adds sparse oct_tree generation in compute

build.rs

@@ -18,7 +18,7 @@ fn main() {
     println!("cargo::rerun-if-changed=shaders/rt_compute_grow_one.comp");
     println!("cargo::rerun-if-changed=shaders/rt_compute_grow_two.comp");
     println!("cargo::rerun-if-changed=shaders/rt_compute_grow_three.comp");
-    println!("cargo::rerun-if-changed=shaders/rt_compute_combine.comp");
+    println!("cargo::rerun-if-changed=shaders/rt_compute_mempos.comp");
 
     std::fs::remove_file("shaders/compiled/geo_cube.spv").unwrap_or(());
     std::fs::remove_file("shaders/compiled/frag_cube.spv").unwrap_or(());
@@ -32,7 +32,7 @@ fn main() {
     std::fs::remove_file("shaders/compiled/rt_compute_grow_one.spv").unwrap_or(());
     std::fs::remove_file("shaders/compiled/rt_compute_grow_two.spv").unwrap_or(());
     std::fs::remove_file("shaders/compiled/rt_compute_grow_three.spv").unwrap_or(());
-    std::fs::remove_file("shaders/compiled/rt_compute_combine.spv").unwrap_or(());
+    std::fs::remove_file("shaders/compiled/rt_compute_mempos.spv").unwrap_or(());
 
     if std::env::consts::OS == "windows" {
         let mut command = Command::new("./shaders/compile.bat");

shaders/compile.bat

@@ -13,4 +13,4 @@ C:/VulkanSDK/1.3.280.0/Bin/glslc.exe shaders/rt_compute_rasterize.comp -o shader
 C:/VulkanSDK/1.3.280.0/Bin/glslc.exe shaders/rt_compute_grow_one.comp -o shaders/compiled/rt_compute_grow_one.spv
 C:/VulkanSDK/1.3.280.0/Bin/glslc.exe shaders/rt_compute_grow_two.comp -o shaders/compiled/rt_compute_grow_two.spv
 C:/VulkanSDK/1.3.280.0/Bin/glslc.exe shaders/rt_compute_grow_three.comp -o shaders/compiled/rt_compute_grow_three.spv
-C:/VulkanSDK/1.3.280.0/Bin/glslc.exe shaders/rt_compute_combine.comp -o shaders/compiled/rt_compute_combine.spv
+C:/VulkanSDK/1.3.280.0/Bin/glslc.exe shaders/rt_compute_mempos.comp -o shaders/compiled/rt_compute_mempos.spv

shaders/compile.sh

@@ -14,4 +14,4 @@ glslc shaders/rt_compute_rasterize.comp -o shaders/compiled/rt_compute_rasterize
 glslc shaders/rt_compute_grow_one.comp -o shaders/compiled/rt_compute_grow_one.spv
 glslc shaders/rt_compute_grow_two.comp -o shaders/compiled/rt_compute_grow_two.spv
 glslc shaders/rt_compute_grow_three.comp -o shaders/compiled/rt_compute_grow_three.spv
-glslc shaders/rt_compute_combine.comp -o shaders/compiled/rt_compute_combine.spv
+glslc shaders/rt_compute_mempos.comp -o shaders/compiled/rt_compute_mempos.spv

shaders/rt_compute_combine.comp

@@ -1,51 +0,0 @@
-#version 450
-
-layout(binding = 0) uniform UniformBufferObject {
-    mat4 model;
-    mat4 geom_rot;
-    mat4 view;
-    mat4 proj;
-    vec3 camera_pos;
-    bool[16] use_geom_shader;
-} ubo;
-
-layout(binding = 3) readonly buffer CompoundBuffer {
-   uint compounds[];
-};
-
-layout(binding = 4) readonly buffer ColorBuffer {
-   uint grid_in[];
-};
-
-layout(binding = 9) readonly buffer TransparentBuffer {
-   bool transparent_grid[];
-};
-
-layout(binding = 8) buffer SizeBuffer3D {
-   uint grid_size_in[];
-};
-
-layout (local_size_x = 16, local_size_y = 1, local_size_z = 1) in;
-
-void main() {
-    uint index = gl_GlobalInvocationID.x;
-    uint output_offset = 0;
-    uint compound_start = 0;
-    // iterate over the compounds and find the work index inside of it
-    while (index > compounds[compound_start] * compounds[compound_start]) {
-        output_offset += compounds[compound_start] * compounds[compound_start] * compounds[compound_start] * 2;
-        index -= compounds[compound_start] * compounds[compound_start];
-        compound_start = compounds[compound_start + 2];
-    }
-    // grid pos in the task
-    uint compound_grid_size = compounds[compound_start];
-    float compound_scale = uintBitsToFloat(compounds[compound_start + 1]);
-    vec3 mid_offset = vec3(compound_scale * 0.5, compound_scale * 0.5, compound_scale * 0.5);
-    uint x = index % compound_grid_size;
-    uint y = ((index) % (compound_grid_size * compound_grid_size) - x) / (compound_grid_size);
-    uint z = (index - x - y * compound_grid_size) / (compound_grid_size * compound_grid_size);
-
-    uint size_x = grid_size_in[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + z) * 3];
-    uint size_y = grid_size_in[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + z) * 3 + 1];
-    uint size_z = grid_size_in[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + z) * 3 + 2];
-}

shaders/rt_compute_grow_one.comp

@@ -43,40 +43,11 @@ void main() {
     vec3 mid_offset = vec3(compound_scale * 0.5, compound_scale * 0.5, compound_scale * 0.5);
     uint y = index % compound_grid_size;
     uint z = (index - y) / compound_grid_size;
-    vec3 compound_pos = vec3(uintBitsToFloat(compounds[compound_start + 5]), uintBitsToFloat(compounds[compound_start + 6]), uintBitsToFloat(compounds[compound_start + 7]));
     // iterate upwards along the x axis
-    bool seen_empty = false;
-    uint start = 0;
-    uint last_col = 0;
+    uint sum = 0;
     for (uint x=0; x < compound_grid_size; x++) {
         uint color_val = grid_in[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z];
-        bool transparent = transparent_grid[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z];
-        grid_out[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z] = 0;
-        // check if we need to stop a volume
-        if (color_val != 0 && !transparent) {
-            // check if we are in a volume right now
-            if (seen_empty) {
-                // close the current volume
-                grid_out[output_offset + start * compound_grid_size * compound_grid_size + y * compound_grid_size + z] = x - start;
-                seen_empty = false;
-                last_col = 0;
-            }
-        } else {
-            // check if transparency changed
-            if (seen_empty && transparent && last_col != color_val) {
-                // if we switch colors close the current volume and prepare for a new one
-                grid_out[output_offset + start * compound_grid_size * compound_grid_size + y * compound_grid_size + z] = x - start;
-                seen_empty = false;
-            }
-            // start a new volume if we are not in one right now
-            if (!seen_empty) {
-                seen_empty = true;
-                start = x;
-                last_col = color_val;
-            }
-        }
-    }
-    if (seen_empty) {
-        grid_out[output_offset + start * compound_grid_size * compound_grid_size + y * compound_grid_size + z] = compound_grid_size - start;
+        sum += uint(color_val != 0);
+        grid_out[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z] = sum;
     }
 }

