Files
test/source/blender/blenkernel/intern/spline_base.cc

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

544 lines
16 KiB
C++
Raw Normal View History

/*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software Foundation,
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*/
#include "BLI_array.hh"
#include "BLI_span.hh"
#include "BLI_task.hh"
#include "BLI_timeit.hh"
#include "BKE_attribute_access.hh"
#include "BKE_attribute_math.hh"
#include "BKE_spline.hh"
#include "FN_generic_virtual_array.hh"
using blender::Array;
using blender::float3;
using blender::IndexRange;
using blender::MutableSpan;
using blender::Span;
Geometry Nodes: refactor virtual array system Goals of this refactor: * Simplify creating virtual arrays. * Simplify passing virtual arrays around. * Simplify converting between typed and generic virtual arrays. * Reduce memory allocations. As a quick reminder, a virtual arrays is a data structure that behaves like an array (i.e. it can be accessed using an index). However, it may not actually be stored as array internally. The two most important implementations of virtual arrays are those that correspond to an actual plain array and those that have the same value for every index. However, many more implementations exist for various reasons (interfacing with legacy attributes, unified iterator over all points in multiple splines, ...). With this refactor the core types (`VArray`, `GVArray`, `VMutableArray` and `GVMutableArray`) can be used like "normal values". They typically live on the stack. Before, they were usually inside a `std::unique_ptr`. This makes passing them around much easier. Creation of new virtual arrays is also much simpler now due to some constructors. Memory allocations are reduced by making use of small object optimization inside the core types. Previously, `VArray` was a class with virtual methods that had to be overridden to change the behavior of a the virtual array. Now,`VArray` has a fixed size and has no virtual methods. Instead it contains a `VArrayImpl` that is similar to the old `VArray`. `VArrayImpl` should rarely ever be used directly, unless a new virtual array implementation is added. To support the small object optimization for many `VArrayImpl` classes, a new `blender::Any` type is added. It is similar to `std::any` with two additional features. It has an adjustable inline buffer size and alignment. The inline buffer size of `std::any` can't be relied on and is usually too small for our use case here. Furthermore, `blender::Any` can store additional user-defined type information without increasing the stack size. Differential Revision: https://developer.blender.org/D12986
2021-11-16 10:15:51 +01:00
using blender::VArray;
using blender::attribute_math::convert_to_static_type;
using blender::bke::AttributeIDRef;
using blender::fn::GMutableSpan;
using blender::fn::GSpan;
using blender::fn::GVArray;
Spline::Type Spline::type() const
{
return type_;
}
void Spline::copy_base_settings(const Spline &src, Spline &dst)
{
dst.normal_mode = src.normal_mode;
dst.is_cyclic_ = src.is_cyclic_;
}
static SplinePtr create_spline(const Spline::Type type)
{
switch (type) {
case Spline::Type::Poly:
return std::make_unique<PolySpline>();
case Spline::Type::Bezier:
return std::make_unique<BezierSpline>();
case Spline::Type::NURBS:
return std::make_unique<NURBSpline>();
}
BLI_assert_unreachable();
return {};
}
SplinePtr Spline::copy() const
{
SplinePtr dst = this->copy_without_attributes();
dst->attributes = this->attributes;
return dst;
}
SplinePtr Spline::copy_only_settings() const
{
SplinePtr dst = create_spline(type_);
this->copy_base_settings(*this, *dst);
this->copy_settings(*dst);
return dst;
}
SplinePtr Spline::copy_without_attributes() const
{
SplinePtr dst = this->copy_only_settings();
this->copy_data(*dst);
/* Though the attributes storage is empty, it still needs to know the correct size. */
dst->attributes.reallocate(dst->size());
return dst;
}
void Spline::translate(const blender::float3 &translation)
{
for (float3 &position : this->positions()) {
position += translation;
}
this->mark_cache_invalid();
}
void Spline::transform(const blender::float4x4 &matrix)
{
for (float3 &position : this->positions()) {
position = matrix * position;
}
this->mark_cache_invalid();
}
void Spline::reverse()
{
this->positions().reverse();
this->radii().reverse();
this->tilts().reverse();
this->attributes.foreach_attribute(
[&](const AttributeIDRef &id, const AttributeMetaData &meta_data) {
std::optional<blender::fn::GMutableSpan> attribute = this->attributes.get_for_write(id);
if (!attribute) {
BLI_assert_unreachable();
return false;
}
convert_to_static_type(meta_data.data_type, [&](auto dummy) {
using T = decltype(dummy);
attribute->typed<T>().reverse();
});
return true;
},
ATTR_DOMAIN_POINT);
this->reverse_impl();
this->mark_cache_invalid();
}
