Vortex2D Versions Save

• velocity interpolation is now configurable (linear or cubic)
• added function to calculate error residue of linear equations
• multigrid can be used as linear solver
• configurable particle size
• added debug data of linear equations (to be viewed as texture in renderdoc)
• support any size of world (not restricted to power of 2)
• improved level set generation from particles
• improved rendering of level sets

• Added SetType to rigidbody
• Additional checks for texture and buffer copies
• Added compute and graphics pipeline cache
• Buffer not cleared on construction anymore
• Fix PCG exit condition
• Not loading surface for unit-tests
• Clear swapchains on startup
• Fix missing transferDst flag to swapchain images
• Documentation improvements
• vk_mem_alloc upgrade to 2.2.0
• Fix prefix scan compute shader constants
• Using spirv-cross version 2019-03-29
• Prefer discrete gpu when creating device
• Use static version of MoltenVK
• Added ParticlePhi method to WaterWorld
• Clear texture on construction only if device local memory
• Fix family queue index selection
• Minimum cmake requirements bumbed to 3.11 (because of FetchContent)
• Fixed high-dpi scaling for example
• Improved cmake install
• Improved dynamic loading of vulkan extension functions

• Improved API of command buffer and render command. All functions return *this.
• Moved ExecuteCommand as Execute in Device.
• Added buffer resize function.
• Added IndexBuffer.
• Better buffer copy methods.
• Fix buffer fill.
• Better texture copy methods.
• Added dynamic viewport and scissor to pipelines.
• Added shape interface
• Removed (emulated) properties of Transformable.
• Added clang-format.
• Fixes around dllexport/dllimport on windows.
• Fix layout for swapchain

• Bicubic interpolation
• Add velocity operator: add/set
• Overhauled rigidbody interface
• Correctly enforce sub-steps
• User setting of solver parameters
• Particle bounds fix
• Minor vulkan fixes

• Support for running simulation with multiple sub-steps per steps
• Rename world function "Solve" to "Step"
• Removed scaling functionality
• Fixed Timer class
• Added option to specify specialization constants in shaders
• CMake improvements when including in other projects
• Initialize to Vulkan 1.1 by default (with fallback to 1.0)
• Improve ComputeSize class

• Added strong rigidbody coupling
• Added support for blend constants
• Scaling of velocities when setting them in rigidbodies
• Using mapbox variant class
• Fix rigidbody div calculation
• Fix rigidbody centre calculation

• Fix broken particle unit-test from glm update
• Fix documentation on glsl compiler

• Faster linear solver
• Added calculation for the CFL number
• Updated glm version to 0.9.9.0
• Fixed initialization of glm variables
• Improved push constant API for compute shader
• Improved Conjugate Gradient exit condition

First release!

• macOS and iOS support via MoltenVK
• compiling shaders with glslangValidator instead of glslc
• improved signed distance field computation from polygons
• added Wait() function to RenderCommand
• correct synchronisation for present queue
• improved cmake configuration