shaders/rt_compute_grow_three.comp

@@ -37,7 +37,7 @@ void main() {
     uint compound_start = 0;
     // iterate over the compounds and find the work index inside of it
     while (index > compounds[compound_start] * compounds[compound_start]) {
-        output_offset += compounds[compound_start] * compounds[compound_start] * compounds[compound_start] * 2;
+        output_offset += compounds[compound_start] * compounds[compound_start] * compounds[compound_start];
         index -= compounds[compound_start] * compounds[compound_start];
         compound_start = compounds[compound_start + 2];
     }
@@ -47,55 +47,10 @@ void main() {
     vec3 mid_offset = vec3(compound_scale * 0.5, compound_scale * 0.5, compound_scale * 0.5);
     uint x = index % compound_grid_size;
     uint y = (index - x) / compound_grid_size;
-    vec3 compound_pos = vec3(uintBitsToFloat(compounds[compound_start + 5]), uintBitsToFloat(compounds[compound_start + 6]), uintBitsToFloat(compounds[compound_start + 7]));
-    // iterate upwards along the x axis
-    bool seen_empty = false;
-    uint start = 0;
-    uint start_x_size = 0;
-    uint start_y_size = 0;
-    uint last_col = 0;
+    // iterate upwards along the z axis
+    uint sum = 0;
     for (uint z=0; z < compound_grid_size; z++) {
-        uint color_val = grid_in[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z];
-        bool transparent = transparent_grid[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z];
-        uint current_x_size = grid_size_in[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + z) * 2];
-        uint current_y_size = grid_size_in[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + z) * 2 + 1];
-        
-        grid_out[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + z) * 3] = 0;
-        grid_out[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + z) * 3 + 1] = 0;
-        grid_out[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + z) * 3 + 2] = 0;
-        // check if we need to stop a volume
-        if (color_val != 0 && !transparent) {
-            // check if we are in a volume right now
-            if (seen_empty) {
-                // close the current volume
-                grid_out[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + start) * 3] = start_x_size;
-                grid_out[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + start) * 3 + 1] = start_y_size;
-                grid_out[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + start) * 3 + 2] = z - start;
-                seen_empty = false;
-                last_col = 0;
-            }
-        } else {
-            // check if transparency changed
-            if (seen_empty && ((transparent && last_col != color_val) || (start_x_size != current_x_size) || (start_y_size != current_y_size))) {
-                // if we switch colors or size close the current volume and prepare for a new one
-                grid_out[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + start) * 3] = start_x_size;
-                grid_out[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + start) * 3 + 1] = start_y_size;
-                grid_out[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + start) * 3 + 2] = z - start;
-                seen_empty = false;
-            }
-            // start a new volume if we are not in one right now
-            if (!seen_empty && current_x_size != 0 && current_y_size != 0) {
-                seen_empty = true;
-                start = z;
-                start_x_size = current_x_size;
-                start_y_size = current_y_size;
-                last_col = color_val;
-            }
-        }
-    }
-    if (seen_empty) {
-        grid_out[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + start) * 3] = start_x_size;
-        grid_out[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + start) * 3 + 1] = start_y_size;
-        grid_out[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + start) * 3 + 2] = compound_grid_size - start;
+        sum += grid_size_in[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z];
+        grid_out[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z] = sum;
     }
 }

shaders/rt_compute_grow_two.comp

@@ -37,7 +37,7 @@ void main() {
     uint compound_start = 0;
     // iterate over the compounds and find the work index inside of it
     while (index > compounds[compound_start] * compounds[compound_start]) {
-        output_offset += compounds[compound_start] * compounds[compound_start] * compounds[compound_start] * 2;
+        output_offset += compounds[compound_start] * compounds[compound_start] * compounds[compound_start];
         index -= compounds[compound_start] * compounds[compound_start];
         compound_start = compounds[compound_start + 2];
     }
@@ -47,48 +47,10 @@ void main() {
     vec3 mid_offset = vec3(compound_scale * 0.5, compound_scale * 0.5, compound_scale * 0.5);
     uint x = index % compound_grid_size;
     uint z = (index - x) / compound_grid_size;
-    vec3 compound_pos = vec3(uintBitsToFloat(compounds[compound_start + 5]), uintBitsToFloat(compounds[compound_start + 6]), uintBitsToFloat(compounds[compound_start + 7]));
-    // iterate upwards along the x axis
-    bool seen_empty = false;
-    uint start = 0;
-    uint start_x_size = 0;
-    uint last_col = 0;
+    // iterate upwards along the y axis
+    uint sum = 0;
     for (uint y=0; y < compound_grid_size; y++) {
-        uint color_val = grid_in[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z];
-        bool transparent = transparent_grid[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z];
-        uint current_x_size = grid_size_in[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z];
-        
-        grid_out[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + z) * 2] = 0;
-        grid_out[output_offset + (x * compound_grid_size * compound_grid_size + y * compound_grid_size + z) * 2 + 1] = 0;
-        // check if we need to stop a volume
-        if (color_val != 0 && !transparent) {
-            // check if we are in a volume right now
-            if (seen_empty) {
-                // close the current volume
-                grid_out[output_offset + (x * compound_grid_size * compound_grid_size + start * compound_grid_size + z) * 2] = start_x_size;
-                grid_out[output_offset + (x * compound_grid_size * compound_grid_size + start * compound_grid_size + z) * 2 + 1] = y - start;
-                seen_empty = false;
-                last_col = 0;
-            }
-        } else {
-            // check if transparency changed
-            if (seen_empty && ((transparent && last_col != color_val) || (start_x_size != current_x_size))) {
-                // if we switch colors or size close the current volume and prepare for a new one
-                grid_out[output_offset + (x * compound_grid_size * compound_grid_size + start * compound_grid_size + z) * 2] = start_x_size;
-                grid_out[output_offset + (x * compound_grid_size * compound_grid_size + start * compound_grid_size + z) * 2 + 1] = y - start;
-                seen_empty = false;
-            }
-            // start a new volume if we are not in one right now
-            if (!seen_empty && current_x_size != 0) {
-                seen_empty = true;
-                start = y;
-                start_x_size = current_x_size;
-                last_col = color_val;
-            }
-        }
-    }
-    if (seen_empty) {
-        grid_out[output_offset + (x * compound_grid_size * compound_grid_size + start * compound_grid_size + z) * 2] = start_x_size;
-        grid_out[output_offset + (x * compound_grid_size * compound_grid_size + start * compound_grid_size + z) * 2 + 1] = compound_grid_size - start;
+        sum += grid_size_in[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z];
+        grid_out[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z] = sum;
     }
 }

