FIFO model
 * Results may not match actual GPU performance
 */
MESHOPTIMIZER_API struct meshopt_VertexCacheStatistics meshopt_analyzeVertexCache(const unsigned int* indices, size_t index_count, size_t vertex_count, unsigned int cache_size, unsigned int warp_size, unsigned int primgroup_size);

struct meshopt_OverdrawStatistics
{
	unsigned int pixels_covered;
	unsigned int pixels_shaded;
	float overdraw; /* shaded pixels / covered pixels; best case 1.0 */
};

/**
 * Overdraw analyzer
 * Returns overdraw statistics using a software rasterizer
 * Results may not match actual GPU performance
 *
 * vertex_positions should have float3 position in the first 12 bytes of each vertex
 */
MESHOPTIMIZER_API struct meshopt_OverdrawStatistics meshopt_analyzeOverdraw(const unsigned int* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride);

struct meshopt_VertexFetchStatistics
{
	unsigned int bytes_fetched;
	float overfetch; /* fetched bytes / vertex buffer size; best case 1.0 (each byte is fetched once) */
};

/**
 * Vertex fetch cache analyzer
 * Returns cache hit statistics using a simplified direct mapped model
 * Results may not match actual GPU performance
 */
MESHOPTIMIZER_API struct meshopt_VertexFetchStatistics meshopt_analyzeVertexFetch(const unsigned int* indices, size_t index_count, size_t vertex_count, size_t vertex_size);

/**
 * Meshlet is a small mesh cluster (subset) that consists of:
 * - triangles, an 8-bit micro triangle (index) buffer, that for each triangle specifies three local vertices to use;
 * - vertices, a 32-bit vertex indirection buffer, that for each local vertex specifies which mesh vertex to fetch vertex attributes from.
 *
 * For efficiency, meshlet triangles and vertices are packed into two large arrays; this structure contains offsets and counts to access the data.
 */
struct meshopt_Meshlet
{
	/* offsets within meshlet_vertices and meshlet_triangles arrays with meshlet data */
	unsigned int vertex_offset;
	unsigned int triangle_offset;

	/* number of vertices and triangles used in the meshlet; data is stored in consecutive range defined by offset and count */
	unsigned int vertex_count;
	unsigned int triangle_count;
};

/**
 * Meshlet builder
 * Splits the mesh into a set of meshlets where each meshlet has a micro index buffer indexing into meshlet vertices that refer to the original vertex buffer
 * The resulting data can be used to render meshes using NVidia programmable mesh shading pipeline, or in other cluster-based renderers.
 * When targeting mesh shading hardware, for maximum efficiency meshlets should be further optimized using meshopt_optimizeMeshlet.
 * When using buildMeshlets, vertex positions need to be provided to minimize the size of the resulting clusters.
 * When using buildMeshletsScan, for maximum efficiency the index buffer being converted has to be optimized for vertex cache first.
 *
 * meshlets must contain enough space for all meshlets, worst case size can be computed with meshopt_buildMeshletsBound
 * meshlet_vertices must contain enough space for all meshlets, worst case size is equal to max_meshlets * max_vertices
 * meshlet_triangles must contain enough space for all meshlets, worst case size is equal to max_meshlets * max_triangles * 3
 * vertex_positions should have float3 position in the first 12 bytes of each vertex
 * max_vertices and max_triangles must not exceed implementation limits (max_vertices <= 255 - not 256!, max_triangles <= 512; max_triangles must be divisible by 4)
 * cone_weight should be set to 0 when cone culling is not used, and a value between 0 and 1 otherwise to balance between cluster size and cone culling efficiency
 */
MESHOPTIMIZER_API size_t meshopt_buildMeshlets(struct meshopt_Meshlet* meshlets, unsigned int* meshlet_vertices, unsigned char* meshlet_triangles, const unsigned int* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, size_t max_vertices, size_t max_triangles, float cone_weight);
MESHOPTIMIZER_API size_t meshopt_buildMeshletsScan(struct meshopt_Meshlet* meshlets, unsigned int* meshlet_vertices, unsigned char* meshlet_triangles, const unsigned int* indices, size_t index_count, size_t vertex_count, size_t max_vertices, size_t max_triangles);
MESHOPTIMIZER_API size_t meshopt_buildMeshletsBound(size_t index_count, size_t max_vertices, size_t max_triangles);

