diff --git a/demo/mod/first_person3d.fnl b/demo/mod/first_person3d.fnl
index fefa7ec..c0051dd 100644
--- a/demo/mod/first_person3d.fnl
+++ b/demo/mod/first_person3d.fnl
@@ -349,11 +349,12 @@
             r2 (* 0.04 (math.sin (+ (* real-time 3.2) phase)))
             light-radius (* 150 (+ 0.95 r1 r2))]
         (lights:clear)
-        (lights:add light-x light-y light-z 0xFFB888 light-intensity light-radius)
+        (lights:add light-x light-y light-z 2 light-intensity light-radius)
 
         (pxl8.push_target)
         (pxl8.begin_frame_3d camera lights {
           :ambient 25
+          :dither true
           :fog_density 0.0
           :celestial_dir [0.5 -0.8 0.3]
           :celestial_intensity 0.3})
diff --git a/src/gfx/pxl8_colormap.h b/src/gfx/pxl8_colormap.h
index fb9f714..584e448 100644
--- a/src/gfx/pxl8_colormap.h
+++ b/src/gfx/pxl8_colormap.h
@@ -37,7 +37,7 @@ typedef struct {
 static const pxl8_rgb pxl8_light_colors[PXL8_LIGHT_COLORS] = {
     {255, 255, 255},
     {255, 64, 64},
-    {255, 160, 64},
+    {255, 192, 64},
     {255, 255, 64},
     {64, 255, 64},
     {64, 255, 255},
diff --git a/src/gfx/pxl8_lights.c b/src/gfx/pxl8_lights.c
index 7180e6a..de3d304 100644
--- a/src/gfx/pxl8_lights.c
+++ b/src/gfx/pxl8_lights.c
@@ -33,16 +33,16 @@ void pxl8_lights_destroy(pxl8_lights* lights) {
     pxl8_free(lights);
 }
 
-void pxl8_lights_add(pxl8_lights* lights, f32 x, f32 y, f32 z, u8 r, u8 g, u8 b, u8 intensity, f32 radius) {
-    (void)r; (void)g; (void)b;
+void pxl8_lights_add(pxl8_lights* lights, f32 x, f32 y, f32 z, u8 color, u8 intensity, f32 radius) {
     if (!lights || lights->count >= lights->capacity) return;
 
     f32 radius_sq = radius * radius;
     pxl8_light* l = &lights->data[lights->count++];
+    l->color = color;
+    l->intensity = (f32)intensity / 255.0f;
+    l->inv_radius_sq = radius_sq > 0.0f ? 1.0f / radius_sq : 0.0f;
     l->position = (pxl8_vec3){{x, y, z}};
     l->radius_sq = radius_sq;
-    l->inv_radius_sq = radius_sq > 0.0f ? 1.0f / radius_sq : 0.0f;
-    l->intensity = (f32)intensity / 255.0f;
 }
 
 void pxl8_lights_clear(pxl8_lights* lights) {
diff --git a/src/gfx/pxl8_lights.h b/src/gfx/pxl8_lights.h
index 5763410..15d861a 100644
--- a/src/gfx/pxl8_lights.h
+++ b/src/gfx/pxl8_lights.h
@@ -6,10 +6,11 @@
 #define PXL8_LIGHTS_MAX 256
 
 typedef struct pxl8_light {
+    u8 color;
+    f32 intensity;
+    f32 inv_radius_sq;
     pxl8_vec3 position;
     f32 radius_sq;
-    f32 inv_radius_sq;
-    f32 intensity;
 } pxl8_light;
 
 typedef struct pxl8_lights pxl8_lights;
@@ -21,7 +22,7 @@ extern "C" {
 pxl8_lights* pxl8_lights_create(u32 capacity);
 void pxl8_lights_destroy(pxl8_lights* lights);
 
-void pxl8_lights_add(pxl8_lights* lights, f32 x, f32 y, f32 z, u8 r, u8 g, u8 b, u8 intensity, f32 radius);
+void pxl8_lights_add(pxl8_lights* lights, f32 x, f32 y, f32 z, u8 color, u8 intensity, f32 radius);
 void pxl8_lights_clear(pxl8_lights* lights);
 u32 pxl8_lights_count(const pxl8_lights* lights);
 const pxl8_light* pxl8_lights_data(const pxl8_lights* lights);
diff --git a/src/gfx/pxl8_shader_builtins.h b/src/gfx/pxl8_shader_builtins.h
index 4efb9bd..a4eaf5f 100644
--- a/src/gfx/pxl8_shader_builtins.h
+++ b/src/gfx/pxl8_shader_builtins.h
@@ -43,11 +43,22 @@ static inline u8 pxl8_sample_indexed(const pxl8_shader_bindings* b, pxl8_vec2 uv
     }
 }
 