shaders/rt_compute_mempos.comp

@@ -0,0 +1,349 @@
+#version 450
+
+layout(binding = 0) uniform UniformBufferObject {
+    mat4 model;
+    mat4 geom_rot;
+    mat4 view;
+    mat4 proj;
+    vec3 camera_pos;
+    bool[16] use_geom_shader;
+} ubo;
+
+layout(binding = 3) readonly buffer CompoundBuffer {
+   uint compounds[];
+};
+
+layout(binding = 4) readonly buffer ColorBuffer {
+   uint grid_in[];
+};
+
+layout(binding = 9) readonly buffer TransparentBuffer {
+   bool transparent_grid[];
+};
+
+layout(binding = 8) readonly buffer SizeBuffer3D {
+   uint grid_size_in[];
+};
+
+layout(binding = 10) buffer OutMemory {
+   uint out_memory[];
+};
+
+layout(binding = 2) readonly buffer SceneInfoBuffer{
+     uint infos[]; 
+} scene_info;
+
+uint max_num_lights = scene_info.infos[0];
+
+layout (local_size_x = 16, local_size_y = 1, local_size_z = 1) in;
+
+uint num_nodes(uint size) {
+   uint nodes = 0;
+   uint add_size = 1;
+   while (size >= 2) {
+      nodes += add_size;
+      add_size = add_size * 8;
+      size = size / 2;
+   }
+
+   return nodes;
+}
+
+layout(binding = 5) buffer SizedVertices {
+   float vertices[];
+};
+
+layout(binding = 6) buffer Indices {
+   uint indices[];
+};
+
+vec3 unpack_color(uint val) {
+    // left most 8 bits first
+    uint val1 = (val >> 24);
+    uint val2 = (val << 8) >> 24;
+    uint val3 = (val << 16) >> 24;
+    uint val4 = (val << 24) >> 24;
+
+    return vec3(val4 / 255.0, val3 / 255.0, val2 / 255.0);
+}
+
+void add_cube(uint cube_num, float scale, vec3 pos, vec3 color) {
+    // add node info for the cube
+    //vertice 0
+    vertices[(cube_num * 8 + 0) * 11 + 0] = pos.x - 0.5 * scale;
+    vertices[(cube_num * 8 + 0) * 11 + 1] = pos.y + 0.5 * scale;
+    vertices[(cube_num * 8 + 0) * 11 + 2] = pos.z + 0.5 * scale;
+
+    vertices[(cube_num * 8 + 0) * 11 + 3] = color.x;
+    vertices[(cube_num * 8 + 0) * 11 + 4] = color.y;
+    vertices[(cube_num * 8 + 0) * 11 + 5] = color.z;
+
+    //vertice 1
+    vertices[(cube_num * 8 + 1) * 11 + 0] = pos.x + 0.5 * scale;
+    vertices[(cube_num * 8 + 1) * 11 + 1] = pos.y + 0.5 * scale;
+    vertices[(cube_num * 8 + 1) * 11 + 2] = pos.z + 0.5 * scale;
+
+    vertices[(cube_num * 8 + 1) * 11 + 3] = color.x;
+    vertices[(cube_num * 8 + 1) * 11 + 4] = color.y;
+    vertices[(cube_num * 8 + 1) * 11 + 5] = color.z;
+
+    //vertice 2
+    vertices[(cube_num * 8 + 2) * 11 + 0] = pos.x - 0.5 * scale;
+    vertices[(cube_num * 8 + 2) * 11 + 1] = pos.y - 0.5 * scale;
+    vertices[(cube_num * 8 + 2) * 11 + 2] = pos.z + 0.5 * scale;
+
+    vertices[(cube_num * 8 + 2) * 11 + 3] = color.x;
+    vertices[(cube_num * 8 + 2) * 11 + 4] = color.y;
+    vertices[(cube_num * 8 + 2) * 11 + 5] = color.z;
+
+    //vertice 3
+    vertices[(cube_num * 8 + 3) * 11 + 0] = pos.x + 0.5 * scale;
+    vertices[(cube_num * 8 + 3) * 11 + 1] = pos.y - 0.5 * scale;
+    vertices[(cube_num * 8 + 3) * 11 + 2] = pos.z + 0.5 * scale;
+
+    vertices[(cube_num * 8 + 3) * 11 + 3] = color.x;
+    vertices[(cube_num * 8 + 3) * 11 + 4] = color.y;
+    vertices[(cube_num * 8 + 3) * 11 + 5] = color.z;
+
+    //vertice 4
+    vertices[(cube_num * 8 + 4) * 11 + 0] = pos.x - 0.5 * scale;
+    vertices[(cube_num * 8 + 4) * 11 + 1] = pos.y + 0.5 * scale;
+    vertices[(cube_num * 8 + 4) * 11 + 2] = pos.z - 0.5 * scale;
+
+    vertices[(cube_num * 8 + 4) * 11 + 3] = color.x;
+    vertices[(cube_num * 8 + 4) * 11 + 4] = color.y;
+    vertices[(cube_num * 8 + 4) * 11 + 5] = color.z;
+
+    //vertice 5
+    vertices[(cube_num * 8 + 5) * 11 + 0] = pos.x + 0.5 * scale;
+    vertices[(cube_num * 8 + 5) * 11 + 1] = pos.y + 0.5 * scale;
+    vertices[(cube_num * 8 + 5) * 11 + 2] = pos.z - 0.5 * scale;
+
+    vertices[(cube_num * 8 + 5) * 11 + 3] = color.x;
+    vertices[(cube_num * 8 + 5) * 11 + 4] = color.y;
+    vertices[(cube_num * 8 + 5) * 11 + 5] = color.z;
+    
+    //vertice 6
+    vertices[(cube_num * 8 + 6) * 11 + 0] = pos.x - 0.5 * scale;
+    vertices[(cube_num * 8 + 6) * 11 + 1] = pos.y - 0.5 * scale;
+    vertices[(cube_num * 8 + 6) * 11 + 2] = pos.z - 0.5 * scale;
+
+    vertices[(cube_num * 8 + 6) * 11 + 3] = color.x;
+    vertices[(cube_num * 8 + 6) * 11 + 4] = color.y;
+    vertices[(cube_num * 8 + 6) * 11 + 5] = color.z;
+
+    //vertice 7
+    vertices[(cube_num * 8 + 7) * 11 + 0] = pos.x + 0.5 * scale;
+    vertices[(cube_num * 8 + 7) * 11 + 1] = pos.y - 0.5 * scale;
+    vertices[(cube_num * 8 + 7) * 11 + 2] = pos.z - 0.5 * scale;
+
+    vertices[(cube_num * 8 + 7) * 11 + 3] = color.x;
+    vertices[(cube_num * 8 + 7) * 11 + 4] = color.y;
+    vertices[(cube_num * 8 + 7) * 11 + 5] = color.z;
+
+    //add indices for the cube
+    //top
+    indices[cube_num * 36 + 0] = cube_num * 8 + 3;
+    indices[cube_num * 36 + 1] = cube_num * 8 + 0;
+    indices[cube_num * 36 + 2] = cube_num * 8 + 2;
+
+    indices[cube_num * 36 + 3] = cube_num * 8 + 3;
+    indices[cube_num * 36 + 4] = cube_num * 8 + 1;
+    indices[cube_num * 36 + 5] = cube_num * 8 + 0;
+
+    //bottom
+    indices[cube_num * 36 + 6] = cube_num * 8 + 6;
+    indices[cube_num * 36 + 7] = cube_num * 8 + 4;
+    indices[cube_num * 36 + 8] = cube_num * 8 + 7;
+
+    indices[cube_num * 36 + 9] = cube_num * 8 + 4;
+    indices[cube_num * 36 + 10] = cube_num * 8 + 5;
+    indices[cube_num * 36 + 11] = cube_num * 8 + 7;
+
+    //left
+    indices[cube_num * 36 + 12] = cube_num * 8 + 0;
+    indices[cube_num * 36 + 13] = cube_num * 8 + 4;
+    indices[cube_num * 36 + 14] = cube_num * 8 + 2;
+
+    indices[cube_num * 36 + 15] = cube_num * 8 + 6;
+    indices[cube_num * 36 + 16] = cube_num * 8 + 2;
+    indices[cube_num * 36 + 17] = cube_num * 8 + 4;
+
+    //right
+    indices[cube_num * 36 + 18] = cube_num * 8 + 1;
+    indices[cube_num * 36 + 19] = cube_num * 8 + 3;
+    indices[cube_num * 36 + 20] = cube_num * 8 + 5;
+
+    indices[cube_num * 36 + 21] = cube_num * 8 + 5;
+    indices[cube_num * 36 + 22] = cube_num * 8 + 3;
+    indices[cube_num * 36 + 23] = cube_num * 8 + 7;
+
+    //near
+    indices[cube_num * 36 + 24] = cube_num * 8 + 6;
+    indices[cube_num * 36 + 25] = cube_num * 8 + 3;
+    indices[cube_num * 36 + 26] = cube_num * 8 + 2;
+
+    indices[cube_num * 36 + 27] = cube_num * 8 + 3;
+    indices[cube_num * 36 + 28] = cube_num * 8 + 6;
+    indices[cube_num * 36 + 29] = cube_num * 8 + 7;
+
+    //far
+    indices[cube_num * 36 + 30] = cube_num * 8 + 0;
+    indices[cube_num * 36 + 31] = cube_num * 8 + 1;
+    indices[cube_num * 36 + 32] = cube_num * 8 + 4;
+
+    indices[cube_num * 36 + 33] = cube_num * 8 + 5;
+    indices[cube_num * 36 + 34] = cube_num * 8 + 4;
+    indices[cube_num * 36 + 35] = cube_num * 8 + 1;
+
+}
+
+void main() {
+   uint index = gl_GlobalInvocationID.x;
+   uint output_offset = 1;
+   uint input_offset = 0;
+   uint compound_start = 0;
+
+   uint nodes = num_nodes(compounds[compound_start]);
+   // iterate over the compounds and find the work index inside of it
+   while (index > nodes) {
+      output_offset += nodes * 9;
+      input_offset += compounds[compound_start] * compounds[compound_start] * compounds[compound_start];
+      index -= nodes;
+      compound_start = compounds[compound_start + 2];
+      nodes = num_nodes(compounds[compound_start]);
+   }
+
+   uint compound_grid_size = compounds[compound_start];
+   uint parent_start = 0;
+   uint cohort_start = 0;
+   uint cohort_index = index;
+   uint size = compounds[compound_start];
+   nodes = 0;
+   uint add_size = 1;
+   while (cohort_index >= add_size) {
+      nodes += add_size;
+      cohort_index -= add_size;
+      parent_start = cohort_start;
+      cohort_start = nodes * 9;
+      add_size *= 8;
+      size = size / 2;
+   }
+
+   uint parent = parent_start + ((cohort_index - cohort_index % 8) / 8) * 9;
+   uint start = cohort_start + 9 * cohort_index;
+   uint steps = compounds[compound_start] / size;
+
+   float compound_scale = uintBitsToFloat(compounds[compound_start + 1]);
+   vec3 mid_offset = vec3(compound_scale * 0.5, compound_scale * 0.5, compound_scale * 0.5);
+   
+   uint x = (cohort_index % steps) * size;
+   uint y = (((cohort_index - (cohort_index % steps)) % (steps * steps)) / (steps)) * size;
+   uint z = (((cohort_index - (cohort_index % (steps * steps)))) / (steps * steps)) * size;
+
+   // plus one size offset, since we want to place the nodes at the far end. This aligns with the iteration directions in the previous shaders
+   x = x  + (size - 1);
+   y = y  + (size - 1);
+   z = z  + (size - 1);
+
+   // sum of all elements with coordinates lower than x, y, z
