diff --git a/src/zm_image.cpp b/src/zm_image.cpp
index 7186446ca..b6443ef18 100644
--- a/src/zm_image.cpp
+++ b/src/zm_image.cpp
@@ -3047,9 +3047,6 @@ __attribute__((noinline,__target__("sse2"))) void sse2_delta8_rgba(const uint8_t
 	"mov $0xff, %%eax\n\t"
 	"movd %%eax, %%xmm0\n\t"
 	"pshufd $0x0, %%xmm0, %%xmm0\n\t"
-	"mov $0x80000000, %%eax\n\t"
-	"movd %%eax, %%xmm5\n\t"
-	"pshufd $0x0, %%xmm5, %%xmm5\n\t"
 	"sub $0x10, %0\n\t"
 	"sub $0x10, %1\n\t"
 	"sub $0x4, %2\n\t"
@@ -3079,20 +3076,9 @@ __attribute__((noinline,__target__("sse2"))) void sse2_delta8_rgba(const uint8_t
 	"psrld $0x10, %%xmm2\n\t"
 	"pand %%xmm0, %%xmm2\n\t"
 	"paddd %%xmm2, %%xmm1\n\t"
-	"movdqa %%xmm1, %%xmm3\n\t"
-	"movdqa %%xmm1, %%xmm2\n\t"
-	"pand %%xmm0, %%xmm3\n\t"
-	"pand %%xmm0, %%xmm2\n\t"
-	"psrldq $0x9, %%xmm2\n\t"
-	"por %%xmm2, %%xmm3\n\t"
-	"movdqa %%xmm1, %%xmm2\n\t"
-	"pand %%xmm0, %%xmm2\n\t"
-	"psrldq $0x6, %%xmm2\n\t"
-	"por %%xmm2, %%xmm3\n\t"
-	"pand %%xmm0, %%xmm1\n\t"
-	"psrldq $0x3, %%xmm1\n\t"
-	"por %%xmm1, %%xmm3\n\t"
-	"movd %%xmm3, %%eax\n\t"
+	"packssdw %%xmm1, %%xmm1\n\t"
+	"packuswb %%xmm1, %%xmm1\n\t"
+	"movd %%xmm1, %%eax\n\t"
 	"movnti %%eax, (%2,%3)\n\t"
 	"sub $0x4, %3\n\t"
 	"jnz sse2_delta8_rgba_iter\n\t"
@@ -3116,9 +3102,6 @@ __attribute__((noinline,__target__("sse2"))) void sse2_delta8_bgra(const uint8_t
 	"mov $0xff, %%eax\n\t"
 	"movd %%eax, %%xmm0\n\t"
 	"pshufd $0x0, %%xmm0, %%xmm0\n\t"
-	"mov $0x80000000, %%eax\n\t"
-	"movd %%eax, %%xmm5\n\t"
-	"pshufd $0x0, %%xmm5, %%xmm5\n\t"
 	"sub $0x10, %0\n\t"
 	"sub $0x10, %1\n\t"
 	"sub $0x4, %2\n\t"
@@ -3148,20 +3131,9 @@ __attribute__((noinline,__target__("sse2"))) void sse2_delta8_bgra(const uint8_t
 	"pand %%xmm0, %%xmm2\n\t"
 	"paddd %%xmm2, %%xmm2\n\t"
 	"paddd %%xmm2, %%xmm1\n\t"
-	"movdqa %%xmm1, %%xmm3\n\t"
-	"movdqa %%xmm1, %%xmm2\n\t"
-	"pand %%xmm0, %%xmm3\n\t"
-	"pand %%xmm0, %%xmm2\n\t"
-	"psrldq $0x9, %%xmm2\n\t"
-	"por %%xmm2, %%xmm3\n\t"
-	"movdqa %%xmm1, %%xmm2\n\t"
-	"pand %%xmm0, %%xmm2\n\t"
-	"psrldq $0x6, %%xmm2\n\t"
-	"por %%xmm2, %%xmm3\n\t"
-	"pand %%xmm0, %%xmm1\n\t"
-	"psrldq $0x3, %%xmm1\n\t"
-	"por %%xmm1, %%xmm3\n\t"
-	"movd %%xmm3, %%eax\n\t"
+	"packssdw %%xmm1, %%xmm1\n\t"
+	"packuswb %%xmm1, %%xmm1\n\t"
+	"movd %%xmm1, %%eax\n\t"
 	"movnti %%eax, (%2,%3)\n\t"
