Fixes and tweeks for things

src/main/java/me/cortex/voxy/client/core/model/TextureUtils.java

@@ -92,10 +92,10 @@ public class TextureUtils {
     private static float u2fdepth(int depth) {
         float depthF = (float) ((double)depth/((1<<24)-1));
         //https://registry.khronos.org/OpenGL-Refpages/gl4/html/glDepthRange.xhtml
-        // due to this and the unsigned bullshit, i believe the depth value needs to get multiplied by 2
+        // due to this and the unsigned bullshit, believe the depth value needs to get multiplied by 2
 
         //Shouldent be needed due to the compute bake copy
-        //depthF *= 2;
+        depthF *= 2;
         if (depthF > 1.00001f) {
             System.err.println("Warning: Depth greater than 1");
             depthF = 1.0f;

src/main/java/me/cortex/voxy/client/core/rendering/building/RenderDataFactory45.java

@@ -359,6 +359,7 @@ public class RenderDataFactory45 {
 
 
             if (false) {//Non fully opaque geometry
+                this.blockMesher.doAuxiliaryFaceOffset = false;
                 //Note: think is ok to just reuse.. blockMesher
                 this.blockMesher.axis = axis;
                 for (int layer = 0; layer < 32; layer++) {
@@ -400,6 +401,7 @@ public class RenderDataFactory45 {
                     }
                     this.blockMesher.finish();
                 }
+                this.blockMesher.doAuxiliaryFaceOffset = true;
             }
         }
     }

src/main/java/me/cortex/voxy/client/core/rendering/hierachical/NodeCleaner.java

@@ -119,6 +119,11 @@ public class NodeCleaner {
         this.setIds(this.freeIds, -1);
 
         if (this.shouldCleanGeometry()) {
+            var gm = this.nodeManager.getGeometryManager();
+
+            int c = (int) (((((double) gm.getUsedCapacity() / gm.geometryCapacity) - 0.75) * 4 * 10) + 1);
+            c = 1;
+            for (int i = 0; i < c; i++) {
                 glMemoryBarrier(GL_SHADER_STORAGE_BARRIER_BIT);
                 this.outputBuffer.fill(this.nodeManager.maxNodeCount - 2);//TODO: maybe dont set to zero??
 
@@ -146,6 +151,7 @@ public class NodeCleaner {
                 //glFinish();
             }
         }
+    }
 
     private boolean shouldCleanGeometry() {
         //// if there is less than 200mb of space, clean

src/main/java/me/cortex/voxy/client/core/rendering/hierachical/TestNodeManager.java

@@ -274,8 +274,8 @@ public class TestNodeManager {
 
     public static void main(String[] args) {
         Logger.INSERT_CLASS = false;
-        int ITER_COUNT = 5000;
+        int ITER_COUNT = 50_000;
-        int INNER_ITER_COUNT = 100_000;
+        int INNER_ITER_COUNT = 500_000;
         boolean GEO_REM = true;
 
         AtomicInteger finished = new AtomicInteger();
@@ -283,7 +283,7 @@ public class TestNodeManager {
 
         Logger.SHUTUP = true;
 
-        if (true) {
+        if (false) {
             for (int q = 0; q < ITER_COUNT; q++) {
                 //Logger.info("Iteration "+ q);
                 if (runTest(INNER_ITER_COUNT, q, seenTraces, GEO_REM)) {

src/main/resources/assets/voxy/shaders/lod/hierarchical/cleaner/result_transformer.comp

@@ -22,7 +22,7 @@ void main() {
     uint id = minVisIds[gl_LocalInvocationID.x];
     uvec4 node = nodes[id];
     uvec2 res = node.xy;
-    if (all(equal(node, uvec4(0)))) {//If its a empty node, TODO: DONT THINK THIS IS ACTUALLY CORRECT
+    if (all(equal(node, uvec4(-1)))) {//If its a empty node, TODO: DONT THINK THIS IS ACTUALLY CORRECT
         res = uvec2(-1);
     }
     outputBuffer[gl_LocalInvocationID.x] = res;//Move the position of the node id into the output buffer
@@ -30,6 +30,6 @@ void main() {
     //This is an evil hack to not spam the same node 500 times in a row
     //TODO: maybe instead set this to the current frame index
 //    visibility[id] += 30;
-    visibility[id] = visibilityCounter;
+    visibility[id] = visibilityCounter + 60;
 
 }

src/main/resources/assets/voxy/shaders/lod/hierarchical/cleaner/sort_visibility.comp

@@ -18,7 +18,8 @@ layout(binding = OUTPUT_BUFFER_BINDING, std430) restrict volatile buffer Minimum
 };
 
 //Returns the id of the max value
-uint atomicDerefMax(uint atId, uint id, uint value) {
+uint atomicDerefMaxExchange(uint atId, uint id) {
+    const uint value = visiblity[id];
     while (true) {
         const uint existingId = minVisIds[atId];
         //Check if the value is less than the dereferenced value, if its not, return our own id
@@ -26,7 +27,7 @@ uint atomicDerefMax(uint atId, uint id, uint value) {
             return id;
         }
         //Attempt to swap, since we know we are less than the existingId
-        uint c = atomicCompSwap(minVisIds[atId], existingId, id);
+        const uint c = atomicCompSwap(minVisIds[atId], existingId, id);
         //Check if we did swap, else if we failed (or got reswapped else where) recheck
 
         //We did swap, (since the original mem contents was the existing id)
@@ -40,7 +41,7 @@ uint atomicDerefMax(uint atId, uint id, uint value) {
 //TODO: optimize
 void bubbleSort(uint start, uint id) {
     for (uint i = start; i < OUTPUT_SIZE; i++) {
-        id = atomicDerefMax(i, id, visiblity[id]);
+        id = atomicDerefMaxExchange(i, id);
     }
 }
 
@@ -50,6 +51,9 @@ void main() {
     //}
     //First do a min sort/set of min OUTPUT_SIZE values of the set
     uint vis = visiblity[gl_GlobalInvocationID.x];
+    if (vis == uint(-1)) {
+        return;
+    }
     if (visiblity[minVisIds[OUTPUT_SIZE-1]] <= vis) {
         return;
     }