+   uint contained_entries = grid_size_in[input_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z];
+   if (z > size) {
+      // remove contained from z neighbor
+      contained_entries = contained_entries - grid_size_in[input_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z - size];
+   }
+
+   if (y > size) {
+      if (z > size) {
+         // add back the section we will remove twice
+         contained_entries = contained_entries + int(grid_size_in[input_offset + x * compound_grid_size * compound_grid_size + (y - size) * compound_grid_size + z - size]);
+      }
+      // remove contained from y neighbor
+      contained_entries = contained_entries - int(grid_size_in[input_offset + x * compound_grid_size * compound_grid_size + (y - size) * compound_grid_size + z]);
+   }
+
+   if (x > size) {
+      if (z > size) {
+         // add the portion already removed through the z neighbor
+         contained_entries = contained_entries + grid_size_in[input_offset + (x - size) * compound_grid_size * compound_grid_size + y * compound_grid_size + z - size];
+      }
+
+      if (y > size) {
+         // add the portion already removed by the y neighbor
+         contained_entries = contained_entries + grid_size_in[input_offset + (x - size) * compound_grid_size * compound_grid_size + (y - size) * compound_grid_size + z];
+
+         if (z > size) {
+            // remove the portion already added through the z neighbor
+            contained_entries = contained_entries - grid_size_in[input_offset + (x - size) * compound_grid_size * compound_grid_size + (y - size) * compound_grid_size + z - size];
+         }
+      }
+
+      // remove contained from x neighbor
+      contained_entries = contained_entries - grid_size_in[input_offset + (x - size) * compound_grid_size * compound_grid_size + y * compound_grid_size + z];
+   }
+
+   if (contained_entries > 0) {
+      out_memory[output_offset + cohort_start + cohort_index * 9 + 0] = parent * uint(size != 64);
+
+      if (size > 2) {
+         // add child node reference
+         uint cohort_end = cohort_start + 9 * add_size;
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 1] = cohort_end + cohort_index * 9 + 9 * 0; // xyz
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 2] = cohort_end + cohort_index * 9 + 9 * 1; // Xyz
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 3] = cohort_end + cohort_index * 9 + 9 * 2; // xYz
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 4] = cohort_end + cohort_index * 9 + 9 * 3; // XYz
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 5] = cohort_end + cohort_index * 9 + 9 * 4; // xyZ
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 6] = cohort_end + cohort_index * 9 + 9 * 5; // XyZ
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 7] = cohort_end + cohort_index * 9 + 9 * 6; // xYZ
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 8] = cohort_end + cohort_index * 9 + 9 * 7; // XYZ
+
+      } else {
+         // copy color values and add cubes to rendering
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 1] = grid_in[input_offset + (x - 1) * compound_grid_size * compound_grid_size + (y - 1) * compound_grid_size + (z - 1)]; // xyz
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 2] = grid_in[input_offset + (x - 0) * compound_grid_size * compound_grid_size + (y - 1) * compound_grid_size + (z - 1)]; // Xyz
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 3] = grid_in[input_offset + (x - 1) * compound_grid_size * compound_grid_size + (y - 0) * compound_grid_size + (z - 1)]; // xYz
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 4] = grid_in[input_offset + (x - 0) * compound_grid_size * compound_grid_size + (y - 0) * compound_grid_size + (z - 1)]; // XYz
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 5] = grid_in[input_offset + (x - 1) * compound_grid_size * compound_grid_size + (y - 1) * compound_grid_size + (z - 0)]; // xyZ
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 6] = grid_in[input_offset + (x - 0) * compound_grid_size * compound_grid_size + (y - 1) * compound_grid_size + (z - 0)]; // XyZ
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 7] = grid_in[input_offset + (x - 1) * compound_grid_size * compound_grid_size + (y - 0) * compound_grid_size + (z - 0)]; // xYZ
+         out_memory[output_offset + cohort_start + cohort_index * 9 + 8] = grid_in[input_offset + (x - 0) * compound_grid_size * compound_grid_size + (y - 0) * compound_grid_size + (z - 0)]; // XYZ
+      
+         vec3 compound_pos = vec3(uintBitsToFloat(compounds[compound_start + 5]), uintBitsToFloat(compounds[compound_start + 6]), uintBitsToFloat(compounds[compound_start + 7]));
+         vec3 check_pos = compound_pos + vec3(float(x) * compound_scale, float(y) * compound_scale, float(z) * compound_scale) + mid_offset;
+         if (out_memory[output_offset + cohort_start + cohort_index * 9 + 1] != 0) {
+            add_cube(input_offset + (z - 1) * compound_grid_size * compound_grid_size + (y - 1) * compound_grid_size + (x - 1), compound_scale, check_pos - vec3(1.0, 1.0, 1.0) * compound_scale, unpack_color(out_memory[output_offset + cohort_start + cohort_index * 9 + 1]));
+         }
+         if (out_memory[output_offset + cohort_start + cohort_index * 9 + 2] != 0) {
+            add_cube(input_offset + (z - 1) * compound_grid_size * compound_grid_size + (y - 1) * compound_grid_size + (x - 0), compound_scale, check_pos - vec3(0.0, 1.0, 1.0) * compound_scale, unpack_color(out_memory[output_offset + cohort_start + cohort_index * 9 + 2]));
+         }
+         if (out_memory[output_offset + cohort_start + cohort_index * 9 + 3] != 0) {
+            add_cube(input_offset + (z - 1) * compound_grid_size * compound_grid_size + (y - 0) * compound_grid_size + (x - 1), compound_scale, check_pos - vec3(1.0, 0.0, 1.0) * compound_scale, unpack_color(out_memory[output_offset + cohort_start + cohort_index * 9 + 3]));
+         }
+         if (out_memory[output_offset + cohort_start + cohort_index * 9 + 4] != 0) {
+            add_cube(input_offset + (z - 1) * compound_grid_size * compound_grid_size + (y - 0) * compound_grid_size + (x - 0), compound_scale, check_pos - vec3(0.0, 0.0, 1.0) * compound_scale, unpack_color(out_memory[output_offset + cohort_start + cohort_index * 9 + 4]));
+         }
+         if (out_memory[output_offset + cohort_start + cohort_index * 9 + 5] != 0) {
+            add_cube(input_offset + (z - 0) * compound_grid_size * compound_grid_size + (y - 1) * compound_grid_size + (x - 1), compound_scale, check_pos - vec3(1.0, 1.0, 0.0) * compound_scale, unpack_color(out_memory[output_offset + cohort_start + cohort_index * 9 + 5]));
+         }
+         if (out_memory[output_offset + cohort_start + cohort_index * 9 + 6] != 0) {
+            add_cube(input_offset + (z - 0) * compound_grid_size * compound_grid_size + (y - 1) * compound_grid_size + (x - 0), compound_scale, check_pos - vec3(0.0, 1.0, 0.0) * compound_scale, unpack_color(out_memory[output_offset + cohort_start + cohort_index * 9 + 6]));
+         }
+         if (out_memory[output_offset + cohort_start + cohort_index * 9 + 7] != 0) {
+            add_cube(input_offset + (z - 0) * compound_grid_size * compound_grid_size + (y - 0) * compound_grid_size + (x - 1), compound_scale, check_pos - vec3(1.0, 0.0, 0.0) * compound_scale, unpack_color(out_memory[output_offset + cohort_start + cohort_index * 9 + 7]));
+         }
+         if (out_memory[output_offset + cohort_start + cohort_index * 9 + 8] != 0) {
+            add_cube(input_offset + (z - 0) * compound_grid_size * compound_grid_size + (y - 0) * compound_grid_size + (x - 0), compound_scale, check_pos - vec3(0.0, 0.0, 0.0) * compound_scale, unpack_color(out_memory[output_offset + cohort_start + cohort_index * 9 + 8]));
+         }
+      }
+   } else {
+      out_memory[output_offset + cohort_start + cohort_index * 9 + 0] = 0;
+      out_memory[output_offset + cohort_start + cohort_index * 9 + 1] = 0;
+      out_memory[output_offset + cohort_start + cohort_index * 9 + 2] = 0;
+      out_memory[output_offset + cohort_start + cohort_index * 9 + 3] = 0;
+      out_memory[output_offset + cohort_start + cohort_index * 9 + 4] = 0;
+      out_memory[output_offset + cohort_start + cohort_index * 9 + 5] = 0;
+      out_memory[output_offset + cohort_start + cohort_index * 9 + 6] = 0;
+      out_memory[output_offset + cohort_start + cohort_index * 9 + 7] = 0;
+      out_memory[output_offset + cohort_start + cohort_index * 9 + 8] = 0;
+   }
+}