/**
 * Experimental: Meshlet optimizer
 * Reorders meshlet vertices and triangles to maximize locality to improve rasterizer throughput
 *
 * meshlet_triangles and meshlet_vertices must refer to meshlet triangle and vertex index data; when buildMeshlets* is used, these
 * need to be computed from meshlet's vertex_offset and triangle_offset
 * triangle_count and vertex_count must not exceed implementation limits (vertex_count <= 255 - not 256!, triangle_count <= 512)
 */
MESHOPTIMIZER_EXPERIMENTAL void meshopt_optimizeMeshlet(unsigned int* meshlet_vertices, unsigned char* meshlet_triangles, size_t triangle_count, size_t vertex_count);

struct meshopt_Bounds
{
	/* bounding sphere, useful for frustum and occlusion culling */
	float center[3];
	float radius;

	/* normal cone, useful for backface culling */
	float cone_apex[3];
	float cone_axis[3];
	float cone_cutoff; /* = cos(angle/2) */

	/* normal cone axis and cutoff, stored in 8-bit SNORM format; decode using x/127.0 */
	signed char cone_axis_s8[3];
	signed char cone_cutoff_s8;
};

/**
 * Cluster bounds generator
 * Creates bounding volumes that can be used for frustum, backface and occlusion culling.
 *
 * For backface culling with orthographic projection, use the following formula to reject backfacing clusters:
 *   dot(view, cone_axis) >= cone_cutoff
 *
 * For perspective projection, you can use the formula that needs cone apex in addition to axis & cutoff:
 *   dot(normalize(cone_apex - camera_position), cone_axis) >= cone_cutoff
 *
 * Alternatively, you can use the formula that doesn't need cone apex and uses bounding sphere instead:
 *   dot(normalize(center - camera_position), cone_axis) >= cone_cutoff + radius / length(center - camera_position)
 * or an equivalent formula that doesn't have a singularity at center = camera_position:
 *   dot(center - camera_position, cone_axis) >= cone_cutoff * length(center - camera_position) + radius
 *
 * The formula that uses the apex is slightly more accurate but needs the apex; if you are already using bounding sphere
 * to do frustum/occlusion culling, the formula that doesn't use the apex may be preferable (for derivation see
 * Real-Time Rendering 4th Edition, section 19.3).
 *
 * vertex_positions should have float3 position in the first 12 bytes of each vertex
 * vertex_count should specify the number of vertices in the entire mesh, not cluster or meshlet
 * index_count/3 and triangle_count must not exceed implementation limits (<= 512)
 */
MESHOPTIMIZER_API struct meshopt_Bounds meshopt_computeClusterBounds(const unsigned int* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride);
MESHOPTIMIZER_API struct meshopt_Bounds meshopt_computeMeshletBounds(const unsigned int* meshlet_vertices, const unsigned char* meshlet_triangles, size_t triangle_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride);

/**
 * Spatial sorter
 * Generates a remap table that can be used to reorder points for spatial locality.
 * Resulting remap table maps old vertices to new vertices and can be used in meshopt_remapVertexBuffer.
 *
 * destination must contain enough space for the resulting remap table (vertex_count elements)
 * vertex_positions should have float3 position in the first 12 bytes of each vertex
 */
MESHOPTIMIZER_API void meshopt_spatialSortRemap(unsigned int* destination, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride);

/**
 * Experimental: Spatial sorter
 * Reorders triangles for spatial locality, and generates a new index buffer. The resulting index buffer can be used with other functions like optimizeVertexCache.
 *
 * destination must contain enough space for the resulting index buffer (index_count elements)
 * vertex_positions should have float3 position in the first 12 bytes of each vertex
 */
MESHOPTIMIZER_EXPERIMENTAL void meshopt_spatialSortTriangles(unsigned int* destination, const unsigned int* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride);

/**
 * Set allocation callbacks
 * These callbacks will be used instead of the default operator new/operator delete for all temporary allocations in the library.
 * Note that all algorithms only allocate memory for temporary use.
 * allocate/deallocate are always called in a stack-like order - last pointer to be allocated is deallocated first.
 */
MESHOPTIMIZER_API void meshopt_setAllocator(void* (MESHOPTIMIZER_ALLOC_CALLCONV* allocate)(size_t), void (MESHOPTIMIZER_ALLOC_CALLCONV* deallocate)(void*));

#ifdef __cplusplus
} /* extern "C" */
#endif

/* Quantization into commonly supported data formats */
#ifdef __cplusplus
/**
 * Quantize a float in [0..1] range into an N-bit fixed point unorm value
 * Assumes reconstruction function (q / (2^N-1)), which is the case for fixed-function normalized fixed point conversion
 * Maximum reconstruction error: 1/2^(N+1)
 */
inline int meshopt_quantizeUnorm(float v, int N);

/**
 * Quantize a float in [-1..1] range into an N-bit fixed point snorm value
 * Assumes reconstruction function (q / (2^(N-1)-1)), which is the case for fixed-function normalized fixed point conversion (except early OpenGL versions)
 * Maximum reconstruction error: 1/2^N
 */
inline int meshopt_quantizeSnorm(float v, int N);

/**
 * Quantize a float into half-precision (as defined by IEEE-754 fp16) floating point value
 * Generates +-inf for overflow, preserves NaN, flushes denormals to zero, rounds to nearest
 * Representable magnitude range: [6e-5; 65504]
 * Maximum relative reconstruction error: 5e-4
 */
MESHOPTIMIZER_API unsigned short meshopt_quantizeHalf(float v);

/**
 * Quantize a float into a floating point value with a limited number of significant mantissa bits, preserving the IEEE-754 fp32 binary representation
 * Generates +-inf for overflow, preserves NaN, flushes denormals to zero, rounds to nearest
 * Assumes N is in a valid mantissa precision range, which is 1..23
 */
MESHOPTIMIZER_API float meshopt_quantizeFloat(float v, int N);

/**
 * Reverse quantization of a half-precision (as defined by IEEE-754 fp16) floating point value
 * Preserves Inf/NaN, flushes denormals to zero
 */
MESHOPTIMIZER_API float meshopt_dequantizeHalf(unsigned short h);
#endif