int Spline::evaluated_edges_size() const
{
const int eval_size = this->evaluated_points_size();
if (eval_size < 2) {
/* Two points are required for an edge. */
return 0;
}
return this->is_cyclic_ ? eval_size : eval_size - 1;
}
float Spline::length() const
{
Span<float> lengths = this->evaluated_lengths();
2021-08-24 10:01:08 -05:00
return lengths.is_empty() ? 0.0f : this->evaluated_lengths().last();
}
int Spline::segments_size() const
{
const int size = this->size();
return is_cyclic_ ? size : size - 1;
}
bool Spline::is_cyclic() const
{
return is_cyclic_;
}
void Spline::set_cyclic(const bool value)
{
is_cyclic_ = value;
}
static void accumulate_lengths(Span<float3> positions,
const bool is_cyclic,
MutableSpan<float> lengths)
{
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
using namespace blender::math;
float length = 0.0f;
for (const int i : IndexRange(positions.size() - 1)) {
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
length += distance(positions[i], positions[i + 1]);
lengths[i] = length;
}
if (is_cyclic) {
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
lengths.last() = length + distance(positions.last(), positions.first());
}
}
Span<float> Spline::evaluated_lengths() const
{
if (!length_cache_dirty_) {
return evaluated_lengths_cache_;
}
std::lock_guard lock{length_cache_mutex_};
if (!length_cache_dirty_) {
return evaluated_lengths_cache_;
}
const int total = evaluated_edges_size();
evaluated_lengths_cache_.resize(total);
if (total != 0) {
Span<float3> positions = this->evaluated_positions();
accumulate_lengths(positions, is_cyclic_, evaluated_lengths_cache_);
}
length_cache_dirty_ = false;
return evaluated_lengths_cache_;
}
static float3 direction_bisect(const float3 &prev, const float3 &middle, const float3 &next)
{
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
using namespace blender::math;
const float3 dir_prev = normalize(middle - prev);
const float3 dir_next = normalize(next - middle);
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
const float3 result = normalize(dir_prev + dir_next);
if (UNLIKELY(is_zero(result))) {
return float3(0.0f, 0.0f, 1.0f);
}
return result;
}
static void calculate_tangents(Span<float3> positions,
const bool is_cyclic,
MutableSpan<float3> tangents)
{
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
using namespace blender::math;
if (positions.size() == 1) {
tangents.first() = float3(0.0f, 0.0f, 1.0f);
return;
}
for (const int i : IndexRange(1, positions.size() - 2)) {
tangents[i] = direction_bisect(positions[i - 1], positions[i], positions[i + 1]);
}
if (is_cyclic) {
const float3 &second_to_last = positions[positions.size() - 2];
const float3 &last = positions.last();
const float3 &first = positions.first();
const float3 &second = positions[1];
tangents.first() = direction_bisect(last, first, second);
tangents.last() = direction_bisect(second_to_last, last, first);
}
else {
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
tangents.first() = normalize(positions[1] - positions[0]);
tangents.last() = normalize(positions.last() - positions[positions.size() - 2]);
}
}
Span<float3> Spline::evaluated_tangents() const
{
if (!tangent_cache_dirty_) {
return evaluated_tangents_cache_;
}
std::lock_guard lock{tangent_cache_mutex_};
if (!tangent_cache_dirty_) {
return evaluated_tangents_cache_;
}
const int eval_size = this->evaluated_points_size();
evaluated_tangents_cache_.resize(eval_size);
Span<float3> positions = this->evaluated_positions();
calculate_tangents(positions, is_cyclic_, evaluated_tangents_cache_);
this->correct_end_tangents();
tangent_cache_dirty_ = false;
return evaluated_tangents_cache_;
}
static float3 rotate_direction_around_axis(const float3 &direction,
const float3 &axis,
const float angle)
{
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
using namespace blender::math;
BLI_ASSERT_UNIT_V3(direction);
BLI_ASSERT_UNIT_V3(axis);
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
const float3 axis_scaled = axis * dot(direction, axis);
const float3 diff = direction - axis_scaled;
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
const float3 cross = blender::math::cross(axis, diff);
return axis_scaled + diff * std::cos(angle) + cross * std::sin(angle);
}
static void calculate_normals_z_up(Span<float3> tangents, MutableSpan<float3> r_normals)
{
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
using namespace blender::math;
BLI_assert(r_normals.size() == tangents.size());
/* Same as in `vec_to_quat`. */
const float epsilon = 1e-4f;
for (const int i : r_normals.index_range()) {
const float3 &tangent = tangents[i];
if (fabsf(tangent.x) + fabsf(tangent.y) < epsilon) {
r_normals[i] = {1.0f, 0.0f, 0.0f};
}
else {
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
r_normals[i] = normalize(float3(tangent.y, -tangent.x, 0.0f));
}
}
}
/**
* Rotate the last normal in the same way the tangent has been rotated.