 	"sub $0x4, %3\n\t"
 	"jnz sse2_delta8_bgra_iter\n\t"
@@ -3185,9 +3157,6 @@ __attribute__((noinline,__target__("sse2"))) void sse2_delta8_argb(const uint8_t
 	"mov $0xff, %%eax\n\t"
 	"movd %%eax, %%xmm0\n\t"
 	"pshufd $0x0, %%xmm0, %%xmm0\n\t"
-	"mov $0x80000000, %%eax\n\t"
-	"movd %%eax, %%xmm5\n\t"
-	"pshufd $0x0, %%xmm5, %%xmm5\n\t"
 	"sub $0x10, %0\n\t"
 	"sub $0x10, %1\n\t"
 	"sub $0x4, %2\n\t"
@@ -3218,20 +3187,9 @@ __attribute__((noinline,__target__("sse2"))) void sse2_delta8_argb(const uint8_t
 	"psrld $0x18, %%xmm2\n\t"
 	"pand %%xmm0, %%xmm2\n\t"
 	"paddd %%xmm2, %%xmm1\n\t"
-	"movdqa %%xmm1, %%xmm3\n\t"
-	"movdqa %%xmm1, %%xmm2\n\t"
-	"pand %%xmm0, %%xmm3\n\t"
-	"pand %%xmm0, %%xmm2\n\t"
-	"psrldq $0x9, %%xmm2\n\t"
-	"por %%xmm2, %%xmm3\n\t"
-	"movdqa %%xmm1, %%xmm2\n\t"
-	"pand %%xmm0, %%xmm2\n\t"
-	"psrldq $0x6, %%xmm2\n\t"
-	"por %%xmm2, %%xmm3\n\t"
-	"pand %%xmm0, %%xmm1\n\t"
-	"psrldq $0x3, %%xmm1\n\t"
-	"por %%xmm1, %%xmm3\n\t"
-	"movd %%xmm3, %%eax\n\t"
+	"packssdw %%xmm1, %%xmm1\n\t"
+	"packuswb %%xmm1, %%xmm1\n\t"
+	"movd %%xmm1, %%eax\n\t"
 	"movnti %%eax, (%2,%3)\n\t"
 	"sub $0x4, %3\n\t"
 	"jnz sse2_delta8_argb_iter\n\t"
@@ -3255,9 +3213,6 @@ __attribute__((noinline,__target__("sse2"))) void sse2_delta8_abgr(const uint8_t
 	"mov $0xff, %%eax\n\t"
 	"movd %%eax, %%xmm0\n\t"
 	"pshufd $0x0, %%xmm0, %%xmm0\n\t"
-	"mov $0x80000000, %%eax\n\t"
-	"movd %%eax, %%xmm5\n\t"
-	"pshufd $0x0, %%xmm5, %%xmm5\n\t"
 	"sub $0x10, %0\n\t"
 	"sub $0x10, %1\n\t"
 	"sub $0x4, %2\n\t"
@@ -3288,20 +3243,9 @@ __attribute__((noinline,__target__("sse2"))) void sse2_delta8_abgr(const uint8_t
 	"pand %%xmm0, %%xmm2\n\t"
 	"paddd %%xmm2, %%xmm2\n\t"
 	"paddd %%xmm2, %%xmm1\n\t"
-	"movdqa %%xmm1, %%xmm3\n\t"
-	"movdqa %%xmm1, %%xmm2\n\t"
-	"pand %%xmm0, %%xmm3\n\t"
-	"pand %%xmm0, %%xmm2\n\t"
-	"psrldq $0x9, %%xmm2\n\t"
-	"por %%xmm2, %%xmm3\n\t"
-	"movdqa %%xmm1, %%xmm2\n\t"
-	"pand %%xmm0, %%xmm2\n\t"
-	"psrldq $0x6, %%xmm2\n\t"
-	"por %%xmm2, %%xmm3\n\t"
-	"pand %%xmm0, %%xmm1\n\t"
-	"psrldq $0x3, %%xmm1\n\t"
-	"por %%xmm1, %%xmm3\n\t"
-	"movd %%xmm3, %%eax\n\t"
+	"packssdw %%xmm1, %%xmm1\n\t"
+	"packuswb %%xmm1, %%xmm1\n\t"
+	"movd %%xmm1, %%eax\n\t"
 	"movnti %%eax, (%2,%3)\n\t"
 	"sub $0x4, %3\n\t"
 	"jnz sse2_delta8_abgr_iter\n\t"