/**
 * C++ template interface
 *
 * These functions mirror the C interface the library provides, providing template-based overloads so that
 * the caller can use an arbitrary type for the index data, both for input and output.
 * When the supplied type is the same size as that of unsigned int, the wrappers are zero-cost; when it's not,
 * the wrappers end up allocating memory and copying index data to convert from one type to another.
 */
#if defined(__cplusplus) && !defined(MESHOPTIMIZER_NO_WRAPPERS)
template <typename T>
inline size_t meshopt_generateVertexRemap(unsigned int* destination, const T* indices, size_t index_count, const void* vertices, size_t vertex_count, size_t vertex_size);
template <typename T>
inline size_t meshopt_generateVertexRemapMulti(unsigned int* destination, const T* indices, size_t index_count, size_t vertex_count, const meshopt_Stream* streams, size_t stream_count);
template <typename T>
inline void meshopt_remapIndexBuffer(T* destination, const T* indices, size_t index_count, const unsigned int* remap);
template <typename T>
inline void meshopt_generateShadowIndexBuffer(T* destination, const T* indices, size_t index_count, const void* vertices, size_t vertex_count, size_t vertex_size, size_t vertex_stride);
template <typename T>
inline void meshopt_generateShadowIndexBufferMulti(T* destination, const T* indices, size_t index_count, size_t vertex_count, const meshopt_Stream* streams, size_t stream_count);
template <typename T>
inline void meshopt_generateAdjacencyIndexBuffer(T* destination, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride);
template <typename T>
inline void meshopt_generateTessellationIndexBuffer(T* destination, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride);
template <typename T>
inline size_t meshopt_generateProvokingIndexBuffer(T* destination, unsigned int* reorder, const T* indices, size_t index_count, size_t vertex_count);
template <typename T>
inline void meshopt_optimizeVertexCache(T* destination, const T* indices, size_t index_count, size_t vertex_count);
template <typename T>
inline void meshopt_optimizeVertexCacheStrip(T* destination, const T* indices, size_t index_count, size_t vertex_count);
template <typename T>
inline void meshopt_optimizeVertexCacheFifo(T* destination, const T* indices, size_t index_count, size_t vertex_count, unsigned int cache_size);
template <typename T>
inline void meshopt_optimizeOverdraw(T* destination, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, float threshold);
template <typename T>
inline size_t meshopt_optimizeVertexFetchRemap(unsigned int* destination, const T* indices, size_t index_count, size_t vertex_count);
template <typename T>
inline size_t meshopt_optimizeVertexFetch(void* destination, T* indices, size_t index_count, const void* vertices, size_t vertex_count, size_t vertex_size);
template <typename T>
inline size_t meshopt_encodeIndexBuffer(unsigned char* buffer, size_t buffer_size, const T* indices, size_t index_count);
template <typename T>
inline int meshopt_decodeIndexBuffer(T* destination, size_t index_count, const unsigned char* buffer, size_t buffer_size);
template <typename T>
inline size_t meshopt_encodeIndexSequence(unsigned char* buffer, size_t buffer_size, const T* indices, size_t index_count);
template <typename T>
inline int meshopt_decodeIndexSequence(T* destination, size_t index_count, const unsigned char* buffer, size_t buffer_size);
template <typename T>
inline size_t meshopt_simplify(T* destination, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, size_t target_index_count, float target_error, unsigned int options = 0, float* result_error = NULL);
template <typename T>
inline size_t meshopt_simplifyWithAttributes(T* destination, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, const float* vertex_attributes, size_t vertex_attributes_stride, const float* attribute_weights, size_t attribute_count, const unsigned char* vertex_lock, size_t target_index_count, float target_error, unsigned int options = 0, float* result_error = NULL);
template <typename T>
inline size_t meshopt_simplifySloppy(T* destination, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, size_t target_index_count, float target_error, float* result_error = NULL);
template <typename T>
inline size_t meshopt_stripify(T* destination, const T* indices, size_t index_count, size_t vertex_count, T restart_index);
template <typename T>
inline size_t meshopt_unstripify(T* destination, const T* indices, size_t index_count, T restart_index);
template <typename T>
inline meshopt_VertexCacheStatistics meshopt_analyzeVertexCache(const T* indices, size_t index_count, size_t vertex_count, unsigned int cache_size, unsigned int warp_size, unsigned int buffer_size);
template <typename T>
inline meshopt_OverdrawStatistics meshopt_analyzeOverdraw(const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride);
template <typename T>
inline meshopt_VertexFetchStatistics meshopt_analyzeVertexFetch(const T* indices, size_t index_count, size_t vertex_count, size_t vertex_size);
template <typename T>
inline size_t meshopt_buildMeshlets(meshopt_Meshlet* meshlets, unsigned int* meshlet_vertices, unsigned char* meshlet_triangles, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, size_t max_vertices, size_t max_triangles, float cone_weight);
template <typename T>
inline size_t meshopt_buildMeshletsScan(meshopt_Meshlet* meshlets, unsigned int* meshlet_vertices, unsigned char* meshlet_triangles, const T* indices, size_t index_count, size_t vertex_count, size_t max_vertices, size_t max_triangles);
template <typename T>
inline meshopt_Bounds meshopt_computeClusterBounds(const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride);
template <typename T>
inline void meshopt_spatialSortTriangles(T* destination, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride);
#endif

/* Inline implementation */
#ifdef __cplusplus
inline int meshopt_quantizeUnorm(float v, int N)
{
	const float scale = float((1 << N) - 1);

	v = (v >= 0) ? v : 0;
	v = (v <= 1) ? v : 1;

	return int(v * scale + 0.5f);
}

inline int meshopt_quantizeSnorm(float v, int N)
{
	const float scale = float((1 << (N - 1)) - 1);

	float round = (v >= 0 ? 0.5f : -0.5f);

	v = (v >= -1) ? v : -1;
	v = (v <= +1) ? v : +1;

	return int(v * scale + round);
}
#endif

/* Internal implementation helpers */
#ifdef __cplusplus
class meshopt_Allocator
{
public:
	template <typename T>
	struct StorageT
	{
		static void* (MESHOPTIMIZER_ALLOC_CALLCONV* allocate)(size_t);
		static void (MESHOPTIMIZER_ALLOC_CALLCONV* deallocate)(void*);
	};

	typedef StorageT<void> Storage;

	meshopt_Allocator()
	    : blocks()
	    , count(0)
	{
	}

	~meshopt_Allocator()
	{
		for (size_t i = count; i > 0; --i)
			Storage::deallocate(blocks[i - 1]);
	}

	template <typename T>
	T* allocate(size_t size)
	{
		assert(count < sizeof(blocks) / sizeof(blocks[0]));
		T* result = static_cast<T*>(Storage::allocate(size > size_t(-1) / sizeof(T) ? size_t(-1) : size * sizeof(T)));
		blocks[count++] = result;
		return result;
	}

	void deallocate(void* ptr)
	{
		assert(count > 0 && blocks[count - 1] == ptr);
		Storage::deallocate(ptr);
		count--;
	}

private:
	void* blocks[24];
	size_t count;
};

