math/project.h

#include <modkit/math/project.h>8 functions

Coordinate space transformations and MVP pipeline.

Coordinate spaces (right-hand system): Local/Object -> Model matrix -> World World -> View matrix -> View/Eye View/Eye -> Projection -> Clip (NDC after perspective divide) Clip -> Viewport -> Screen The typical rendering pipeline uses: MVP = Projection * View * Model The GPU backend computes modelViewProj automatically from:mk_view_transform(view, projection)mk_encoder_set_transform(model)

Functions

mk_compute_model_viewfunction

void mk_compute_model_view(const mk_mat4 *model, const mk_mat4 *view, mk_mat4 *dest)

Compute model-view matrix: result = view * model.

Parameter	Type	Description
model	const mk_mat4 *	Value for model.
view	const mk_mat4 *	Value for view.
dest	mk_mat4 *	Value for dest.

mk_compute_mvpfunction

void mk_compute_mvp(const mk_mat4 *model, const mk_mat4 *view, const mk_mat4 *projection, mk_mat4 *dest)

Compute MVP matrix: result = projection * view * model.

Parameter	Type	Description
model	const mk_mat4 *	Value for model.
view	const mk_mat4 *	Value for view.
projection	const mk_mat4 *	Value for projection.
dest	mk_mat4 *	Value for dest.

mk_compute_view_projectionfunction

void mk_compute_view_projection(const mk_mat4 *view, const mk_mat4 *projection, mk_mat4 *dest)

Compute view-projection matrix: result = projection * view.

Parameter	Type	Description
view	const mk_mat4 *	Value for view.
projection	const mk_mat4 *	Value for projection.
dest	mk_mat4 *	Value for dest.

mk_project_world_to_screenfunction

void mk_project_world_to_screen(const mk_vec3 *world_pos, const mk_mat4 *mvp, const mk_vec4 *viewport, mk_vec3 *screen_pos)

Project world coordinate to screen coordinate.

Parameter	Type	Description
world_pos	const mk_vec3 *	Position in world space
mvp	const mk_mat4 *	Model-View-Projection matrix
viewport	const mk_vec4 *	Viewport as [x, y, width, height]
screen_pos	mk_vec3 *	Output screen position (z = normalized depth)

mk_screen_to_rayfunction

void mk_screen_to_ray(float screen_x, float screen_y, const mk_camera *camera, const mk_vec4 *viewport, mk_vec3 *ray_origin, mk_vec3 *ray_dir)

Create ray from screen position (for picking/raycasting).

Screen coordinates are top-left origin (y increases downward), matching mk_mouse_position(); the Y flip to clip space is handled internally.

Parameter	Type	Description
screen_x	float	Screen X coordinate
screen_y	float	Screen Y coordinate
camera	const mk_camera *	Camera to create ray from
viewport	const mk_vec4 *	Viewport as [x, y, width, height]
ray_origin	mk_vec3 *	Output ray origin (camera position)
ray_dir	mk_vec3 *	Output ray direction (normalized)

mk_unproject_screen_to_worldfunction

void mk_unproject_screen_to_world(const mk_vec3 *screen_pos, const mk_mat4 *inv_mvp, const mk_vec4 *viewport, mk_vec3 *world_pos)

Unproject screen coordinate to world coordinate.

Parameter	Type	Description
screen_pos	const mk_vec3 *	Screen position (z = normalized depth 0-1)
inv_mvp	const mk_mat4 *	Inverse of MVP matrix
viewport	const mk_vec4 *	Viewport as [x, y, width, height]
world_pos	mk_vec3 *	Output world position

mk_viewportfunction

mk_vec4 mk_viewport(float x, float y, float width, float height)

Create viewport vec4 from dimensions.

Parameter	Type	Description
x	float	Value for x.
y	float	Value for y.
width	float	Value for width.
height	float	Value for height.

Returns The resulting value.

mk_viewport_fullfunction

mk_vec4 mk_viewport_full(float width, float height)

Create viewport from 0,0 with given dimensions.

Parameter	Type	Description
width	float	Value for width.
height	float	Value for height.

Returns The resulting value.