-static inline u8 pxl8_colormap_lookup(const pxl8_shader_bindings* b, u8 color, u8 light) {
+static inline u8 pxl8_colormap_lookup(const pxl8_shader_bindings* b, u8 color, u8 light_color, u8 intensity) {
     if (!b) return color;
     const u8* cm = b->colormap;
     if (!cm) return color;
-    u32 row = light >> 5;
+    u32 row = ((u32)light_color << 3) + (intensity >> 5);
+    return cm[(row << 8) | (u32)color];
+}
+
+static inline u8 pxl8_colormap_lookup_dithered(const pxl8_shader_bindings* b, u8 color, u8 light_color, u8 intensity, u32 x, u32 y) {
+    if (!b) return color;
+    const u8* cm = b->colormap;
+    if (!cm) return color;
+    u32 base_row = intensity >> 5;
+    u32 frac = intensity & 31;
+    u32 threshold = PXL8_BAYER_4X4[(y & 3) * 4 + (x & 3)] * 2;
+    u32 row = ((u32)light_color << 3) + (frac > threshold && base_row < 7 ? base_row + 1 : base_row);
     return cm[(row << 8) | (u32)color];
 }
 
diff --git a/src/gfx/pxl8_shader_registry.c b/src/gfx/pxl8_shader_registry.c
index 370f183..0461b4e 100644
--- a/src/gfx/pxl8_shader_registry.c
+++ b/src/gfx/pxl8_shader_registry.c
@@ -2,14 +2,14 @@
 
 #include <string.h>
 
-extern u8 pxl8_shader_lit(pxl8_shader_ctx* ctx, const pxl8_shader_bindings* bindings, const pxl8_shader_uniforms* uniforms);
-extern u8 pxl8_shader_unlit(pxl8_shader_ctx* ctx, const pxl8_shader_bindings* bindings, const pxl8_shader_uniforms* uniforms);
+extern void pxl8_shader_lit(const pxl8_shader_ctx* ctx, const pxl8_shader_bindings* bindings, const pxl8_shader_uniforms* uniforms, u8* colors_out);
+extern void pxl8_shader_unlit(const pxl8_shader_ctx* ctx, const pxl8_shader_bindings* bindings, const pxl8_shader_uniforms* uniforms, u8* colors_out);
 
 void pxl8_shader_registry_init(void) {}
 void pxl8_shader_registry_reload(void) {}
 
 pxl8_shader_fn pxl8_shader_registry_get(const char* name) {
-    if (strcmp(name, "lit") == 0) return (pxl8_shader_fn)pxl8_shader_lit;
-    if (strcmp(name, "unlit") == 0) return (pxl8_shader_fn)pxl8_shader_unlit;
+    if (strcmp(name, "lit") == 0) return pxl8_shader_lit;
+    if (strcmp(name, "unlit") == 0) return pxl8_shader_unlit;
     return NULL;
 }
diff --git a/src/gfx/shaders/cpu/lit.c b/src/gfx/shaders/cpu/lit.c
index 58aaf5a..b776fef 100644
--- a/src/gfx/shaders/cpu/lit.c
+++ b/src/gfx/shaders/cpu/lit.c
@@ -32,6 +32,8 @@ void pxl8_shader_lit(
     }
 
     pxl8_f32_simd light = ctx->v_light;
+    f32 max_strength[4] = {0, 0, 0, 0};
+    u8 dominant_color[4] = {0, 0, 0, 0};
 
     if (uniforms) {
         pxl8_f32_simd ambient = pxl8_f32_simd_set((f32)uniforms->ambient / 255.0f);
@@ -72,11 +74,32 @@ void pxl8_shader_lit(
             );
             falloff = pxl8_f32_simd_max(falloff, pxl8_f32_simd_zero());
 
+            if (uniforms->dither) {
+                f32 falloff_arr[4];
+                pxl8_f32_simd_store(falloff_arr, falloff);
+                for (u32 j = 0; j < 4; j++) {
+                    if (falloff_arr[j] < 0.5f) {
+                        f32 threshold = (PXL8_BAYER_4X4[((u32)py[j] & 3) * 4 + ((u32)px[j] & 3)] + 0.5f) * (1.0f / 16.0f);
+                        if (falloff_arr[j] < threshold * 0.5f) falloff_arr[j] = 0.0f;
+                    }
+                }
+                falloff = pxl8_f32_simd_load(falloff_arr);
+            }
+
             pxl8_f32_simd strength = pxl8_f32_simd_mul(
                 pxl8_f32_simd_mul(pxl8_f32_simd_set(l->intensity), falloff),
                 ndotl
             );
 
+            f32 strength_arr[4];
+            pxl8_f32_simd_store(strength_arr, strength);
+            for (u32 j = 0; j < 4; j++) {
+                if (strength_arr[j] > max_strength[j]) {
+                    max_strength[j] = strength_arr[j];
+                    dominant_color[j] = l->color;
+                }
+            }
+
             light = pxl8_f32_simd_add(light, strength);
         }
     }
@@ -88,13 +111,12 @@ void pxl8_shader_lit(
     pxl8_f32_simd_store(light_arr, light_f);
 
     for (u32 i = 0; i < ctx->color_count; i++) {
-        u8 light_u8;
+        u8 light_u8 = (u8)light_arr[i];
         if (uniforms && uniforms->dither) {
-            light_u8 = pxl8_gfx_dither(light_arr[i], (u32)px[i], (u32)py[i]);
+            colors_out[i] = pxl8_colormap_lookup_dithered(bindings, tex_idx[i], dominant_color[i], light_u8, (u32)px[i], (u32)py[i]);
         } else {
-            light_u8 = (u8)light_arr[i];
+            colors_out[i] = pxl8_colormap_lookup(bindings, tex_idx[i], dominant_color[i], light_u8);
         }
-        colors_out[i] = pxl8_colormap_lookup(bindings, tex_idx[i], light_u8);
     }
 
