diff --git a/src/pixie/images.nim b/src/pixie/images.nim
index d2871e4..8212040 100644
--- a/src/pixie/images.nim
+++ b/src/pixie/images.nim
@@ -77,23 +77,25 @@ proc isOpaque*(image: Image): bool {.raises: [].} =
 
 proc flipHorizontal*(image: Image) {.raises: [].} =
   ## Flips the image around the Y axis.
-  let w = image.width div 2
+  let halfWidth = image.width div 2
   for y in 0 ..< image.height:
-    for x in 0 ..< w:
-      swap(
-        image.data[image.dataIndex(x, y)],
-        image.data[image.dataIndex(image.width - x - 1, y)]
-      )
+    var
+      left = image.dataIndex(0, y)
+      right = left + image.width - 1
+    for x in 0 ..< halfWidth:
+      swap(image.data[left], image.data[right])
+      inc left
+      dec right
 
 proc flipVertical*(image: Image) {.raises: [].} =
   ## Flips the image around the X axis.
-  let h = image.height div 2
-  for y in 0 ..< h:
+  let halfHeight = image.height div 2
+  for y in 0 ..< halfHeight:
+    let
+      topStart = image.dataIndex(0, y)
+      bottomStart = image.dataIndex(0, image.height - y - 1)
     for x in 0 ..< image.width:
-      swap(
-        image.data[image.dataIndex(x, y)],
-        image.data[image.dataIndex(x, image.height - y - 1)]
-      )
+      swap(image.data[topStart + x], image.data[bottomStart + x])
 
 proc rotate90*(image: Image) {.raises: [PixieError].} =
   ## Rotates the image 90 degrees clockwise.
diff --git a/src/pixie/simd/avx2.nim b/src/pixie/simd/avx2.nim
index 3e36f8f..b79f695 100644
--- a/src/pixie/simd/avx2.nim
+++ b/src/pixie/simd/avx2.nim
@@ -110,7 +110,7 @@ proc toPremultipliedAlphaAvx2*(data: var seq[ColorRGBA | ColorRGBX]) {.simd.} =
     oddMask = mm256_set1_epi16(0xff00)
     vec128 = mm256_set1_epi16(128)
     hiMask = mm256_set1_epi16(255 shl 8)
-    iterations = data.len div 8
+    iterations = (data.len - i) div 8
   for _ in 0 ..< iterations:
     let
       values = mm256_load_si256(cast[pointer](p))
@@ -163,7 +163,7 @@ proc invertAvx2*(image: Image) {.simd.} =
 
   let
     vec255 = mm256_set1_epi8(255)
-    iterations = image.data.len div 16
+    iterations = (image.data.len - i) div 16
   for _ in 0 ..< iterations:
     let
       a = mm256_load_si256(cast[pointer](p))
@@ -211,7 +211,7 @@ proc applyOpacityAvx2*(image: Image, opacity: float32) {.simd.} =
     div255 = mm256_set1_epi16(0x8081)
     zeroVec = mm256_setzero_si256()
     opacityVec = mm256_slli_epi16(mm256_set1_epi16(opacity), 8)
-    iterations = image.data.len div 8
+    iterations = (image.data.len - i) div 8
   for _ in 0 ..< iterations:
     let
       values = mm256_load_si256(cast[pointer](p))
@@ -257,7 +257,7 @@ proc ceilAvx2*(image: Image) {.simd.} =
   let
     vecZero = mm256_setzero_si256()
     vec255 = mm256_set1_epi8(255)
-    iterations = image.data.len div 8
+    iterations = (image.data.len - i) div 8
   for _ in 0 ..< iterations:
     var values = mm256_load_si256(cast[pointer](p))
     values = mm256_cmpeq_epi8(values, vecZero)
@@ -330,9 +330,8 @@ proc minifyBy2Avx2*(image: Image, power = 1): Image {.simd.} =
           addedOddDiv4 = mm256_srli_epi16(addedOdd, 2)
           merged = mm256_or_si256(addedEvenDiv4, mm256_slli_epi16(addedOddDiv4, 8))
           # Merged has the correct values for the next two pixels at
-          # index 0, 2, 4, 6 so mask the others out and permute into position
-          masked = mm256_and_si256(merged, mergedMask)
-          permuted = mm_256_permutevar8x32_epi32(masked, permuteControl)
+          # index 0, 2, 4, 6 so permute into position and store
+          permuted = mm_256_permutevar8x32_epi32(merged, permuteControl)
         mm_storeu_si128(
           result.data[result.dataIndex(x, y)].addr,
           mm256_castsi256_si128(permuted)
diff --git a/src/pixie/simd/neon.nim b/src/pixie/simd/neon.nim
index e28572e..6b83ee5 100644
--- a/src/pixie/simd/neon.nim
+++ b/src/pixie/simd/neon.nim
@@ -236,5 +236,30 @@ proc applyOpacityNeon*(image: Image, opacity: float32) {.simd.} =
     rgbx.a = ((rgbx.a * opacity) div 255).uint8
     image.data[i] = rgbx
 