// This makes sure that allocate/deallocate are lazily generated in translation units that need them and are deduplicated by the linker
template <typename T>
void* (MESHOPTIMIZER_ALLOC_CALLCONV* meshopt_Allocator::StorageT<T>::allocate)(size_t) = operator new;
template <typename T>
void (MESHOPTIMIZER_ALLOC_CALLCONV* meshopt_Allocator::StorageT<T>::deallocate)(void*) = operator delete;
#endif

/* Inline implementation for C++ templated wrappers */
#if defined(__cplusplus) && !defined(MESHOPTIMIZER_NO_WRAPPERS)
template <typename T, bool ZeroCopy = sizeof(T) == sizeof(unsigned int)>
struct meshopt_IndexAdapter;

template <typename T>
struct meshopt_IndexAdapter<T, false>
{
	T* result;
	unsigned int* data;
	size_t count;

	meshopt_IndexAdapter(T* result_, const T* input, size_t count_)
	    : result(result_)
	    , data(NULL)
	    , count(count_)
	{
		size_t size = count > size_t(-1) / sizeof(unsigned int) ? size_t(-1) : count * sizeof(unsigned int);

		data = static_cast<unsigned int*>(meshopt_Allocator::Storage::allocate(size));

		if (input)
		{
			for (size_t i = 0; i < count; ++i)
				data[i] = input[i];
		}
	}

	~meshopt_IndexAdapter()
	{
		if (result)
		{
			for (size_t i = 0; i < count; ++i)
				result[i] = T(data[i]);
		}

		meshopt_Allocator::Storage::deallocate(data);
	}
};

template <typename T>
struct meshopt_IndexAdapter<T, true>
{
	unsigned int* data;

	meshopt_IndexAdapter(T* result, const T* input, size_t)
	    : data(reinterpret_cast<unsigned int*>(result ? result : const_cast<T*>(input)))
	{
	}
};

template <typename T>
inline size_t meshopt_generateVertexRemap(unsigned int* destination, const T* indices, size_t index_count, const void* vertices, size_t vertex_count, size_t vertex_size)
{
	meshopt_IndexAdapter<T> in(NULL, indices, indices ? index_count : 0);

	return meshopt_generateVertexRemap(destination, indices ? in.data : NULL, index_count, vertices, vertex_count, vertex_size);
}

template <typename T>
inline size_t meshopt_generateVertexRemapMulti(unsigned int* destination, const T* indices, size_t index_count, size_t vertex_count, const meshopt_Stream* streams, size_t stream_count)
{
	meshopt_IndexAdapter<T> in(NULL, indices, indices ? index_count : 0);

	return meshopt_generateVertexRemapMulti(destination, indices ? in.data : NULL, index_count, vertex_count, streams, stream_count);
}

template <typename T>
inline void meshopt_remapIndexBuffer(T* destination, const T* indices, size_t index_count, const unsigned int* remap)
{
	meshopt_IndexAdapter<T> in(NULL, indices, indices ? index_count : 0);
	meshopt_IndexAdapter<T> out(destination, 0, index_count);

	meshopt_remapIndexBuffer(out.data, indices ? in.data : NULL, index_count, remap);
}

template <typename T>
inline void meshopt_generateShadowIndexBuffer(T* destination, const T* indices, size_t index_count, const void* vertices, size_t vertex_count, size_t vertex_size, size_t vertex_stride)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, index_count);

	meshopt_generateShadowIndexBuffer(out.data, in.data, index_count, vertices, vertex_count, vertex_size, vertex_stride);
}

template <typename T>
inline void meshopt_generateShadowIndexBufferMulti(T* destination, const T* indices, size_t index_count, size_t vertex_count, const meshopt_Stream* streams, size_t stream_count)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, index_count);

