---
title: "Technical Notes"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Technical Notes}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```

```{r setup}
library(nara)
```



### What is a native raster image and why is it fast?


A native raster image is a built-in datatype in R.

It is an integer matrix where each integer represents the RGBA color at 
a single pixel.  The 32-bit integer at each location is interpreted within
R to be four color channels (RGBA) represented by 8 bits each.

This way of encoding color information is closer to the internal representation
used by graphics devices, and therefore can be faster to render and manipulate.

Native rasters do **not** use pre-multiplied alpha.

### In-place operation

`{nara}` tries to minimise
R function calls and memory allocations in order to be fast.

When updating native raster image with this package, changes are done
*in place* on the current image i.e. a new image is **not** created.

### Anti-aliasing/Interpolation

No anti-aliasing is done by the draw methods in this package.

No interpolation is done - `x` and `y` values for drawing coordinates are
always rounded to integers.

### Dimension ordering

All arguments specifying dimensions are in the order **horizontal** then **vertical** i.e.

* x, y
* width, height
* hjust, vjust

### Coordinate System

The coordinate system for `nara` native raster image has the origin
at the **top left corner** with coordinates `(0, 0)`.

This is equivalent to `{grid}` graphics using `native` units.

It is also how `{magick}` represents image coordinates, as well as the majority
of C graphics libraries.