1st, 3rd, 4th dimension #5
Conversation
finished i think? probably some bugs lingering I need to add an example
surprised the github test didnt find it
examples/point-cloud/pointcloud.cpp
Outdated
| #if 0 //internal testing | ||
| noiz::Noise2f noise2; | ||
| noiz::Noise_Processor2<float> noise_processor2{noise2}; | ||
|
|
||
| noiz::Vec3f testing_point; | ||
| noise_processor.basic_processing(testing_point); | ||
| noise_processor.billowy_processing(testing_point); | ||
| noise_processor.hybrid_multi_fractal_processing(testing_point); | ||
| noise_processor.rigid_processing(testing_point); | ||
| noise_processor.turbulence_processing(testing_point); | ||
| noise_processor.raw_noise(testing_point); | ||
|
|
||
| noiz::Vec2f testing_point2; | ||
| noise_processor2.basic_processing(testing_point2); | ||
| noise_processor2.billowy_processing(testing_point2); | ||
| noise_processor2.hybrid_multi_fractal_processing(testing_point2); | ||
| noise_processor2.rigid_processing(testing_point2); | ||
| noise_processor2.turbulence_processing(testing_point2); | ||
| noise_processor2.raw_noise(testing_point2); | ||
| #endif |
There was a problem hiding this comment.
Commented code, should remove before merging.
noiz/include/noiz/cell3.hpp
Outdated
| Type left_top_below{}; | ||
| Type right_top_below{}; | ||
| Type left_bottom_below{}; | ||
| Type right_bottom_below{}; | ||
|
|
||
| Type left_top_above{}; | ||
| Type right_top_above{}; | ||
| Type left_bottom_above{}; | ||
| Type right_bottom_above{}; |
There was a problem hiding this comment.
I don't really have the visualization in my head like you do, but shouldn't the suffixes be front / back or something? Since below is basically = bottom.
| /// | ||
| /// \brief Obtain the next random unit Vec. | ||
| /// \returns The next random unit Vec. | ||
| /// | ||
| /// //i dont think this is used anywhere |
There was a problem hiding this comment.
Yeah it's not used by the library itself, but is available for users.
noiz/include/noiz/detail/grid4.hpp
Outdated
| //if this is multi-threaded, accessing these variables directly could lead to strange behavior | ||
| //i.e. a handful of threads processing noise, then another comes in and changes octave?? | ||
| //not sure if thats a use case worth covering? | ||
| noiz::Noise2<Type>& noise; |
There was a problem hiding this comment.
Storing a reference member deletes assignment, which is OK if this class isn't intended to be stored / used as a member itself. But if it is, then it will be annoying to deal with:
auto np = NoiseProcessor2{};
np = NoiseProcessor2{}; // compile errorAnd the same error will cascade through any types that have it as a member.
To avoid that, use std::reference_wrapper<Noise2<Type>> or store a pointer (but that will then require null checks everywhere, so it's worse).
| auto raw_noise(noiz::Vec2<Type> const& point) -> Type{ | ||
| //redundant | ||
| return noise.at(point * step); | ||
| } |
There was a problem hiding this comment.
Looks like this function could be const?
There was a problem hiding this comment.
Lots of comments here, not sure if you want to commit all of them. File is not fully formatted either.
In general you want to keep member functions of a class const unless they change visible state of the type (ie users can notice the change). Since Noise_Processor2 stores a reference (wrapper) and exposes a bunch of knobs and tweaks, and most of its member functions don't change any of that state (ie don't rebind the reference to another Noise2 instance, and don't modify the knobs), all of them should be const.
noiz/include/noiz/cell4.hpp
Outdated
| //16 is 2^4 | ||
| std::array<Type, 16> corners; //NOLINT | ||
| //16 is 2^4 | ||
| std::array<Type, 16> corners; |
There was a problem hiding this comment.
Should initialize this with {}, would risk being garbage otherwise.
### grid4 was missing a few ) wasnt calling index.positions.at(), but instead calling index.at(), now fixed An example for noise4 would expose similar bugs laying around ### Processor processor2/3/4 abstracted to Processor changes in noise2/3/4 and vec2/3/4 to reflect the processor abstraction PointCloud example updated to include new usage
clang pls
Pull Request
1st dimension wasn't added.
The performance difference between a 1st dimension section and a 2nd dimension with only 1 depth in the y dimension should be negligible.
3rd dimension
rebased from a separate branch
point cloud example moved from a separate branch.
Point Cloud example builds a obj file
4th dimension
4th dimension added
no viable 4th dimension example is currently known. I believe the limiting factor is that we either need a custom file format to export the data, or we need a 3d rendering framework/engine to directly render the data without worrying about transferring between programs.