	meshopt_generateShadowIndexBufferMulti(out.data, in.data, index_count, vertex_count, streams, stream_count);
}

template <typename T>
inline void meshopt_generateAdjacencyIndexBuffer(T* destination, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, index_count * 2);

	meshopt_generateAdjacencyIndexBuffer(out.data, in.data, index_count, vertex_positions, vertex_count, vertex_positions_stride);
}

template <typename T>
inline void meshopt_generateTessellationIndexBuffer(T* destination, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, index_count * 4);

	meshopt_generateTessellationIndexBuffer(out.data, in.data, index_count, vertex_positions, vertex_count, vertex_positions_stride);
}

template <typename T>
inline size_t meshopt_generateProvokingIndexBuffer(T* destination, unsigned int* reorder, const T* indices, size_t index_count, size_t vertex_count)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, index_count);

	size_t bound = vertex_count + (index_count / 3);
	assert(size_t(T(bound - 1)) == bound - 1); // bound - 1 must fit in T
	(void)bound;

	return meshopt_generateProvokingIndexBuffer(out.data, reorder, in.data, index_count, vertex_count);
}

template <typename T>
inline void meshopt_optimizeVertexCache(T* destination, const T* indices, size_t index_count, size_t vertex_count)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, index_count);

	meshopt_optimizeVertexCache(out.data, in.data, index_count, vertex_count);
}

template <typename T>
inline void meshopt_optimizeVertexCacheStrip(T* destination, const T* indices, size_t index_count, size_t vertex_count)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, index_count);

	meshopt_optimizeVertexCacheStrip(out.data, in.data, index_count, vertex_count);
}

template <typename T>
inline void meshopt_optimizeVertexCacheFifo(T* destination, const T* indices, size_t index_count, size_t vertex_count, unsigned int cache_size)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, index_count);

	meshopt_optimizeVertexCacheFifo(out.data, in.data, index_count, vertex_count, cache_size);
}

template <typename T>
inline void meshopt_optimizeOverdraw(T* destination, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, float threshold)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, index_count);

	meshopt_optimizeOverdraw(out.data, in.data, index_count, vertex_positions, vertex_count, vertex_positions_stride, threshold);
}

template <typename T>
inline size_t meshopt_optimizeVertexFetchRemap(unsigned int* destination, const T* indices, size_t index_count, size_t vertex_count)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);

	return meshopt_optimizeVertexFetchRemap(destination, in.data, index_count, vertex_count);
}

template <typename T>
inline size_t meshopt_optimizeVertexFetch(void* destination, T* indices, size_t index_count, const void* vertices, size_t vertex_count, size_t vertex_size)
{
	meshopt_IndexAdapter<T> inout(indices, indices, index_count);

	return meshopt_optimizeVertexFetch(destination, inout.data, index_count, vertices, vertex_count, vertex_size);
}

template <typename T>
inline size_t meshopt_encodeIndexBuffer(unsigned char* buffer, size_t buffer_size, const T* indices, size_t index_count)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);

	return meshopt_encodeIndexBuffer(buffer, buffer_size, in.data, index_count);
}

template <typename T>
inline int meshopt_decodeIndexBuffer(T* destination, size_t index_count, const unsigned char* buffer, size_t buffer_size)
{
	char index_size_valid[sizeof(T) == 2 || sizeof(T) == 4 ? 1 : -1];
	(void)index_size_valid;

	return meshopt_decodeIndexBuffer(destination, index_count, sizeof(T), buffer, buffer_size);
}

template <typename T>
inline size_t meshopt_encodeIndexSequence(unsigned char* buffer, size_t buffer_size, const T* indices, size_t index_count)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);

	return meshopt_encodeIndexSequence(buffer, buffer_size, in.data, index_count);
}

template <typename T>
inline int meshopt_decodeIndexSequence(T* destination, size_t index_count, const unsigned char* buffer, size_t buffer_size)
{
	char index_size_valid[sizeof(T) == 2 || sizeof(T) == 4 ? 1 : -1];
	(void)index_size_valid;

	return meshopt_decodeIndexSequence(destination, index_count, sizeof(T), buffer, buffer_size);
}

template <typename T>
inline size_t meshopt_simplify(T* destination, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, size_t target_index_count, float target_error, unsigned int options, float* result_error)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, index_count);

