diff --git a/system/camerad/sensors/ar0231.cc b/system/camerad/sensors/ar0231.cc
index a107a6e1b9..e4ef3af404 100644
--- a/system/camerad/sensors/ar0231.cc
+++ b/system/camerad/sensors/ar0231.cc
@@ -126,8 +126,8 @@ AR0231::AR0231() {
     0x00000fbc, 0x000000bb, 0x00000009,
     0x00000fb6, 0x00000fe0, 0x000000ea,
   };
-  for (int i = 0; i < 64; i++) {
-    float fx = i / 63.0;
+  for (int i = 0; i < 65; i++) {
+    float fx = i / 64.0;
     const float gamma_k = 0.75;
     const float gamma_b = 0.125;
     const float mp = 0.01; // ideally midpoint should be adaptive
@@ -138,6 +138,7 @@ AR0231::AR0231() {
       ((rk * (fx-mp) * (gamma_k*mp+gamma_b) * (1+1/(rk*mp)) / (1-rk*(fx-mp))) + gamma_k*mp + gamma_b);
     gamma_lut_rgb.push_back((uint32_t)(fx*1023.0 + 0.5));
   }
+  prepare_gamma_lut();
   linearization_lut = {
     0x02000000, 0x02000000, 0x02000000, 0x02000000,
     0x020007ff, 0x020007ff, 0x020007ff, 0x020007ff,
diff --git a/system/camerad/sensors/os04c10.cc b/system/camerad/sensors/os04c10.cc
index 08530989fe..53ac37f08a 100644
--- a/system/camerad/sensors/os04c10.cc
+++ b/system/camerad/sensors/os04c10.cc
@@ -68,10 +68,11 @@ OS04C10::OS04C10() {
     0x00000fa7, 0x000000d9, 0x00001000,
     0x00000fca, 0x00000fef, 0x000000c7,
   };
-  for (int i = 0; i < 64; i++) {
-    float fx = i / 63.0;
+  for (int i = 0; i < 65; i++) {
+    float fx = i / 64.0;
     gamma_lut_rgb.push_back((uint32_t)(pow(fx, 0.7)*1023.0 + 0.5));
   }
+  prepare_gamma_lut();
   linearization_lut = {
     0x02000000, 0x02000000, 0x02000000, 0x02000000,
     0x020007ff, 0x020007ff, 0x020007ff, 0x020007ff,
diff --git a/system/camerad/sensors/ox03c10.cc b/system/camerad/sensors/ox03c10.cc
index 16ab0a65bc..19742661b1 100644
--- a/system/camerad/sensors/ox03c10.cc
+++ b/system/camerad/sensors/ox03c10.cc
@@ -70,11 +70,12 @@ OX03C10::OX03C10() {
     0x00000fcc, 0x000000b9, 0x00000ffb,
     0x00000fc2, 0x00000ff6, 0x000000c9,
   };
-  for (int i = 0; i < 64; i++) {
-    float fx = i / 63.0;
+  for (int i = 0; i < 65; i++) {
+    float fx = i / 64.0;
     fx = -0.507089*exp(-12.54124638*fx) + 0.9655*pow(fx, 0.5) - 0.472597*fx + 0.507089;
     gamma_lut_rgb.push_back((uint32_t)(fx*1023.0 + 0.5));
   }
+  prepare_gamma_lut();
   linearization_lut = {
     0x00200000, 0x00200000, 0x00200000, 0x00200000,
     0x00404080, 0x00404080, 0x00404080, 0x00404080,
diff --git a/system/camerad/sensors/sensor.h b/system/camerad/sensors/sensor.h
index d94cd57c8d..dc2aadfe13 100644
--- a/system/camerad/sensors/sensor.h
+++ b/system/camerad/sensors/sensor.h
@@ -72,6 +72,12 @@ public:
   uint32_t black_level;
   std::vector<uint32_t> color_correct_matrix;  // 3x3
   std::vector<uint32_t> gamma_lut_rgb;         // gamma LUTs are length 64 * sizeof(uint32_t); same for r/g/b here
+  void prepare_gamma_lut() {
+    for (int i = 0; i < 64; i++) {
+      gamma_lut_rgb[i] |= ((uint32_t)(gamma_lut_rgb[i+1] - gamma_lut_rgb[i]) << 10);
+    }
+    gamma_lut_rgb.pop_back();
+  }
   std::vector<uint32_t> linearization_lut;     // length 288
   std::vector<uint32_t> linearization_pts;     // length 4
   std::vector<uint32_t> vignetting_lut;        // 2x length 884