std.image
std.image
std.image loads, inspects, saves, and converts images. BMP is always
available (native); PNG and JPEG require libpng/libjpeg at build
time.
Reference
load(path) # -> {width, height, format, pixels} (pixels: RGBA8 binary)
info(path) # -> {width, height, format}
save(path, img[, quality = 90]) # img: load()-shaped object
save(path, w, h, pixels[, quality = 90]) # legacy raw-pixel form
convert(in, out[, quality = 90])
capability() # -> {bmp, png, jpg}
| Function | Description |
|---|---|
load(path) |
Load image → {width, height, format, pixels} |
info(path) |
Returns {width, height, format} (the file is still fully decoded; only the pixel buffer is omitted) |
save(path, img[, q]) |
Save a load()-shaped {width, height, pixels} object; format from file extension (the object’s format field is informational and ignored) |
save(path, w, h, pixels[, q]) |
Legacy raw-pixel form |
convert(in, out[, q]) |
Convert between formats |
capability() |
Which formats are available |
pixels is an RGBA8 binary string of width * height * 4 bytes. quality
(1–100, default 90) affects JPEG output.
Decoded images are capped at 30000 pixels per side and 64M pixels total
(e.g. 30000×30000 exceeds the budget and is rejected).
In guarded engines, load/info need filesystem-read plus native-parse,
save needs filesystem-write, and convert needs all three; missing
capabilities strip the function.
Examples
var caps = std.image.capability()
print(caps.bmp, caps.png, caps.jpg)
var img = std.image.load("photo.png")
print(img.width, "x", img.height, img.format) # e.g. 800 x 600 png
std.image.save("copy.png", img)
std.image.convert("photo.png", "photo.jpg", 85)
# inspect without decoding everything
var meta = std.image.info("a.jpg")
print(meta.width, meta.format)
Paths accept ~ and $VAR expansion.
See also
Last updated:
Recent commits