*/
static float3 calculate_next_normal(const float3 &last_normal,
const float3 &last_tangent,
const float3 &current_tangent)
{
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
using namespace blender::math;
if (is_zero(last_tangent) || is_zero(current_tangent)) {
return last_normal;
}
const float angle = angle_normalized_v3v3(last_tangent, current_tangent);
if (angle != 0.0) {
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
const float3 axis = normalize(cross(last_tangent, current_tangent));
return rotate_direction_around_axis(last_normal, axis, angle);
}
2021-06-18 14:27:43 +10:00
return last_normal;
}
static void calculate_normals_minimum(Span<float3> tangents,
const bool cyclic,
MutableSpan<float3> r_normals)
{
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
using namespace blender::math;
BLI_assert(r_normals.size() == tangents.size());
if (r_normals.is_empty()) {
return;
}
const float epsilon = 1e-4f;
/* Set initial normal. */
const float3 &first_tangent = tangents[0];
if (fabs(first_tangent.x) + fabs(first_tangent.y) < epsilon) {
r_normals[0] = {1.0f, 0.0f, 0.0f};
}
else {
BLI: Refactor vector types & functions to use templates This patch implements the vector types (i.e:`float2`) by making heavy usage of templating. All vector functions are now outside of the vector classes (inside the `blender::math` namespace) and are not vector size dependent for the most part. In the ongoing effort to make shaders less GL centric, we are aiming to share more code between GLSL and C++ to avoid code duplication. ####Motivations: - We are aiming to share UBO and SSBO structures between GLSL and C++. This means we will use many of the existing vector types and others we currently don't have (uintX, intX). All these variations were asking for many more code duplication. - Deduplicate existing code which is duplicated for each vector size. - We also want to share small functions. Which means that vector functions should be static and not in the class namespace. - Reduce friction to use these types in new projects due to their incompleteness. - The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a bit of a let down. Most clases are incomplete, out of sync with each others with different codestyles, and some functions that should be static are not (i.e: `float3::reflect()`). ####Upsides: - Still support `.x, .y, .z, .w` for readability. - Compact, readable and easilly extendable. - All of the vector functions are available for all the vectors types and can be restricted to certain types. Also template specialization let us define exception for special class (like mpq). - With optimization ON, the compiler unroll the loops and performance is the same. ####Downsides: - Might impact debugability. Though I would arge that the bugs are rarelly caused by the vector class itself (since the operations are quite trivial) but by the type conversions. - Might impact compile time. I did not saw a significant impact since the usage is not really widespread. - Functions needs to be rewritten to support arbitrary vector length. For instance, one can't call `len_squared_v3v3` in `math::length_squared()` and call it a day. - Type cast does not work with the template version of the `math::` vector functions. Meaning you need to manually cast `float *` and `(float *)[3]` to `float3` for the function calls. i.e: `math::distance_squared(float3(nearest.co), positions[i]);` - Some parts might loose in readability: `float3::dot(v1.normalized(), v2.normalized())` becoming `math::dot(math::normalize(v1), math::normalize(v2))` But I propose, when appropriate, to use `using namespace blender::math;` on function local or file scope to increase readability. `dot(normalize(v1), normalize(v2))` ####Consideration: - Include back `.length()` method. It is quite handy and is more C++ oriented. - I considered the GLM library as a candidate for replacement. It felt like too much for what we need and would be difficult to extend / modify to our needs. - I used Macros to reduce code in operators declaration and potential copy paste bugs. This could reduce debugability and could be reverted. - This touches `delaunay_2d.cc` and the intersection code. I would like to know @howardt opinion on the matter. - The `noexcept` on the copy constructor of `mpq(2|3)` is being removed. But according to @JacquesLucke it is not a real problem for now. I would like to give a huge thanks to @JacquesLucke who helped during this and pushed me to reduce the duplication further. Reviewed By: brecht, sergey, JacquesLucke Differential Revision: https://developer.blender.org/D13791
2022-01-12 12:57:07 +01:00
r_normals[0] = normalize(float3(first_tangent.y, -first_tangent.x, 0.0f));
}
/* Forward normal with minimum twist along the entire spline. */