shaders/rt_compute_rasterize.comp

@@ -349,7 +349,7 @@ void main() {
         if (render) {
             grid[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z] = color_int;
             transparent_grid[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z] = transparent;
-            add_cube(output_offset + index * compound_grid_size + z, compound_scale, check_pos, color);
+            //add_cube(output_offset + index * compound_grid_size + z, compound_scale, check_pos, color);
         } else {
             grid[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z] = 0;
             transparent_grid[output_offset + x * compound_grid_size * compound_grid_size + y * compound_grid_size + z] = false;

src/app_data.rs

@@ -27,7 +27,7 @@ pub struct AppData {
     pub pipeline_compute_grow_one: vk::Pipeline,
     pub pipeline_compute_grow_two: vk::Pipeline,
     pub pipeline_compute_grow_three: vk::Pipeline,
-    pub pipeline_compute_combine: vk::Pipeline,
+    pub pipeline_compute_mempos: vk::Pipeline,
 
     pub framebuffers: Vec<vk::Framebuffer>,
     pub command_pool: vk::CommandPool,
@@ -60,6 +60,9 @@ pub struct AppData {
     pub compute_out_storage_buffers_size_three: Vec<vk::Buffer>,
     pub compute_out_storage_buffers_memory_size_three: Vec<vk::DeviceMemory>,
 
+    pub compute_out_storage_buffers_oct_tree: Vec<vk::Buffer>,
+    pub compute_out_storage_buffers_memory_oct_tree: Vec<vk::DeviceMemory>,
+
     pub compute_out_cuboid_buffers: Vec<vk::Buffer>,
     pub compute_out_cuboid_buffers_memory: Vec<vk::DeviceMemory>,
 
@@ -92,6 +95,8 @@ pub struct AppData {
     pub compute_task_one_size: usize,
     pub compute_task_one_out_buffer_size: u64,
     pub compute_task_one_out_size: u64,
+    pub compute_task_oct_tree_size: u64,
+    pub compute_task_oct_tree_nodes: u64,
     // values passed to shader
     pub num_lights_per_volume: u32,
     pub min_light_weight: f32,

src/buffer.rs

@@ -206,7 +206,7 @@ pub unsafe fn create_descriptor_set_layout(
         .binding(2)
         .descriptor_type(vk::DescriptorType::STORAGE_BUFFER)
         .descriptor_count(1)
-        .stage_flags(vk::ShaderStageFlags::FRAGMENT);
+        .stage_flags(vk::ShaderStageFlags::FRAGMENT | vk::ShaderStageFlags::COMPUTE);
 
     let storage_binding_compute_in = vk::DescriptorSetLayoutBinding::builder()
         .binding(3)
@@ -250,7 +250,13 @@ pub unsafe fn create_descriptor_set_layout(
         .descriptor_count(1)
         .stage_flags(vk::ShaderStageFlags::FRAGMENT | vk::ShaderStageFlags::COMPUTE);
 
-    let bindings = &[ubo_binding, sampler_binding, storage_binding_render, storage_binding_compute_in, storage_binding_compute_out_color, storage_binding_compute_cuboid_out, storage_binding_compute_cuboid_index_out, storage_binding_compute_out_size_two, storage_binding_compute_out_size_three, storage_binding_compute_out_size_transparent];
+    let storage_binding_compute_out_oct_tree = vk::DescriptorSetLayoutBinding::builder()
+        .binding(10)
+        .descriptor_type(vk::DescriptorType::STORAGE_BUFFER)
+        .descriptor_count(1)
+        .stage_flags(vk::ShaderStageFlags::FRAGMENT | vk::ShaderStageFlags::COMPUTE);
+
+    let bindings = &[ubo_binding, sampler_binding, storage_binding_render, storage_binding_compute_in, storage_binding_compute_out_color, storage_binding_compute_cuboid_out, storage_binding_compute_cuboid_index_out, storage_binding_compute_out_size_two, storage_binding_compute_out_size_three, storage_binding_compute_out_size_transparent, storage_binding_compute_out_oct_tree];
     let info = vk::DescriptorSetLayoutCreateInfo::builder()
         .bindings(bindings);
     