	return meshopt_simplify(out.data, in.data, index_count, vertex_positions, vertex_count, vertex_positions_stride, target_index_count, target_error, options, result_error);
}

template <typename T>
inline size_t meshopt_simplifyWithAttributes(T* destination, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, const float* vertex_attributes, size_t vertex_attributes_stride, const float* attribute_weights, size_t attribute_count, const unsigned char* vertex_lock, size_t target_index_count, float target_error, unsigned int options, float* result_error)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, index_count);

	return meshopt_simplifyWithAttributes(out.data, in.data, index_count, vertex_positions, vertex_count, vertex_positions_stride, vertex_attributes, vertex_attributes_stride, attribute_weights, attribute_count, vertex_lock, target_index_count, target_error, options, result_error);
}

template <typename T>
inline size_t meshopt_simplifySloppy(T* destination, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, size_t target_index_count, float target_error, float* result_error)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, index_count);

	return meshopt_simplifySloppy(out.data, in.data, index_count, vertex_positions, vertex_count, vertex_positions_stride, target_index_count, target_error, result_error);
}

template <typename T>
inline size_t meshopt_stripify(T* destination, const T* indices, size_t index_count, size_t vertex_count, T restart_index)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, (index_count / 3) * 5);

	return meshopt_stripify(out.data, in.data, index_count, vertex_count, unsigned(restart_index));
}

template <typename T>
inline size_t meshopt_unstripify(T* destination, const T* indices, size_t index_count, T restart_index)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, (index_count - 2) * 3);

	return meshopt_unstripify(out.data, in.data, index_count, unsigned(restart_index));
}

template <typename T>
inline meshopt_VertexCacheStatistics meshopt_analyzeVertexCache(const T* indices, size_t index_count, size_t vertex_count, unsigned int cache_size, unsigned int warp_size, unsigned int buffer_size)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);

	return meshopt_analyzeVertexCache(in.data, index_count, vertex_count, cache_size, warp_size, buffer_size);
}

template <typename T>
inline meshopt_OverdrawStatistics meshopt_analyzeOverdraw(const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);

	return meshopt_analyzeOverdraw(in.data, index_count, vertex_positions, vertex_count, vertex_positions_stride);
}

template <typename T>
inline meshopt_VertexFetchStatistics meshopt_analyzeVertexFetch(const T* indices, size_t index_count, size_t vertex_count, size_t vertex_size)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);

	return meshopt_analyzeVertexFetch(in.data, index_count, vertex_count, vertex_size);
}

template <typename T>
inline size_t meshopt_buildMeshlets(meshopt_Meshlet* meshlets, unsigned int* meshlet_vertices, unsigned char* meshlet_triangles, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride, size_t max_vertices, size_t max_triangles, float cone_weight)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);

	return meshopt_buildMeshlets(meshlets, meshlet_vertices, meshlet_triangles, in.data, index_count, vertex_positions, vertex_count, vertex_positions_stride, max_vertices, max_triangles, cone_weight);
}

template <typename T>
inline size_t meshopt_buildMeshletsScan(meshopt_Meshlet* meshlets, unsigned int* meshlet_vertices, unsigned char* meshlet_triangles, const T* indices, size_t index_count, size_t vertex_count, size_t max_vertices, size_t max_triangles)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);

	return meshopt_buildMeshletsScan(meshlets, meshlet_vertices, meshlet_triangles, in.data, index_count, vertex_count, max_vertices, max_triangles);
}

template <typename T>
inline meshopt_Bounds meshopt_computeClusterBounds(const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);

	return meshopt_computeClusterBounds(in.data, index_count, vertex_positions, vertex_count, vertex_positions_stride);
}

template <typename T>
inline void meshopt_spatialSortTriangles(T* destination, const T* indices, size_t index_count, const float* vertex_positions, size_t vertex_count, size_t vertex_positions_stride)
{
	meshopt_IndexAdapter<T> in(NULL, indices, index_count);
	meshopt_IndexAdapter<T> out(destination, NULL, index_count);

	meshopt_spatialSortTriangles(out.data, in.data, index_count, vertex_positions, vertex_count, vertex_positions_stride);
}
#endif

/**
 * Copyright (c) 2016-2024 Arseny Kapoulkine
 *
 * 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.
 */