diff --git a/src/encode.rs b/src/encode.rs
index 604620e..1e80cc0 100644
--- a/src/encode.rs
+++ b/src/encode.rs
@@ -4,6 +4,8 @@ use core::convert::TryFrom;
 #[cfg(feature = "std")]
 use std::io::Write;
 
+use bytemuck::Pod;
+
 use crate::consts::{QOI_HEADER_SIZE, QOI_OP_INDEX, QOI_OP_RUN, QOI_PADDING, QOI_PADDING_SIZE};
 use crate::error::{Error, Result};
 use crate::header::Header;
@@ -17,6 +19,7 @@ use crate::utils::{unlikely, BytesMut, Writer};
 fn encode_impl<W: Writer, const N: usize>(mut buf: W, data: &[u8]) -> Result<usize>
 where
     Pixel<N>: SupportedChannels,
+    [u8; N]: Pod,
 {
     let cap = buf.capacity();
 
diff --git a/src/pixel.rs b/src/pixel.rs
index 90674e4..784726e 100644
--- a/src/pixel.rs
+++ b/src/pixel.rs
@@ -1,6 +1,7 @@
 use crate::consts::{QOI_OP_DIFF, QOI_OP_LUMA, QOI_OP_RGB, QOI_OP_RGBA};
 use crate::error::Result;
 use crate::utils::Writer;
+use bytemuck::{cast, Pod};
 
 #[derive(Copy, Clone, PartialEq, Eq, Debug)]
 #[repr(transparent)]
@@ -107,12 +108,18 @@ impl<const N: usize> Pixel<N> {
     }
 
     #[inline]
-    pub const fn hash_index(self) -> u8 {
-        let r = self.r().wrapping_mul(3);
-        let g = self.g().wrapping_mul(5);
-        let b = self.b().wrapping_mul(7);
-        let a = self.a_or(0xff).wrapping_mul(11);
-        r.wrapping_add(g).wrapping_add(b).wrapping_add(a) % 64
+    pub fn hash_index(self) -> u8
+    where
+        [u8; N]: Pod,
+    {
+        // credits for the initial idea: @zakarumych
+        let v = if N == 4 {
+            u32::from_ne_bytes(cast(self.0))
+        } else {
+            u32::from_ne_bytes([self.0[0], self.0[1], self.0[2], 0xff])
+        } as u64;
+        let s = ((v & 0xff00_ff00) << 32) | (v & 0x00ff_00ff);
+        s.wrapping_mul(0x0300_0700_0005_000b_u64).to_le().swap_bytes() as u8 & 63
     }
 
     #[inline]