diff --git a/src/libImaging/Bands.c b/src/libImaging/Bands.c
index e1b16b34a..4f50c85a8 100644
--- a/src/libImaging/Bands.c
+++ b/src/libImaging/Bands.c
@@ -17,6 +17,16 @@
 
 #include "Imaging.h"
 
+#include <emmintrin.h>
+#include <mmintrin.h>
+#include <smmintrin.h>
+
+#if defined(__AVX2__)
+    #include <immintrin.h>
+#endif
+
+
+
 Imaging
 ImagingGetBand(Imaging imIn, int band) {
     Imaging imOut;
@@ -52,8 +62,10 @@ ImagingGetBand(Imaging imIn, int band) {
         UINT8 *out = imOut->image8[y];
         x = 0;
         for (; x < imIn->xsize - 3; x += 4) {
-            UINT32 v = MAKE_UINT32(in[0], in[4], in[8], in[12]);
-            memcpy(out + x, &v, sizeof(v));
+            __m128i source = _mm_loadu_si128((__m128i *) in);
+            *((UINT32*) (out + x)) = _mm_cvtsi128_si32(
+                _mm_shuffle_epi8(source, _mm_set_epi8(
+                    -1,-1,-1,-1, -1,-1,-1,-1, -1,-1,-1,-1, 12,8,4,0)));
             in += 16;
         }
         for (; x < imIn->xsize; x++) {
@@ -99,10 +111,12 @@ ImagingSplit(Imaging imIn, Imaging bands[4]) {
             UINT8 *out1 = bands[1]->image8[y];
             x = 0;
             for (; x < imIn->xsize - 3; x += 4) {
-                UINT32 v = MAKE_UINT32(in[0], in[4], in[8], in[12]);
-                memcpy(out0 + x, &v, sizeof(v));
-                v = MAKE_UINT32(in[0 + 3], in[4 + 3], in[8 + 3], in[12 + 3]);
-                memcpy(out1 + x, &v, sizeof(v));
+                __m128i source = _mm_loadu_si128((__m128i *) in);
+                source = _mm_shuffle_epi8(source, _mm_set_epi8(
+                    15,11,7,3, 14,10,6,2, 13,9,5,1, 12,8,4,0));
+                *((UINT32*) (out0 + x)) = _mm_cvtsi128_si32(source);
+                *((UINT32*) (out1 + x)) = _mm_cvtsi128_si32(
+                    _mm_srli_si128(source, 12));
                 in += 16;
             }
             for (; x < imIn->xsize; x++) {
@@ -119,12 +133,14 @@ ImagingSplit(Imaging imIn, Imaging bands[4]) {
             UINT8 *out2 = bands[2]->image8[y];
             x = 0;
             for (; x < imIn->xsize - 3; x += 4) {
-                UINT32 v = MAKE_UINT32(in[0], in[4], in[8], in[12]);
-                memcpy(out0 + x, &v, sizeof(v));
-                v = MAKE_UINT32(in[0 + 1], in[4 + 1], in[8 + 1], in[12 + 1]);
-                memcpy(out1 + x, &v, sizeof(v));
-                v = MAKE_UINT32(in[0 + 2], in[4 + 2], in[8 + 2], in[12 + 2]);
-                memcpy(out2 + x, &v, sizeof(v));
+                __m128i source = _mm_loadu_si128((__m128i *) in);
+                source = _mm_shuffle_epi8(source, _mm_set_epi8(
+                    15,11,7,3, 14,10,6,2, 13,9,5,1, 12,8,4,0));
+                *((UINT32*) (out0 + x)) = _mm_cvtsi128_si32(source);
+                *((UINT32*) (out1 + x)) = _mm_cvtsi128_si32(
+                    _mm_srli_si128(source, 4));
+                *((UINT32*) (out2 + x)) = _mm_cvtsi128_si32(
+                    _mm_srli_si128(source, 8));
                 in += 16;
             }
             for (; x < imIn->xsize; x++) {
@@ -143,14 +159,16 @@ ImagingSplit(Imaging imIn, Imaging bands[4]) {
             UINT8 *out3 = bands[3]->image8[y];
             x = 0;
             for (; x < imIn->xsize - 3; x += 4) {
-                UINT32 v = MAKE_UINT32(in[0], in[4], in[8], in[12]);
-                memcpy(out0 + x, &v, sizeof(v));
-                v = MAKE_UINT32(in[0 + 1], in[4 + 1], in[8 + 1], in[12 + 1]);
-                memcpy(out1 + x, &v, sizeof(v));
-                v = MAKE_UINT32(in[0 + 2], in[4 + 2], in[8 + 2], in[12 + 2]);
-                memcpy(out2 + x, &v, sizeof(v));
-                v = MAKE_UINT32(in[0 + 3], in[4 + 3], in[8 + 3], in[12 + 3]);
-                memcpy(out3 + x, &v, sizeof(v));
+                __m128i source = _mm_loadu_si128((__m128i *) in);
+                source = _mm_shuffle_epi8(source, _mm_set_epi8(
+                    15,11,7,3, 14,10,6,2, 13,9,5,1, 12,8,4,0));
+                *((UINT32*) (out0 + x)) = _mm_cvtsi128_si32(source);
+                *((UINT32*) (out1 + x)) = _mm_cvtsi128_si32(
+                    _mm_srli_si128(source, 4));
+                *((UINT32*) (out2 + x)) = _mm_cvtsi128_si32(
+                    _mm_srli_si128(source, 8));
+                *((UINT32*) (out3 + x)) = _mm_cvtsi128_si32(
+                    _mm_srli_si128(source, 12));
                 in += 16;
             }
             for (; x < imIn->xsize; x++) {