for (const int i : IndexRange(1, r_normals.size() - 1)) {
r_normals[i] = calculate_next_normal(r_normals[i - 1], tangents[i - 1], tangents[i]);
}
if (!cyclic) {
return;
}
/* Compute how much the first normal deviates from the normal that has been forwarded along the
* entire cyclic spline. */
const float3 uncorrected_last_normal = calculate_next_normal(
r_normals.last(), tangents.last(), tangents[0]);
float correction_angle = angle_signed_on_axis_v3v3_v3(
r_normals[0], uncorrected_last_normal, tangents[0]);
if (correction_angle > M_PI) {
correction_angle = correction_angle - 2 * M_PI;
}
/* Gradually apply correction by rotating all normals slightly. */
const float angle_step = correction_angle / r_normals.size();
for (const int i : r_normals.index_range()) {
const float angle = angle_step * i;
r_normals[i] = rotate_direction_around_axis(r_normals[i], tangents[i], angle);
}
}
Span<float3> Spline::evaluated_normals() const
{
if (!normal_cache_dirty_) {
return evaluated_normals_cache_;
}
std::lock_guard lock{normal_cache_mutex_};
if (!normal_cache_dirty_) {
return evaluated_normals_cache_;
}
const int eval_size = this->evaluated_points_size();
evaluated_normals_cache_.resize(eval_size);
Span<float3> tangents = this->evaluated_tangents();
MutableSpan<float3> normals = evaluated_normals_cache_;
/* Only Z up normals are supported at the moment. */
switch (this->normal_mode) {
case ZUp: {
calculate_normals_z_up(tangents, normals);
break;
}
case Minimum: {
calculate_normals_minimum(tangents, is_cyclic_, normals);
break;
}
case Tangent: {
/* Tangent mode is not yet supported. */
calculate_normals_z_up(tangents, normals);
break;
}
}
/* Rotate the generated normals with the interpolated tilt data. */
Geometry Nodes: refactor virtual array system Goals of this refactor: * Simplify creating virtual arrays. * Simplify passing virtual arrays around. * Simplify converting between typed and generic virtual arrays. * Reduce memory allocations. As a quick reminder, a virtual arrays is a data structure that behaves like an array (i.e. it can be accessed using an index). However, it may not actually be stored as array internally. The two most important implementations of virtual arrays are those that correspond to an actual plain array and those that have the same value for every index. However, many more implementations exist for various reasons (interfacing with legacy attributes, unified iterator over all points in multiple splines, ...). With this refactor the core types (`VArray`, `GVArray`, `VMutableArray` and `GVMutableArray`) can be used like "normal values". They typically live on the stack. Before, they were usually inside a `std::unique_ptr`. This makes passing them around much easier. Creation of new virtual arrays is also much simpler now due to some constructors. Memory allocations are reduced by making use of small object optimization inside the core types. Previously, `VArray` was a class with virtual methods that had to be overridden to change the behavior of a the virtual array. Now,`VArray` has a fixed size and has no virtual methods. Instead it contains a `VArrayImpl` that is similar to the old `VArray`. `VArrayImpl` should rarely ever be used directly, unless a new virtual array implementation is added. To support the small object optimization for many `VArrayImpl` classes, a new `blender::Any` type is added. It is similar to `std::any` with two additional features. It has an adjustable inline buffer size and alignment. The inline buffer size of `std::any` can't be relied on and is usually too small for our use case here. Furthermore, `blender::Any` can store additional user-defined type information without increasing the stack size. Differential Revision: https://developer.blender.org/D12986
2021-11-16 10:15:51 +01:00
VArray<float> tilts = this->interpolate_to_evaluated(this->tilts());
for (const int i : normals.index_range()) {
normals[i] = rotate_direction_around_axis(normals[i], tangents[i], tilts[i]);
}
normal_cache_dirty_ = false;
return evaluated_normals_cache_;
}
Spline::LookupResult Spline::lookup_evaluated_factor(const float factor) const
{
return this->lookup_evaluated_length(this->length() * factor);
}
Spline::LookupResult Spline::lookup_evaluated_length(const float length) const
{
BLI_assert(length >= 0.0f && length <= this->length());
Span<float> lengths = this->evaluated_lengths();
const float *offset = std::lower_bound(lengths.begin(), lengths.end(), length);
const int index = offset - lengths.begin();
const int next_index = (index == this->evaluated_points_size() - 1) ? 0 : index + 1;
const float previous_length = (index == 0) ? 0.0f : lengths[index - 1];
const float length_in_segment = length - previous_length;
const float segment_length = lengths[index] - previous_length;