 #else
@@ -111,6 +133,8 @@ void pxl8_shader_lit(
     }
 
     f32 light = ctx->v_light;
+    f32 max_strength = 0;
+    u8 dominant_color = 0;
 
     if (uniforms) {
         f32 ambient = (f32)uniforms->ambient / 255.0f;
@@ -143,8 +167,16 @@ void pxl8_shader_lit(
 
             f32 falloff = 1.0f - dist_sq * l->inv_radius_sq;
             if (falloff <= 0.0f) continue;
+            if (uniforms->dither && falloff < 0.5f) {
+                f32 threshold = (PXL8_BAYER_4X4[((u32)ctx->y & 3) * 4 + ((u32)ctx->x & 3)] + 0.5f) * (1.0f / 16.0f);
+                if (falloff < threshold * 0.5f) continue;
+            }
 
             f32 strength = l->intensity * falloff * ndotl;
+            if (strength > max_strength) {
+                max_strength = strength;
+                dominant_color = l->color;
+            }
             light += strength;
         }
     }
@@ -155,9 +187,9 @@ void pxl8_shader_lit(
     f32 light_f = light * 255.0f;
     u8 light_u8 = (u8)light_f;
     if (uniforms && uniforms->dither) {
-        light_u8 = pxl8_gfx_dither(light_f, (u32)ctx->x, (u32)ctx->y);
+        colors_out[0] = pxl8_colormap_lookup_dithered(bindings, tex_idx, dominant_color, light_u8, (u32)ctx->x, (u32)ctx->y);
+    } else {
+        colors_out[0] = pxl8_colormap_lookup(bindings, tex_idx, dominant_color, light_u8);
     }
-
-    colors_out[0] = pxl8_colormap_lookup(bindings, tex_idx, light_u8);
 #endif
 }
diff --git a/src/lua/pxl8/effects.lua b/src/lua/pxl8/effects.lua
index f38381d..567d64c 100644
--- a/src/lua/pxl8/effects.lua
+++ b/src/lua/pxl8/effects.lua
@@ -14,16 +14,8 @@ function Lights.new(capacity)
     return setmetatable({ _ptr = ptr }, Lights)
 end
 
-function Lights:add(x, y, z, r, g, b, intensity, radius)
-    if r and r > 255 then
-        local rgb = r
-        intensity = g
-        radius = b
-        r = bit.band(bit.rshift(rgb, 16), 0xFF)
-        g = bit.band(bit.rshift(rgb, 8), 0xFF)
-        b = bit.band(rgb, 0xFF)
-    end
-    C.pxl8_lights_add(self._ptr, x, y, z, r or 255, g or 255, b or 255, intensity or 255, radius or 10)
+function Lights:add(x, y, z, color, intensity, radius)
+    C.pxl8_lights_add(self._ptr, x, y, z, color or 0, intensity or 255, radius or 10)
 end
 
 function Lights:clear()
diff --git a/src/script/pxl8_script_ffi.h b/src/script/pxl8_script_ffi.h
index 036da56..a7577a3 100644
--- a/src/script/pxl8_script_ffi.h
+++ b/src/script/pxl8_script_ffi.h
@@ -205,18 +205,17 @@ static const char* pxl8_ffi_cdefs =
 "typedef struct { float m[16]; } pxl8_mat4;\n"
 "\n"
 "typedef struct pxl8_light {\n"
-"    pxl8_vec3 position;\n"
+"    u8 color;\n"
+"    f32 intensity;\n"
 "    f32 inv_radius_sq;\n"
-"    u8 r, g, b;\n"
-"    u8 intensity;\n"
-"    f32 radius;\n"
+"    pxl8_vec3 position;\n"
 "    f32 radius_sq;\n"
 "} pxl8_light;\n"
 "\n"
 "typedef struct pxl8_lights pxl8_lights;\n"
 "pxl8_lights* pxl8_lights_create(u32 capacity);\n"
 "void pxl8_lights_destroy(pxl8_lights* lights);\n"
-"void pxl8_lights_add(pxl8_lights* lights, f32 x, f32 y, f32 z, u8 r, u8 g, u8 b, u8 intensity, f32 radius);\n"
+"void pxl8_lights_add(pxl8_lights* lights, f32 x, f32 y, f32 z, u8 color, u8 intensity, f32 radius);\n"
 "void pxl8_lights_clear(pxl8_lights* lights);\n"
 "u32 pxl8_lights_count(const pxl8_lights* lights);\n"
 "const pxl8_light* pxl8_lights_data(const pxl8_lights* lights);\n"