@@ -311,6 +317,9 @@ pub unsafe fn create_storage_buffers(
     data.compute_out_cuboid_index_buffers.clear();
     data.compute_out_cuboid_index_buffers_memory.clear();
 
+    data.compute_out_storage_buffers_oct_tree.clear();
+    data.compute_out_storage_buffers_memory_oct_tree.clear();
+
     for _ in 0..data.swapchain_images.len() {
         let (storage_buffer, storage_buffer_memory) = create_buffer(
             instance,
@@ -363,7 +372,7 @@ pub unsafe fn create_storage_buffers(
             instance,
             device,
             data,
-            (size_of::<u32>() * 2) as u64 * data.compute_task_one_out_buffer_size.max(1),
+            (size_of::<u32>()) as u64 * data.compute_task_one_out_buffer_size.max(1),
             vk::BufferUsageFlags::STORAGE_BUFFER,
             vk::MemoryPropertyFlags::DEVICE_LOCAL,
         )?;
@@ -375,7 +384,7 @@ pub unsafe fn create_storage_buffers(
             instance,
             device,
             data,
-            (size_of::<u32>() * 3) as u64 * data.compute_task_one_out_buffer_size.max(1),
+            (size_of::<u32>()) as u64 * data.compute_task_one_out_buffer_size.max(1),
             vk::BufferUsageFlags::STORAGE_BUFFER,
             vk::MemoryPropertyFlags::DEVICE_LOCAL,
         )?;
@@ -383,6 +392,18 @@ pub unsafe fn create_storage_buffers(
         data.compute_out_storage_buffers_size_three.push(storage_buffer);
         data.compute_out_storage_buffers_memory_size_three.push(storage_buffer_memory);
 
+        let (storage_buffer, storage_buffer_memory) = create_buffer(
+            instance,
+            device,
+            data,
+            (size_of::<u32>()) as u64 * data.compute_task_oct_tree_size.max(1),
+            vk::BufferUsageFlags::STORAGE_BUFFER,
+            vk::MemoryPropertyFlags::DEVICE_LOCAL,
+        )?;
+
+        data.compute_out_storage_buffers_oct_tree.push(storage_buffer);
+        data.compute_out_storage_buffers_memory_oct_tree.push(storage_buffer_memory);
+
         let (storage_buffer, storage_buffer_memory) = create_buffer(
             instance,
             device,
@@ -551,8 +572,12 @@ pub unsafe fn create_descriptor_pool(device: &Device, data: &mut app_data::AppDa
     let compute_out_storage_transparent_size = vk::DescriptorPoolSize::builder()
         .type_(vk::DescriptorType::STORAGE_BUFFER)
         .descriptor_count(data.swapchain_images.len() as u32);
+
+    let compute_out_storage_oct_tree = vk::DescriptorPoolSize::builder()
+        .type_(vk::DescriptorType::STORAGE_BUFFER)
+        .descriptor_count(data.swapchain_images.len() as u32);
     
-    let pool_sizes = &[ubo_size, sampler_size, render_storage_size, compute_in_storage_size, compute_out_storage_color_size, compute_out_cuboid_size, compute_out_cuboid_index_size, compute_out_storage_size_two_size, compute_out_storage_size_three_size, compute_out_storage_transparent_size];
+    let pool_sizes = &[ubo_size, sampler_size, render_storage_size, compute_in_storage_size, compute_out_storage_color_size, compute_out_cuboid_size, compute_out_cuboid_index_size, compute_out_storage_size_two_size, compute_out_storage_size_three_size, compute_out_storage_transparent_size, compute_out_storage_oct_tree];
     let info = vk::DescriptorPoolCreateInfo::builder()
         .pool_sizes(pool_sizes)
         .max_sets(data.swapchain_images.len() as u32);
@@ -625,7 +650,7 @@ pub unsafe fn create_descriptor_sets(device: &Device, data: &mut app_data::AppDa
         let info = vk::DescriptorBufferInfo::builder()
             .buffer(data.compute_out_storage_buffers_color[i])
             .offset(0)
-            .range((size_of::<u32>() * 3) as u64 * data.compute_task_one_out_buffer_size.max(1));
+            .range((size_of::<u32>()) as u64 * data.compute_task_one_out_buffer_size.max(1));
         let storage_info = &[info];
 
         let storage_write_compute_out_color = vk::WriteDescriptorSet::builder()
@@ -638,7 +663,7 @@ pub unsafe fn create_descriptor_sets(device: &Device, data: &mut app_data::AppDa
         let info = vk::DescriptorBufferInfo::builder()
             .buffer(data.compute_out_storage_buffers_size_two[i])
             .offset(0)
-            .range((size_of::<u32>() * 2) as u64 * data.compute_task_one_out_buffer_size.max(1));
+            .range((size_of::<u32>()) as u64 * data.compute_task_one_out_buffer_size.max(1));
         let storage_info = &[info];
 
         let storage_write_compute_out_size_two = vk::WriteDescriptorSet::builder()
@@ -651,7 +676,7 @@ pub unsafe fn create_descriptor_sets(device: &Device, data: &mut app_data::AppDa
         let info = vk::DescriptorBufferInfo::builder()
             .buffer(data.compute_out_storage_buffers_size_three[i])
             .offset(0)
-            .range((size_of::<u32>() * 3) as u64 * data.compute_task_one_out_buffer_size.max(1));
+            .range((size_of::<u32>()) as u64 * data.compute_task_one_out_buffer_size.max(1));
         let storage_info = &[info];
 
         let storage_write_compute_out_size_three = vk::WriteDescriptorSet::builder()
@@ -664,7 +689,7 @@ pub unsafe fn create_descriptor_sets(device: &Device, data: &mut app_data::AppDa
         let info = vk::DescriptorBufferInfo::builder()
             .buffer(data.compute_out_storage_buffers_transparent[i])
             .offset(0)
-            .range((size_of::<bool>() * 3) as u64 * data.compute_task_one_out_buffer_size.max(1));
+            .range((size_of::<bool>()) as u64 * data.compute_task_one_out_buffer_size.max(1));
         let storage_info = &[info];
 
         let storage_write_compute_out_transparent = vk::WriteDescriptorSet::builder()
@@ -674,6 +699,19 @@ pub unsafe fn create_descriptor_sets(device: &Device, data: &mut app_data::AppDa
             .descriptor_type(vk::DescriptorType::STORAGE_BUFFER)
             .buffer_info(storage_info);
 
+        let info = vk::DescriptorBufferInfo::builder()
+            .buffer(data.compute_out_storage_buffers_oct_tree[i])
+            .offset(0)
+            .range((size_of::<u32>()) as u64 * data.compute_task_oct_tree_size.max(1));
+        let storage_info = &[info];
+
+        let storage_write_compute_out_oct_tree = vk::WriteDescriptorSet::builder()
+            .dst_set(data.descriptor_sets[i])
+            .dst_binding(10)
+            .dst_array_element(0)
+            .descriptor_type(vk::DescriptorType::STORAGE_BUFFER)
+            .buffer_info(storage_info);
+
         let info = vk::DescriptorBufferInfo::builder()
             .buffer(data.compute_out_cuboid_buffers[i])
             .offset(0)
@@ -702,7 +740,7 @@ pub unsafe fn create_descriptor_sets(device: &Device, data: &mut app_data::AppDa
         
 
         device.update_descriptor_sets(
-            &[ubo_write, sampler_write, storage_write_render, storage_write_compute_in, storage_write_compute_out_color, storage_write_compute_cuboid_out, storage_write_compute_cuboid_index_out, storage_write_compute_out_size_two, storage_write_compute_out_size_three, storage_write_compute_out_transparent],
+            &[ubo_write, sampler_write, storage_write_render, storage_write_compute_in, storage_write_compute_out_color, storage_write_compute_cuboid_out, storage_write_compute_cuboid_index_out, storage_write_compute_out_size_two, storage_write_compute_out_size_three, storage_write_compute_out_transparent, storage_write_compute_out_oct_tree],
             &[] as &[vk::CopyDescriptorSet],
         );
     }

src/command_buffer.rs

@@ -156,6 +156,13 @@ pub unsafe fn create_command_buffers(device: &Device, data: &mut app_data::AppDa
                     &[]);
             
             device.cmd_dispatch(*command_buffer, (data.compute_task_one_size as f64 / 16.0).ceil() as u32, 1, 1);
+            
+            let buffer_memory_barrier_in = vk::BufferMemoryBarrier::builder()
+                .buffer(data.compute_out_storage_buffers_size_three[i])
+                .src_access_mask(vk::AccessFlags::SHADER_READ)
+                .dst_access_mask(vk::AccessFlags::SHADER_WRITE)
+                .size(vk::WHOLE_SIZE as u64)
+                .build();
 
             let buffer_memory_barrier_out = vk::BufferMemoryBarrier::builder()
                 .buffer(data.compute_out_storage_buffers_size_two[i])
@@ -169,7 +176,7 @@ pub unsafe fn create_command_buffers(device: &Device, data: &mut app_data::AppDa
                 vk::PipelineStageFlags::COMPUTE_SHADER,
                 vk::DependencyFlags::DEVICE_GROUP,
                 &[] as &[vk::MemoryBarrier],
-                &[buffer_memory_barrier_out],
+                &[buffer_memory_barrier_in, buffer_memory_barrier_out],
                 &[] as &[vk::ImageMemoryBarrier]);
 
             // grow z axis
@@ -185,6 +192,13 @@ pub unsafe fn create_command_buffers(device: &Device, data: &mut app_data::AppDa
                     &[]);
             
             device.cmd_dispatch(*command_buffer, (data.compute_task_one_size as f64 / 16.0).ceil() as u32, 1, 1);
+            
+            let buffer_memory_barrier_in = vk::BufferMemoryBarrier::builder()
+                .buffer(data.compute_out_storage_buffers_size_two[i])
+                .src_access_mask(vk::AccessFlags::SHADER_READ)
+                .dst_access_mask(vk::AccessFlags::SHADER_WRITE)
+                .size(vk::WHOLE_SIZE as u64)
+                .build();
 
             let buffer_memory_barrier_out = vk::BufferMemoryBarrier::builder()
                 .buffer(data.compute_out_storage_buffers_size_three[i])
@@ -198,12 +212,12 @@ pub unsafe fn create_command_buffers(device: &Device, data: &mut app_data::AppDa
                 vk::PipelineStageFlags::COMPUTE_SHADER,
                 vk::DependencyFlags::DEVICE_GROUP,
                 &[] as &[vk::MemoryBarrier],
-                &[buffer_memory_barrier_out],
+                &[buffer_memory_barrier_in, buffer_memory_barrier_out],
                 &[] as &[vk::ImageMemoryBarrier]);
 
-            // combine element
+            // calculate mem size
             device.cmd_bind_pipeline(
-                *command_buffer, vk::PipelineBindPoint::COMPUTE, data.pipeline_compute_combine);
+                *command_buffer, vk::PipelineBindPoint::COMPUTE, data.pipeline_compute_mempos);
 
             device.cmd_bind_descriptor_sets(
                 *command_buffer, 
@@ -213,10 +227,17 @@ pub unsafe fn create_command_buffers(device: &Device, data: &mut app_data::AppDa
                 &[data.descriptor_sets[i]],
                     &[]);
             
-            device.cmd_dispatch(*command_buffer, (data.compute_task_one_size as f64 / 16.0).ceil() as u32, 1, 1);
+            device.cmd_dispatch(*command_buffer, data.compute_task_oct_tree_nodes as u32, 1, 1);
+            
+            let buffer_memory_barrier_in = vk::BufferMemoryBarrier::builder()
+                .buffer(data.compute_out_storage_buffers_size_three[i])
+                .src_access_mask(vk::AccessFlags::SHADER_READ)
+                .dst_access_mask(vk::AccessFlags::SHADER_WRITE)
+                .size(vk::WHOLE_SIZE as u64)
+                .build();
 
             let buffer_memory_barrier_out = vk::BufferMemoryBarrier::builder()
-                .buffer(data.render_storage_buffers[i])
+                .buffer(data.compute_out_storage_buffers_size_two[i])
                 .src_access_mask(vk::AccessFlags::SHADER_WRITE)
                 .dst_access_mask(vk::AccessFlags::SHADER_READ)
                 .size(vk::WHOLE_SIZE as u64)
@@ -224,10 +245,10 @@ pub unsafe fn create_command_buffers(device: &Device, data: &mut app_data::AppDa
 
             device.cmd_pipeline_barrier(*command_buffer,
                 vk::PipelineStageFlags::COMPUTE_SHADER,
-                vk::PipelineStageFlags::FRAGMENT_SHADER,
+                vk::PipelineStageFlags::COMPUTE_SHADER,
                 vk::DependencyFlags::DEVICE_GROUP,
                 &[] as &[vk::MemoryBarrier],
-                &[buffer_memory_barrier_out],
+                &[buffer_memory_barrier_in, buffer_memory_barrier_out],
                 &[] as &[vk::ImageMemoryBarrier]);
         }
         // start render pass

src/main.rs

@@ -441,13 +441,20 @@ impl App {
             .iter()
             .for_each(|m| self.device.free_memory(*m, None));
 
-            self.data.compute_out_storage_buffers_size_three
+        self.data.compute_out_storage_buffers_size_three
             .iter()
             .for_each(|b| self.device.destroy_buffer(*b, None));
         self.data.compute_out_storage_buffers_memory_size_three
             .iter()
             .for_each(|m| self.device.free_memory(*m, None));
 
+        self.data.compute_out_storage_buffers_oct_tree
+            .iter()
+            .for_each(|b| self.device.destroy_buffer(*b, None));
+        self.data.compute_out_storage_buffers_memory_oct_tree
+            .iter()
+            .for_each(|m| self.device.free_memory(*m, None));
+
         self.data.compute_out_cuboid_buffers
             .iter()
             .for_each(|b| self.device.destroy_buffer(*b, None));
@@ -474,7 +481,7 @@ impl App {
         self.device.destroy_pipeline(self.data.pipeline_compute_grow_one, None);
         self.device.destroy_pipeline(self.data.pipeline_compute_grow_two, None);
         self.device.destroy_pipeline(self.data.pipeline_compute_grow_three, None);
-        self.device.destroy_pipeline(self.data.pipeline_compute_combine, None);
+        self.device.destroy_pipeline(self.data.pipeline_compute_mempos, None);
 
         self.device.destroy_pipeline_layout(self.data.pipeline_layout, None);
         self.device.destroy_render_pass(self.data.render_pass, None);
@@ -897,13 +904,13 @@ unsafe fn create_pipeline(device: &Device, data: &mut app_data::AppData) -> Resu
         .name(b"main\0");
 
     // load the byte data
-    let compute_bytes = include_bytes!("../shaders/compiled/rt_compute_combine.spv");
+    let compute_bytes = include_bytes!("../shaders/compiled/rt_compute_mempos.spv");
     // create the shaders
-    let compute_shader_module_combine = create_shader_module(device, &compute_bytes[..])?;
+    let compute_shader_module_mempos = create_shader_module(device, &compute_bytes[..])?;
     //create the shader stage for the compute shader
-    let compute_stage_combine = vk::PipelineShaderStageCreateInfo::builder()
+    let compute_stage_mempos = vk::PipelineShaderStageCreateInfo::builder()
         .stage(vk::ShaderStageFlags::COMPUTE)
-        .module(compute_shader_module_combine)
+        .module(compute_shader_module_mempos)
         .name(b"main\0");
 