+proc ceilNeon*(image: Image) {.simd.} =
+  var
+    i: int
+    p = cast[uint](image.data[0].addr)
+
+  let
+    zeroVec = vmovq_n_u8(0)
+    vec255 = vmovq_n_u8(255)
+    iterations = image.data.len div 4
+  for _ in 0 ..< iterations:
+    var values = vld1q_u8(cast[pointer](p))
+    values = vceqq_u8(values, zeroVec)
+    values = vbicq_u8(vec255, values)
+    vst1q_u8(cast[pointer](p), values)
+    p += 16
+  i += 4 * iterations
+
+  for i in i ..< image.data.len:
+    var rgbx = image.data[i]
+    rgbx.r = if rgbx.r == 0: 0 else: 255
+    rgbx.g = if rgbx.g == 0: 0 else: 255
+    rgbx.b = if rgbx.b == 0: 0 else: 255
+    rgbx.a = if rgbx.a == 0: 0 else: 255
+    image.data[i] = rgbx
+
 when defined(release):
   {.pop.}
diff --git a/src/pixie/simd/sse2.nim b/src/pixie/simd/sse2.nim
index 815b880..84140de 100644
--- a/src/pixie/simd/sse2.nim
+++ b/src/pixie/simd/sse2.nim
@@ -212,7 +212,7 @@ proc invertSse2*(image: Image) {.simd.} =
 
   let
     vec255 = mm_set1_epi8(255)
-    iterations = image.data.len div 16
+    iterations = (image.data.len - i) div 16
   for _ in 0 ..< iterations:
     let
       a = mm_load_si128(cast[pointer](p))
@@ -264,7 +264,7 @@ proc applyOpacitySse2*(image: Image, opacity: float32) {.simd.} =
     div255 = mm_set1_epi16(0x8081)
     zeroVec = mm_setzero_si128()
     opacityVec = mm_slli_epi16(mm_set1_epi16(opacity), 8)
-    iterations = image.data.len div 4
+    iterations = (image.data.len - i) div 4
   for _ in 0 ..< iterations:
     let values = mm_loadu_si128(cast[pointer](p))
     if mm_movemask_epi8(mm_cmpeq_epi16(values, zeroVec)) != 0xffff:
@@ -308,7 +308,7 @@ proc ceilSse2*(image: Image) {.simd.} =
   let
     vecZero = mm_setzero_si128()
     vec255 = mm_set1_epi8(255)
-    iterations = image.data.len div 8
+    iterations = (image.data.len - i) div 8
   for _ in 0 ..< iterations:
     var
       values0 = mm_loadu_si128(cast[pointer](p))
@@ -383,13 +383,10 @@ proc minifyBy2Sse2*(image: Image, power = 1): Image {.simd.} =
           addedOddDiv4 = mm_srli_epi16(addedOdd, 2)
           merged = mm_or_si128(addedEvenDiv4, mm_slli_epi16(addedOddDiv4, 8))
           # Merged has the correct values for the next two pixels at
-          # index 0 and 2 so mask the others out and shift 0 and 2 into
-          # position and store
-          masked = mm_and_si128(merged, mergedMask)
-        mm_storeu_si128(
-          result.data[result.dataIndex(x, y)].addr,
-          mm_shuffle_epi32(masked, MM_SHUFFLE(3, 3, 2, 0))
-        )
+          # index 0 and 2 so shift 0 and 2 into position and store
+          shuffled = mm_shuffle_epi32(merged, MM_SHUFFLE(3, 3, 2, 0))
+          lower = mm_cvtsi128_si64(shuffled)
+        copyMem(result.data[result.dataIndex(x, y)].addr, lower.unsafeAddr, 8)
         x += 2
 
       for x in x ..< resultEvenWidth: