diff --git a/.idea/misc.xml b/.idea/misc.xml
new file mode 100644
index 0000000..79b3c94
--- /dev/null
+++ b/.idea/misc.xml
@@ -0,0 +1,4 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="CMakeWorkspace" PROJECT_DIR="$PROJECT_DIR$" />
+</project>
\ No newline at end of file
diff --git a/.idea/raymarcher.iml b/.idea/raymarcher.iml
deleted file mode 100644
index bc2cd87..0000000
--- a/.idea/raymarcher.iml
+++ /dev/null
@@ -1,8 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<module type="CPP_MODULE" version="4">
-  <component name="NewModuleRootManager">
-    <content url="file://$MODULE_DIR$" />
-    <orderEntry type="inheritedJdk" />
-    <orderEntry type="sourceFolder" forTests="false" />
-  </component>
-</module>
\ No newline at end of file
diff --git a/Makefile b/Makefile
index 273fe2f..311d849 100644
--- a/Makefile
+++ b/Makefile
@@ -1,8 +1,8 @@
 
 
-OPTIMIZATION=-O3
+OPTIMIZATION=-O3 -flto
 CC=gcc
-CFLAGS=-Isrc/ -lm -Wall -Wextra -pedantic-errors $(OPTIMIZATION)
+CFLAGS=-Isrc/ -lm -lpthread -Wall -Wextra -pedantic-errors $(OPTIMIZATION)
 
 .PHONY: directories
 
@@ -21,7 +21,6 @@ obj/camera.o: src/camera.c src/camera.h
 obj/images.o: images/src/images.c images/src/images.h
 	$(CC) $(CFLAGS) -c -o $@ images/src/images.c
 
-
 march: obj/camera.o obj/scene.o obj/point.o obj/images.o
 	$(CC) $(CFLAGS) -o out/march $^ marcher.c
 
diff --git a/bench.c b/bench.c
index 20bc774..138e588 100644
--- a/bench.c
+++ b/bench.c
@@ -11,10 +11,6 @@
 #include "src/point.h"
 
 
-typedef int bool;
-#define true 1
-#define false 0
-
 #define BENCH_VERSION "1.0"
 
 
@@ -28,12 +24,12 @@ typedef int bool;
 
 */
 double mandelbulb_dist(struct point pt, struct scene_object *self) {
-    int iters = self->args[0];
+    int iters = (int) self->args[0];
     double power = self->args[1];
 
     struct point z = pt;
-	float dr = 1.0;
-	float r = 0.0;
+	double dr = 1.0;
+	double r = 0.0;
 	for (int i = 0; i < iters ; i++) {
 		r = pt_length(z);
 
@@ -42,12 +38,12 @@ double mandelbulb_dist(struct point pt, struct scene_object *self) {
         }
 		
 		// convert to polar coordinates
-		float theta = acos(z.z/r);
-		float phi = atan2(z.y,z.x);
+		double theta = acos(z.z/r);
+		double phi = atan2(z.y,z.x);
 		dr =  pow(r, power-1.0)*power*dr + 1.0;
 		
 		// scale and rotate the point
-		float zr = pow(r, power);
+		double zr = pow(r, power);
 		theta = theta*power;
 		phi = phi*power;
 		
@@ -81,11 +77,11 @@ struct scene_object mandelbulb_new(struct point location, int iters, double powe
     return so;
 }
 
-int run_bench(int size, float pow, int threads, const char path[], bool save) {
-    float cam_position = 1.15;
+int run_bench(int size, double pow, int threads, const char path[], int save) {
+    double cam_position = 1.15;
     int steps = 2000;
     int iters = 1000;
-    float threshold = 0.0001;
+    double threshold = 0.0001;
 
     struct camera cam;
     cam.fov = 90;
@@ -107,7 +103,7 @@ int run_bench(int size, float pow, int threads, const char path[], bool save) {
         image_save_bmp(*img, path);
     }
 
-    image_destroy_shared(*img);
+    image_destroy(*img);
     scene_destroy(scene);
 
     return 0;  
@@ -134,11 +130,10 @@ void timer_print(struct timer t) {
     printf("\nBenchmark %s took %ldms (%.2fs)\n", t.name, time, time / 1000.0f);
 }
 
-int main(int argc, char *argv[])
+int main()
 {
-    int threads = get_nprocs();
+    int threads = get_nprocs() / 2;
     struct timer bench;
-    int size = 1080;
 
     printf("Mandelbulb Benchmark v%s\n\nDetected %d threads...\n", BENCH_VERSION, threads);
     
@@ -146,7 +141,7 @@ int main(int argc, char *argv[])
 
     bench.name = "1080px render with saving";
     timer_start(&bench);
-    run_bench(1080, 3.0, threads, "bench-pow3-1080p.bmp", true);
+    run_bench(1080, 3.0, threads, "bench-pow3-1080p.bmp", 1);
     timer_end(&bench);
     timer_print(bench);
 
@@ -154,7 +149,7 @@ int main(int argc, char *argv[])
 
     bench.name = "1080px render without saving";
     timer_start(&bench);
-    run_bench(1080, 3.0, threads, "", false);
+    run_bench(1080, 3.0, threads, "", 0);
     timer_end(&bench);
     timer_print(bench);
 
@@ -162,7 +157,7 @@ int main(int argc, char *argv[])
 
     bench.name = "10 megapixel render with saving";
     timer_start(&bench);
-    run_bench(3162, 3.0, threads, "bench-pow3-10mpx.bmp", true);
+    run_bench(3162, 3.0, threads, "bench-pow3-10mpx.bmp", 1);
     timer_end(&bench);
     timer_print(bench);
 
@@ -170,7 +165,7 @@ int main(int argc, char *argv[])
 
     bench.name = "40 megapixel render with saving";
     timer_start(&bench);
-    run_bench(6324, 3.0, threads, "bench-pow3-40mpx.bmp", true);
+    run_bench(6324, 3.0, threads, "bench-pow3-40mpx.bmp", 1);
     timer_end(&bench);
     timer_print(bench);
 
@@ -178,7 +173,7 @@ int main(int argc, char *argv[])
 
     bench.name = "1080px render single threaded without saving";
     timer_start(&bench);
-    run_bench(1080, 3.0, 1, "", false);
+    run_bench(1080, 3.0, 1, "", 0);
     timer_end(&bench);
     timer_print(bench);
 
diff --git a/marcher.c b/marcher.c
index c32562b..4a786ae 100644
--- a/marcher.c
+++ b/marcher.c
@@ -160,7 +160,7 @@ int main(int argc, char* argv[]) {
 
     image_save_bmp(*img, path);
 
-    image_destroy_shared(*img);
+    image_destroy(*img);
     scene_destroy(scene);
 
     return 0;  
diff --git a/src/camera.c b/src/camera.c
index 54a493b..7bcfe6e 100644
--- a/src/camera.c
+++ b/src/camera.c
@@ -1,11 +1,25 @@
-#include <unistd.h>
 #include <sys/wait.h>
 #include <sys/mman.h>
 #include <stdlib.h>
 #include <stdio.h>
+#include <pthread.h>
 #include "point.h"
 #include "camera.h"
 
+struct thread_args {
+    struct point start;
+    int thread_id;
+    int thread_count;
+    int height;
+    int width;
+    struct point move_up;
+    struct point move_right;
+    void (*callback)(struct point, int, int);
+};
+
+static void * camera_iterate_rays_const_dist_thread(void* args);
+
+
 struct camera camera_new(struct point direction, unsigned int fov) {
     struct camera camera;
     camera.location = (struct point) {
@@ -26,10 +40,9 @@ void camera_set_looking_at(struct camera *cam, struct point origin, struct point
     cam->direction = pt_normalize(pt_sub(thing, origin));
 }
 
-
-void camera_iterate_rays_const_dist(struct camera camera, int width, int height, int threads, void (*callback)(struct point, int, int)) {
-    // negative threads => single threaded.
-    if (threads < 0) threads = 0;
+void camera_iterate_rays_const_dist(struct camera camera, int width, int height, int thread_count, void (*callback)(struct point, int, int)) {
+    // negative thread_count => single threaded.
+    if (thread_count < 0) thread_count = 0;
 
     struct point span_z, span_xy;
 
@@ -52,46 +65,88 @@ void camera_iterate_rays_const_dist(struct camera camera, int width, int height,
     starting_point = pt_add(starting_point, pt_mult(move_right, - width / (double) 2));
     starting_point = pt_add(starting_point, pt_mult(move_up, - height / (double) 2));
 
-    // initialize threads
-    int thread_id = 0;
-    for (int i = 0; i < threads - 1; i++) {
-        if (fork() == 0) {
-            thread_id = i + 1;
-            break;
-        }
+    if (thread_count < 2) {
+        //TODO implement single threaded work here
+        struct thread_args arg = {
+                .start = starting_point,
+                .thread_id = 0,
+                .thread_count = 1,
+                .height = height,
+                .width = width,
+                .move_up = move_up,
+                .move_right = move_right,
+                .callback = callback
+        };
+
+        camera_iterate_rays_const_dist_thread(&arg);
+
+        return;
+    }
+
+    // initialize thread_count
+    pthread_t * threads = malloc(sizeof(pthread_t) * thread_count);
+    struct thread_args* args = malloc(sizeof(struct thread_args) * thread_count);
+
+    for (int i = 0; i < thread_count; i++) {
+        args[i] = (struct thread_args) {
+            .start = starting_point,
+            .thread_id = i,
+            .thread_count = thread_count,
+            .height = height,
+            .width = width,
+            .move_up = move_up,
+            .move_right = move_right,
+            .callback = callback
+        };
+
+        pthread_create(threads + i, NULL, camera_iterate_rays_const_dist_thread, (void*) (args + i));
+    }
+
+    //struct thread_args {
+    //    struct point start;
+    //    int thread_id;
+    //    int thread_count;
+    //    int height;
+    //    int width;
+    //    struct point move_up;
+    //    struct point move_right;
+    //    void (*callback)(struct point, int, int);
+    //};
+
+
+    for (int i = 0; i < thread_count; i++) {
+        pthread_join(threads[i], NULL);
     }
+}
+
+static void * camera_iterate_rays_const_dist_thread(void* arg_ptr) {
+    // explicit cast to make gcc happy
+    struct thread_args* args = (struct thread_args*) arg_ptr;
 
     // this point is moved for every pixel
-    struct point curr_pt = starting_point;
+    struct point curr_pt = args->start;
 
     // (0,0) screenspace is bottom left corner
-    for (int y = 0; y < height; y++) {
+    for (int y = 0; y < args->height; y++) {
         // move one row up (this has to be done in every thread!)
-        starting_point = pt_add(starting_point, move_up);
+        args->start = pt_add(args->start, args->move_up);
 
         // only render the lines this thread is responsible for
-        if (y % threads != thread_id) continue;
-        
+        if (y % args->thread_count != args->thread_id) continue;
+
         // display progress in percent
-        if (height > 200 && y % (height / 100) == 0 && y != 0) {
-            printf("\r%02i%%", (y * 100) / height);
+        if (args->height > 200 && y % (args->height / 100) == 0 && y != 0) {
+            printf("\r%02i%%", (y * 100) / args->height);
             fflush(stdout);
         }
 
         // actually iterate this line
-        curr_pt = starting_point;
-        for (int x = 0; x < width; x++) {
-            callback(curr_pt, x, y);
-            curr_pt = pt_add(curr_pt, move_right); // move pt right to next pt
+        curr_pt = args->start;
+        for (int x = 0; x < args->width; x++) {
+            args->callback(curr_pt, x, y);
+            curr_pt = pt_add(curr_pt, args->move_right); // move pt right to next pt
         }
     }
 
-    if (thread_id != 0) {
-        exit(0);
-    }
-
-    int status;
-    for (int i = 0; i < threads - 1; i++) {
-        while(wait(&status) > 0) {}
-    }
-}
\ No newline at end of file
+    return NULL;
+}
diff --git a/src/camera.h b/src/camera.h
index b766c3f..2fb6318 100644
--- a/src/camera.h
+++ b/src/camera.h
@@ -8,4 +8,4 @@ struct camera {
 
 struct camera camera_new(struct point direction, unsigned int fov);
 void camera_set_looking_at(struct camera *cam, struct point origin, struct point thing);
-void camera_iterate_rays_const_dist(struct camera camera, int width, int height, int threads, void (*callback)(struct point, int, int));
\ No newline at end of file
+void camera_iterate_rays_const_dist(struct camera camera, int width, int height, int thread_count, void (*callback)(struct point, int, int));
\ No newline at end of file
diff --git a/src/point.c b/src/point.c
index 219a70c..bb928f4 100644
--- a/src/point.c
+++ b/src/point.c
@@ -2,17 +2,10 @@
 #include <stdio.h>
 #include "point.h"
 
-// scale vector to length
-struct point pt_scale(struct point pt, double length) {
-    double f = length / pt_length(pt);
-    return (struct point) {
-        .x = pt.x * f,
-        .y = pt.y * f,
-        .z = pt.z * f
-    };
-}
+// get the length of vector
+inline double pt_length_inline (struct point pt) __attribute__((always_inline));
 
-struct point pt_mult(struct point pt, double scalar) {
+inline struct point pt_mult(struct point pt, double scalar) {
     return (struct point) {
         .x = pt.x * scalar,
         .y = pt.y * scalar,
@@ -21,7 +14,7 @@ struct point pt_mult(struct point pt, double scalar) {
 }
 
 // return internal angle between a and b
-double pt_angle(struct point a, struct point b) {
+inline double pt_angle(struct point a, struct point b) {
     return acos(pt_dot(
         pt_normalize(a),
         pt_normalize(b)
@@ -29,12 +22,16 @@ double pt_angle(struct point a, struct point b) {
 }
 
 // get the length of vector
-double pt_length(struct point pt) {
+inline double pt_length_inline (struct point pt) {
     return sqrt((pt.x * pt.x) + (pt.y * pt.y) + (pt.z * pt.z));
 }
 
+double pt_length(struct point pt) {
+    return pt_length_inline(pt);
+}
+
 // add the vector add to the vector pt
-struct point pt_add(struct point pt, struct point add) {
+inline struct point pt_add(struct point pt, struct point add) {
     return (struct point) {
         .x = pt.x + add.x,
         .y = pt.y + add.y,
@@ -43,7 +40,7 @@ struct point pt_add(struct point pt, struct point add) {
 }
 
 // add the vector add to the vector pt
-struct point pt_sub(struct point pt, struct point sub) {
+inline struct point pt_sub(struct point pt, struct point sub) {
     return (struct point) {
         .x = pt.x - sub.x,
         .y = pt.y - sub.y,
@@ -51,22 +48,38 @@ struct point pt_sub(struct point pt, struct point sub) {
     };
 }
 
-double pt_dist(struct point p1, struct point p2) {
-    return pt_length(pt_sub(p1, p2));
+inline double pt_dist(struct point p1, struct point p2) {
+    return pt_length_inline(pt_sub(p1, p2));
 }
 
 // normalize a vector
-struct point pt_normalize(struct point pt) {
-    return pt_scale(pt, 1);
+inline struct point pt_normalize(struct point pt) {
+    double length = pt_length(pt);
+
+    return (struct point) {
+        .x = pt.x / length,
+        .y = pt.y / length,
+        .z = pt.z / length
+    };
+}
+
+// scale vector to length
+inline struct point pt_scale(struct point pt, double length) {
+    double f = length / pt_length_inline(pt);
+    return (struct point) {
+            .x = pt.x * f,
+            .y = pt.y * f,
+            .z = pt.z * f
+    };
 }
 
 // dot product of two vectors
-double pt_dot(struct point a, struct point b) {
+inline double pt_dot(struct point a, struct point b) {
     return a.x*b.x + a.y*b.y + a.z*b.z;
 }
 
 // cross product of two vectors
-struct point pt_cross(struct point a, struct point b) {
+inline struct point pt_cross(struct point a, struct point b) {
     return (struct point) {
         .x = a.y*b.z - a.z*b.y,
         .y = a.z*b.x - a.x*b.z,
@@ -74,29 +87,29 @@ struct point pt_cross(struct point a, struct point b) {
     };
 }
 
-void pt_print(struct point pt) {
+inline void pt_print(struct point pt) {
     printf("(%f, %f, %f)\n", pt.x, pt.y, pt.z);
 }
 
-void pt_print_n(const char* name, struct point pt) {
+inline void pt_print_n(const char* name, struct point pt) {
     printf("%s: (%f, %f, %f)\n", name, pt.x, pt.y, pt.z);
 }
 
 // find two vectors that span the orthogonal plane, where 
 // span_xy is a vector lying on the xy-plane (and pointing left)
 // and span_z is orthogonal to span_xy pointing "upwards"
-void pt_orthogonal_plane(struct point pt, struct point *span_z, struct point *span_xy) {
+inline void pt_orthogonal_plane(struct point pt, struct point *span_z, struct point *span_xy) {
     pt = pt_normalize(pt);
 
     // get the vector lying on the xy axis
-    // this is done by 
-    *span_xy = pt_normalize(pt_cross((struct point){.x = 0, .y = 0, .z = 1}, pt)); // points to the "left" (of the viewing direction)
+    // this is done by
+    *span_xy = pt_normalize(pt_cross(PT_NEW(0,0,1), pt)); // points to the "left" (of the viewing direction)
 
     // now use this, to find the vector 
     *span_z = pt_normalize(pt_cross(pt, *span_xy));
 }
 
-struct point pt_mod(struct point pt, double mod) {
+inline struct point pt_mod(struct point pt, double mod) {
     return (struct point) {
         .x = fabs(fmod(pt.x, mod)),
         .y = fabs(fmod(pt.y, mod)),
diff --git a/src/point.h b/src/point.h
index 69cb378..62bde2d 100644
--- a/src/point.h
+++ b/src/point.h
@@ -8,6 +8,7 @@ struct point {
 };
 
 #define PT_ZERO ((struct point) {.x=0, .y=0, .z=0})
+#define PT_NEW(x,y,z) ((struct point){(x), (y), (z)})
 
 struct point pt_scale(struct point pt, double length);
 struct point pt_normalize(struct point pt);
diff --git a/src/scene.c b/src/scene.c
index bad4222..b62fab6 100644
--- a/src/scene.c
+++ b/src/scene.c
@@ -51,7 +51,7 @@ Image* render_scene(struct scene *scene, struct camera *camera, unsigned int thr
     current_camera= camera;
 
     // initialize shared pixel buffer
-    image_new_shared(scene->width, scene->height, current_image);
+    image_new(scene->width, scene->height, current_image);
 
     // iterate over the rays
     camera_iterate_rays_const_dist(*camera, scene->width, scene->height, threads, camera_iter_callback);