const float factor = segment_length == 0.0f ? 0.0f : length_in_segment / segment_length;
return LookupResult{index, next_index, factor};
}
Array<float> Spline::sample_uniform_index_factors(const int samples_size) const
{
const Span<float> lengths = this->evaluated_lengths();
BLI_assert(samples_size > 0);
Array<float> samples(samples_size);
samples[0] = 0.0f;
if (samples_size == 1) {
return samples;
}
const float total_length = this->length();
const float sample_length = total_length / (samples_size - (is_cyclic_ ? 0 : 1));
/* Store the length at the previous evaluated point in a variable so it can
* start out at zero (the lengths array doesn't contain 0 for the first point). */
float prev_length = 0.0f;
int i_sample = 1;
for (const int i_evaluated : IndexRange(this->evaluated_edges_size())) {
const float length = lengths[i_evaluated];
/* Add every sample that fits in this evaluated edge. */
while ((sample_length * i_sample) < length && i_sample < samples_size) {
const float factor = (sample_length * i_sample - prev_length) / (length - prev_length);
samples[i_sample] = i_evaluated + factor;
i_sample++;
}
prev_length = length;
}
2021-11-01 14:00:58 +11:00
/* Zero lengths or float inaccuracies can cause invalid values, or simply
* skip some, so set the values that weren't completed in the main loop. */
for (const int i : IndexRange(i_sample, samples_size - i_sample)) {
samples[i] = float(samples_size);
}
if (!is_cyclic_) {
/* In rare cases this can prevent overflow of the stored index. */
samples.last() = lengths.size();
}
return samples;
}
Spline::LookupResult Spline::lookup_data_from_index_factor(const float index_factor) const
{
const int eval_size = this->evaluated_points_size();
if (is_cyclic_) {
if (index_factor < eval_size) {
const int index = std::floor(index_factor);
const int next_index = (index < eval_size - 1) ? index + 1 : 0;
return LookupResult{index, next_index, index_factor - index};
}
return LookupResult{eval_size - 1, 0, 1.0f};
}
if (index_factor < eval_size - 1) {
const int index = std::floor(index_factor);
const int next_index = index + 1;
return LookupResult{index, next_index, index_factor - index};
}
return LookupResult{eval_size - 2, eval_size - 1, 1.0f};
}
void Spline::bounds_min_max(float3 &min, float3 &max, const bool use_evaluated) const
{
Span<float3> positions = use_evaluated ? this->evaluated_positions() : this->positions();
for (const float3 &position : positions) {
minmax_v3v3_v3(min, max, position);
}
}
Geometry Nodes: refactor virtual array system Goals of this refactor: * Simplify creating virtual arrays. * Simplify passing virtual arrays around. * Simplify converting between typed and generic virtual arrays. * Reduce memory allocations. As a quick reminder, a virtual arrays is a data structure that behaves like an array (i.e. it can be accessed using an index). However, it may not actually be stored as array internally. The two most important implementations of virtual arrays are those that correspond to an actual plain array and those that have the same value for every index. However, many more implementations exist for various reasons (interfacing with legacy attributes, unified iterator over all points in multiple splines, ...). With this refactor the core types (`VArray`, `GVArray`, `VMutableArray` and `GVMutableArray`) can be used like "normal values". They typically live on the stack. Before, they were usually inside a `std::unique_ptr`. This makes passing them around much easier. Creation of new virtual arrays is also much simpler now due to some constructors. Memory allocations are reduced by making use of small object optimization inside the core types. Previously, `VArray` was a class with virtual methods that had to be overridden to change the behavior of a the virtual array. Now,`VArray` has a fixed size and has no virtual methods. Instead it contains a `VArrayImpl` that is similar to the old `VArray`. `VArrayImpl` should rarely ever be used directly, unless a new virtual array implementation is added. To support the small object optimization for many `VArrayImpl` classes, a new `blender::Any` type is added. It is similar to `std::any` with two additional features. It has an adjustable inline buffer size and alignment. The inline buffer size of `std::any` can't be relied on and is usually too small for our use case here. Furthermore, `blender::Any` can store additional user-defined type information without increasing the stack size. Differential Revision: https://developer.blender.org/D12986
2021-11-16 10:15:51 +01:00
GVArray Spline::interpolate_to_evaluated(GSpan data) const
{
