Im3d may be configured by modifying im3d_config.h.

• IM3D_ASSERT(e) - Application-defined assertion handler (default is cassert assert()).
• IM3D_MALLOC(size), IM3D_FREE(ptr) - Application-defined malloc/free (default is cstdlib malloc()/free()).
• IM3D_THREAD_LOCAL_CONTEXT_PTR - Define to enable a thread-local context ptr. This is a prerequisite for multi threaded use of Im3d via multiple contexts. Note that although this makes Im3d's internal context ptr thread local, the application must allocate and set per-thread context instances.
• IM3D_MATRIX_ROW_MAJOR - Use row-major internal matrix layout.
• IM3D_VERTEX_ALIGNMENT - Force vertex data alignment (default is 4 bytes). Whether this is required depends on the graphics API and rendering method used by the application, e.g. examples/OpenGL31 requires 16 byte alignment for vertex data.
• IM3D_CULL_PRIMITIVES - Enable internal culling for primitives (see Culling).
• IM3D_CULL_GIZMOS - Enable internal culling for gizmos (see Culling).

Some interfaces in the API take internal math types as arguments (Im3d::Vec2, Im3d::Vec4, etc.). It is therefore useful to define implicit conversion between Im3d and application math types. This is done by defining an IM3D_X_APP macro (where X is the type name) which is injected into the Im3d type's definition. E.g.

#define IM3D_VEC2_APP \
	Vec2(const linalg::vec2& _v)       { x = _v.x; y = _v.y;     } \
	operator linalg::vec2() const      { return linalg::vec2(x, y); }