aznyan is a collection of image filters for R that wraps OpenCV, ported from 5PB-3-4/AviUtl_OpenCV_Scripts.
Still in development. It will probably work, but the API is likely to change.
aznyan provides functions that take a nativeRaster of image data as
their first argument and return a nativeRaster after applying the
effect. All these functions return a new nativeRaster and do not
modify the input object in place.
A nativeRaster is a built-in datatype in R that stores image data as
an integer matrix. There are several common ways to obtain a
nativeRaster object:
- Generate or manipulate a
nativeRasterusing nara - Capture the contents of a graphics device with
ragg::agg_capture() - Read image files using functions such as
jpeg::readJPEG(native = TRUE)png::readPNG(native = TRUE)fastpng::read_png(type = "nativeraster")aznyan::read_still()
For example, you can read a PNG image as a nativeRaster like this:
library(ggplot2)
library(patchwork) # for layouting native rasters
png <-
aznyan::read_still(system.file("images/aznyan-256x256.png", package = "aznyan"))This png object corresponds to the following image:
With aznyan, you can easily apply various effects to a nativeRaster
image.
blur <-
wrap_elements(full = aznyan::median_blur(png, ksize = 8)) +
labs(title = "Median Blur")
diffusion <-
wrap_elements(full = aznyan::diffusion_filter(png, factor = 8)) +
labs(title = "Diffusion Filter")
morph <-
wrap_elements(full = aznyan::morphology(png, ksize = c(4, 4, 4))) +
labs(title = "Morphological\nTransformation")
(blur | diffusion | morph)
In addition to general-purpose image effects, aznyan also provides photo-style color filters ported from Rustagram, such as the following:
In practice, a convenient way to use aznyan is as a custom filter via
ggfx::with_custom(),
which allows you to apply effects to arbitrary ggplot layers. For
example:
library(ggfx)
my_effect <- \(x, ...) {
vp <- get_viewport_area(x)
vp <- aznyan::stylize(vp, ...)
set_viewport_area(x, vp)
}
ggplot(mpg) +
with_custom(
geom_density(aes(cty, fill = drv), alpha = 0.4, adjust = 2),
filter = my_effect
) +
scale_fill_viridis_d(option = "plasma")
MIT License.