Geometry Nodes: refactor virtual array system Goals of this refactor: * Simplify creating virtual arrays. * Simplify passing virtual arrays around. * Simplify converting between typed and generic virtual arrays. * Reduce memory allocations. As a quick reminder, a virtual arrays is a data structure that behaves like an array (i.e. it can be accessed using an index). However, it may not actually be stored as array internally. The two most important implementations of virtual arrays are those that correspond to an actual plain array and those that have the same value for every index. However, many more implementations exist for various reasons (interfacing with legacy attributes, unified iterator over all points in multiple splines, ...). With this refactor the core types (`VArray`, `GVArray`, `VMutableArray` and `GVMutableArray`) can be used like "normal values". They typically live on the stack. Before, they were usually inside a `std::unique_ptr`. This makes passing them around much easier. Creation of new virtual arrays is also much simpler now due to some constructors. Memory allocations are reduced by making use of small object optimization inside the core types. Previously, `VArray` was a class with virtual methods that had to be overridden to change the behavior of a the virtual array. Now,`VArray` has a fixed size and has no virtual methods. Instead it contains a `VArrayImpl` that is similar to the old `VArray`. `VArrayImpl` should rarely ever be used directly, unless a new virtual array implementation is added. To support the small object optimization for many `VArrayImpl` classes, a new `blender::Any` type is added. It is similar to `std::any` with two additional features. It has an adjustable inline buffer size and alignment. The inline buffer size of `std::any` can't be relied on and is usually too small for our use case here. Furthermore, `blender::Any` can store additional user-defined type information without increasing the stack size. Differential Revision: https://developer.blender.org/D12986
2021-11-16 10:15:51 +01:00
return this->interpolate_to_evaluated(GVArray::ForSpan(data));
}
void Spline::sample_with_index_factors(const GVArray &src,
Span<float> index_factors,
GMutableSpan dst) const
{
BLI_assert(src.size() == this->evaluated_points_size());
blender::attribute_math::convert_to_static_type(src.type(), [&](auto dummy) {
using T = decltype(dummy);
Geometry Nodes: refactor virtual array system Goals of this refactor: * Simplify creating virtual arrays. * Simplify passing virtual arrays around. * Simplify converting between typed and generic virtual arrays. * Reduce memory allocations. As a quick reminder, a virtual arrays is a data structure that behaves like an array (i.e. it can be accessed using an index). However, it may not actually be stored as array internally. The two most important implementations of virtual arrays are those that correspond to an actual plain array and those that have the same value for every index. However, many more implementations exist for various reasons (interfacing with legacy attributes, unified iterator over all points in multiple splines, ...). With this refactor the core types (`VArray`, `GVArray`, `VMutableArray` and `GVMutableArray`) can be used like "normal values". They typically live on the stack. Before, they were usually inside a `std::unique_ptr`. This makes passing them around much easier. Creation of new virtual arrays is also much simpler now due to some constructors. Memory allocations are reduced by making use of small object optimization inside the core types. Previously, `VArray` was a class with virtual methods that had to be overridden to change the behavior of a the virtual array. Now,`VArray` has a fixed size and has no virtual methods. Instead it contains a `VArrayImpl` that is similar to the old `VArray`. `VArrayImpl` should rarely ever be used directly, unless a new virtual array implementation is added. To support the small object optimization for many `VArrayImpl` classes, a new `blender::Any` type is added. It is similar to `std::any` with two additional features. It has an adjustable inline buffer size and alignment. The inline buffer size of `std::any` can't be relied on and is usually too small for our use case here. Furthermore, `blender::Any` can store additional user-defined type information without increasing the stack size. Differential Revision: https://developer.blender.org/D12986
2021-11-16 10:15:51 +01:00
const VArray<T> src_typed = src.typed<T>();
MutableSpan<T> dst_typed = dst.typed<T>();
if (src.size() == 1) {
dst_typed.fill(src_typed[0]);
return;
}
blender::threading::parallel_for(dst_typed.index_range(), 1024, [&](IndexRange range) {
for (const int i : range) {
const LookupResult interp = this->lookup_data_from_index_factor(index_factors[i]);
dst_typed[i] = blender::attribute_math::mix2(interp.factor,
src_typed[interp.evaluated_index],
src_typed[interp.next_evaluated_index]);
}
});
});
}