diff --git a/Cargo.toml b/Cargo.toml
index 721b047..ed1b715 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "fimg"
-version = "0.4.22"
+version = "0.4.23"
 authors = ["bend-n <bend.n@outlook.com>"]
 license = "MIT"
 edition = "2021"
@@ -17,6 +17,7 @@ fontdue = { version = "0.7.3", optional = true }
 vecto = "0.1.0"
 umath = "0.0.7"
 fr = { version = "0.1.1", package = "fer", optional = true }
+stackblur-iter = { version = "0.2.0", features = ["simd"], optional = true }
 
 [dev-dependencies]
 iai = { git = "https://github.com/bend-n/iai.git" }
@@ -50,6 +51,7 @@ harness = false
 scale = ["fr"]
 save = ["png"]
 text = ["fontdue"]
+blur = ["stackblur-iter"]
 default = ["save", "scale"]
 
 [profile.release]
diff --git a/README.md b/README.md
index 7ad63a6..90be93f 100644
--- a/README.md
+++ b/README.md
@@ -14,4 +14,5 @@ quick simple image operations
 - [x] box drawing
 - [x] polygon drawing
 - [x] circle drawing
-- [x] text drawing
\ No newline at end of file
+- [x] text drawing
+- [x] blur
\ No newline at end of file
diff --git a/src/blur.rs b/src/blur.rs
new file mode 100644
index 0000000..62ec9fa
--- /dev/null
+++ b/src/blur.rs
@@ -0,0 +1,51 @@
+use stackblur_iter::imgref::ImgRefMut;
+
+use crate::{pixels::convert::PFrom, Image};
+
+impl<T: AsMut<[u32]> + AsRef<[u32]>> Image<T, 1> {
+    /// Blur a image of packed 32 bit integers, `[0xAARRGGBB]`.
+    pub fn blur_argb(&mut self, radius: usize) {
+        let w = self.width() as usize;
+        let h = self.height() as usize;
+        stackblur_iter::simd_blur_argb::<4>(&mut ImgRefMut::new(self.buffer.as_mut(), w, h), radius)
+    }
+}
+
+impl<const N: usize> Image<Box<[u8]>, N>
+where
+    [u8; 4]: PFrom<N>,
+    [u8; N]: PFrom<4>,
+{
+    /// Blur a image.
+    /// ```
+    /// # use fimg::Image;
+    /// let mut i = Image::alloc(300, 300).boxed();
+    /// // draw a lil pentagon
+    /// i.poly((150., 150.), 5, 100.0, 0.0, [255]);
+    /// // give it some blur
+    /// i.blur(25);
+    /// assert_eq!(include_bytes!("../tdata/blurred_pentagon.imgbuf"), i.bytes())
+    /// ```
+    pub fn blur(&mut self, radius: usize) {
+        // you know, i optimized blurslice a fair bit, and yet, despite all the extra bit twiddling stackblur-iter is faster.
+        let mut argb = Image::<Box<[u32]>, 1>::from(self.as_ref());
+        argb.blur_argb(radius);
+        for (i, n) in crate::convert::unpack_all(&argb.buffer).enumerate() {
+            *unsafe { self.buffer.get_unchecked_mut(i) } = n;
+        }
+    }
+}
+
+impl<const N: usize> Image<&[u8], N>
+where
+    [u8; 4]: PFrom<N>,
+    [u8; N]: PFrom<4>,
+{
+    /// Blur a image.
+    pub fn blur(self, radius: usize) -> Image<Box<[u8]>, N> {
+        let mut argb = Image::<Box<[u32]>, 1>::from(self);
+        argb.blur_argb(radius);
+        // SAFETY: ctor
+        unsafe { Image::new(argb.width, argb.height, &**argb.buffer()) }.into()
+    }
+}
diff --git a/src/convert.rs b/src/convert.rs
index 5704f26..6f44116 100644
--- a/src/convert.rs
+++ b/src/convert.rs
@@ -62,3 +62,57 @@ boxconv!(3 => 4);
 boxconv!(4 => 1);
 boxconv!(4 => 2);
 boxconv!(4 => 3);
+
+#[inline]
+fn pack([r, g, b, a]: [u8; 4]) -> u32 {
+    ((a as u32) << 24) | ((r as u32) << 16) | ((g as u32) << 8) | (b as u32)
+}
+
+#[inline]
+fn unpack(n: u32) -> [u8; 4] {
+    [
+        ((n >> 16) & 0xFF) as u8,
+        ((n >> 8) & 0xFF) as u8,
+        (n & 0xFF) as u8,
+        ((n >> 24) & 0xFF) as u8,
+    ]
+}
+
+impl<const N: usize> From<Image<&[u8], N>> for Image<Box<[u32]>, 1>
+where
+    [u8; 4]: PFrom<N>,
+{
+    /// Pack into ARGB.
+    fn from(value: Image<&[u8], N>) -> Self {
+        let buf = value
+            .chunked()
+            .copied()
+            .map(PFrom::pfrom)
+            .map(pack)
+            .collect();
+        // SAFETY: ctor
+        unsafe { Self::new(value.width, value.height, buf) }
+    }
+}
+
+pub fn unpack_all<const N: usize>(buffer: &[u32]) -> impl Iterator<Item = u8> + '_
+where
+    [u8; N]: PFrom<4>,
+{
+    buffer
+        .iter()
+        .copied()
+        .map(unpack)
+        .flat_map(<[u8; N] as PFrom<4>>::pfrom)
+}
+
+impl<const N: usize> From<Image<&[u32], 1>> for Image<Box<[u8]>, N>
+where
+    [u8; N]: PFrom<4>,
+{
+    fn from(value: Image<&[u32], 1>) -> Self {
+        let buf = unpack_all(value.buffer).collect();
+        // SAFETY: ctor
+        unsafe { Self::new(value.width, value.height, buf) }
+    }
+}
diff --git a/src/lib.rs b/src/lib.rs
index e325c6c..52bc7ed 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -34,6 +34,7 @@
 //! Misc image ops:
 //! - [`Image::repeated`]
 //! - [`Image::overlay`](Overlay), [`Image::overlay_at`](OverlayAt), [`Image::overlay_blended`](BlendingOverlay)
+//! - [`Image::blur`]
 #![feature(
     slice_swap_unchecked,
     generic_const_exprs,
@@ -60,6 +61,8 @@
 use std::{num::NonZeroU32, slice::SliceIndex};
 
 mod affine;
+#[cfg(feature = "blur")]
+mod blur;
 #[doc(hidden)]
 pub mod builder;
 #[doc(hidden)]
diff --git a/src/pixels/convert.rs b/src/pixels/convert.rs
index 55c9338..13b7e62 100644
--- a/src/pixels/convert.rs
+++ b/src/pixels/convert.rs
@@ -24,7 +24,7 @@ impl PFrom<2> for Y {
 
 impl PFrom<3> for Y {
     fn pfrom([r, g, b]: RGB) -> Self {
-        [((2126 * r as u16 + 7152 * g as u16 + 722 * b as u16) / 10000) as u8]
+        [((2126 * r as u32 + 7152 * g as u32 + 722 * b as u32) / 10000) as u8]
     }
 }
 
diff --git a/tdata/blurred_pentagon.imgbuf b/tdata/blurred_pentagon.imgbuf
new file mode 100644
index 0000000..9580840
Binary files /dev/null and b/tdata/blurred_pentagon.imgbuf differ