Struct Camera2d 

pub struct Camera2d {
    pub center: Vec2,
    pub offset: Vec2,
    pub units_per_pixel: f32,
    pub zoom: f32,
}

2D orthographic camera. Owned by the UI (signals), copied into the snapshot per frame.

The camera’s look point is effective_center (center + offset); zoom and units_per_pixel scale the view uniformly. All three viewport consumers (canvas drawing, the GPU batch matrix, and pointer picks) derive from these fields through one shared transform, so they can never disagree.

Fields

center: Vec2

World-space point the camera follows (the follow target, room center, or player position). Defaults to (0, 0).

Scroll limits (see apply_limits) clamp this field; put momentary displacement such as trauma shake in offset instead so it can push past the limits.

offset: Vec2

Momentary displacement added to center. Defaults to (0, 0).

Useful for looking around or camera shake animations: applied after limits, so a shake impulse still moves the view even when the follow target is pinned at a scroll edge. Decays back to zero under game-side trauma handling.

units_per_pixel: f32

World units per screen pixel at zoom 1. Defaults to 1.0.

Combined with the viewport size to build the orthographic projection: a smaller value shows less of the world (larger sprites). Together with zoom, the visible world width is viewport_dp * units_per_pixel / zoom. Non-positive values fall back to 1.0 so a bad snapshot can never divide by zero.

zoom: f32

Magnification. Defaults to 1.0.

Higher values zoom in: 2.0 shows half the world width on each axis (a quarter of the area); 0.5 shows twice as much. Applied on canvas and GPU alike. Non-positive values fall back to 1.0.

Implementations

impl Camera2d

pub fn effective_center(&self) -> [f32; 2]

Effective look point: center + offset.

All framing (canvas drawing, the GPU batch matrix, pointer picks, actor surfaces) uses this value, so shake and look-around applied through offset move every consumer together.

pub fn view_proj(&self, viewport_px: [f32; 2]) -> Mat4

Y-down view-projection: world +y points screen-down. Guards degenerate inputs (zero viewport/zoom) so a bad snapshot can never div-by-zero; callers driving viewports should prefer fit_view_proj (contain-fit + look), which is the single framing contract canvas, GPU, and picks share.

pub fn screen_to_world(&self, viewport_px: [f32; 2], px: [f32; 2]) -> Vec2

World-space position under a viewport-pixel cursor position. Guards zero viewports (returns the look point instead of NaN).

Trait Implementations

impl Clone for Camera2d

fn clone(&self) -> Camera2d

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for Camera2d

impl Debug for Camera2d

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl Default for Camera2d

fn default() -> Camera2d

Returns the “default value” for a type.