     // define input assembly and object type. This is altered when using geometry shader
@@ -1062,11 +1069,11 @@ unsafe fn create_pipeline(device: &Device, data: &mut app_data::AppData) -> Resu
         .stage(compute_stage_grow_three)
         .layout(data.pipeline_layout);
 
-    let info_compute_combine = vk::ComputePipelineCreateInfo::builder()
-        .stage(compute_stage_combine)
+    let info_compute_mempos = vk::ComputePipelineCreateInfo::builder()
+        .stage(compute_stage_mempos)
         .layout(data.pipeline_layout);
 
-    let compute_pipelines = device.create_compute_pipelines(vk::PipelineCache::null(), &[info_compute_rasterize, info_compute_grow_one, info_compute_grow_two, info_compute_grow_three, info_compute_combine], None)?.0;
+    let compute_pipelines = device.create_compute_pipelines(vk::PipelineCache::null(), &[info_compute_rasterize, info_compute_grow_one, info_compute_grow_two, info_compute_grow_three, info_compute_mempos], None)?.0;
 
     data.pipeline_cube = pipelines[0];
     data.pipeline_cuboid = pipelines[1];
@@ -1076,7 +1083,7 @@ unsafe fn create_pipeline(device: &Device, data: &mut app_data::AppData) -> Resu
     data.pipeline_compute_grow_one = compute_pipelines[1];
     data.pipeline_compute_grow_two = compute_pipelines[2];
     data.pipeline_compute_grow_three = compute_pipelines[3];
-    data.pipeline_compute_combine = compute_pipelines[4];
+    data.pipeline_compute_mempos = compute_pipelines[4];
 
     device.destroy_shader_module(vert_shader_module_cube, None);
     device.destroy_shader_module(geo_shader_module_cube, None);
@@ -1093,7 +1100,7 @@ unsafe fn create_pipeline(device: &Device, data: &mut app_data::AppData) -> Resu
     device.destroy_shader_module(compute_shader_module_grow_one, None);
     device.destroy_shader_module(compute_shader_module_grow_two, None);
     device.destroy_shader_module(compute_shader_module_grow_three, None);
-    device.destroy_shader_module(compute_shader_module_combine, None);
+    device.destroy_shader_module(compute_shader_module_mempos, None);
 
     Ok(())
 }

src/scene/generators.rs

@@ -127,12 +127,12 @@ pub fn generate_test_scene(scene: &mut Scene, data: &mut AppData) -> Result<(Poi
     let mut comp = ShapeComposition::new(64);
     comp.included_shapes.push(Rc::new(RefCell::new(Cone::new(Vector3 { x: 20.0 + grid_size as f32, y: 5.0 + grid_size as f32, z: 10.0 }, Vector3 { x: 0.0, y: 0.0, z: 0.0 }, 0.0, 2.5, Vector3 { x: 0.0, y: 10.0, z: 0.0 },Vector3 { x: 0, y: 255, z: 0 }, 64, false))));
     comp.excluded_shapes.push(Rc::new(RefCell::new(Cone::new(Vector3 { x: 20.0 + grid_size as f32, y: 5.0 + grid_size as f32, z: 10.0 }, Vector3 { x: 0.0, y: 0.0, z: 0.0 }, 0.0, 1.5, Vector3 { x: 0.0, y: 10.0, z: 0.0 },Vector3 { x: 0, y: 255, z: 0 }, 64, false))));
-    scene.volumetrics.push(Rc::new(RefCell::new(comp)));
+    //scene.volumetrics.push(Rc::new(RefCell::new(comp)));
 
     let mut comp = ShapeComposition::new(64);
     comp.included_shapes.push(Rc::new(RefCell::new(Rect::new(Vector3 { x: -5.0 + grid_size as f32, y: 5.0 + grid_size as f32, z: 10.0 }, Vector3 { x: 0.0, y: 0.0, z: 0.0 }, Vector3 { x: 5.0, y: 10.0, z: 2.0 },Vector3 { x: 0, y: 0, z: 255 }, 64, false))));
     comp.excluded_shapes.push(Rc::new(RefCell::new(Rect::new(Vector3 { x: -5.0 + grid_size as f32, y: 5.0 + grid_size as f32, z: 10.0 }, Vector3 { x: 0.0, y: 0.0, z: 0.0 }, Vector3 { x: 3.0, y: 8.0, z: 2.0 },Vector3 { x: 0, y: 0, z: 255 }, 64, false))));
-    scene.volumetrics.push(Rc::new(RefCell::new(comp)));
+    //scene.volumetrics.push(Rc::new(RefCell::new(comp)));
 
     Ok((cgmath::point3(5.0, 5.0, 10.0)))
 }

src/scene/mod.rs

@@ -227,7 +227,12 @@ impl Scene {
 
         let mut compute_task_one_size = 0;
         let mut compute_task_one_out_size = 0;
+        let mut target_index = 1;
+        let mut node_count = 0;
         for compound in &self.volumetrics {
+            compound.borrow_mut().target_memory_start = target_index;
+            target_index += compound.borrow().get_target_buffer_mem_size();
+            node_count += compound.borrow().get_num_nodes();
             volumetrics_memory = compound.borrow_mut().insert_into_memory(volumetrics_memory, data, &self);
             compute_task_one_size += compound.borrow().size.pow(2) as usize;
             compute_task_one_out_size += compound.borrow().size.pow(3) as usize;
@@ -238,6 +243,8 @@ impl Scene {
         data.compute_task_one_size = compute_task_one_size;
         data.compute_task_one_out_buffer_size = (compute_task_one_out_size * 4) as u64;
         data.compute_task_one_out_size = compute_task_one_out_size as u64;
+        data.compute_task_oct_tree_size = target_index as u64;
+        data.compute_task_oct_tree_nodes = node_count as u64;
     }
 
     pub unsafe fn destroy(&mut self, device: &vulkanalia::Device) {

src/scene/volumetrics/mod.rs

@@ -31,6 +31,7 @@ enum ShapeTypes {
 #[derive(Clone, Debug)]
 pub struct ShapeComposition {
     memory_start: usize,
+    pub target_memory_start: u32,
     prev_memory_size: u32,
     pub size: u32,
     pub included_shapes: Vec<Rc<RefCell<dyn Volumetrics>>>,
@@ -40,7 +41,7 @@ pub struct ShapeComposition {
 
 impl ShapeComposition {
     pub fn new(size: u32) -> Self {
-        Self { memory_start: 0, prev_memory_size: 0, size: size, included_shapes: vec![], excluded_shapes: vec![], dirty: true }
+        Self { memory_start: 0, target_memory_start: 0, prev_memory_size: 0, size: size, included_shapes: vec![], excluded_shapes: vec![], dirty: true }
     }
 }
 
@@ -142,6 +143,25 @@ impl Memorizable for ShapeComposition {
     }
 }
 
+impl ShapeComposition {
+    pub fn get_num_nodes(&self) -> u32 {
+        let mut nodes = 0;
+        let mut add_size = 1;
+        let mut size = self.size;
+        while  size >= 2 {
+            nodes += add_size;
+            add_size *= 8;
+            size /= 2;
+        }
+
+        nodes
+    }
+
+    pub fn get_target_buffer_mem_size(&self) -> u32 {
+        self.get_num_nodes() * 9
+    }
+}
+
 #[derive(Clone, Debug, PartialEq)]
 pub struct Sphere {
     pos: Vector3<f32>,