You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

600 lines
18 KiB

float DistanceBetweenPoints(float3 a, float3 b) {
return fast_distance(a, b);
//return sqrt(pow(a.x - b.x, 2) + pow(a.y - b.y, 2) + pow(a.z - b.z, 2));
}
float Distance(float3 a) {
return fast_length(a);
//return sqrt(pow(a.x, 2) + pow(a.y, 2) + pow(a.z, 2));
}
// Naive incident ray light
float4 white_light(float4 input, float3 light, int3 mask) {
input.w = input.w + acos(
dot(
normalize(light),
normalize(convert_float3(mask * (-mask)))
)
) / 32;
input.w += 0.25f;
return input;
}
// Phong + diffuse lighting function for g
// 0 1 2 3 4 5 6 7 8 9
// {r, g, b, i, x, y, z, x', y', z'}
float4 view_light(float4 in_color, float3 light, float4 light_color, float3 view, int3 mask) {
if (all(light == (0.0f,0.0f,0.0f)))
return (0,0,0,0);
float d = Distance(light) / 100.0f;
d *= d;
float diffuse = max(dot(normalize(convert_float3(mask)), normalize(light)), 0.0f);
in_color += diffuse * light_color * 0.5f / d;
if (dot(light, normalize(convert_float3(mask))) > 0.0f)
{
// Small dots of light are caused by floating point error
// flipping bits on the face mask and screwing up this calculation
float3 halfwayVector = normalize(normalize(light) + normalize(view));
float specTmp = max(dot(normalize(convert_float3(mask)), halfwayVector), 0.0f);
in_color += pow(specTmp, 8.0f) * light_color * 0.5f / d;
}
if (in_color.w > 1.0f){
in_color.xyz *= in_color.w;
}
return in_color;
}
int rand(int* seed) // 1 <= *seed < m
{
int const a = 16807; //ie 7**5
int const m = 2147483647; //ie 2**31-1
*seed = ((*seed) * a) % m;
return(*seed);
}
bool get_oct_vox(
int3 position,
global ulong *octree_descriptor_buffer,
global uint *octree_attachment_lookup_buffer,
global ulong *octree_attachment_buffer,
global ulong *settings_buffer
){
// (X, Y, Z) mask for the idx
const uchar idx_set_x_mask = 0x1;
const uchar idx_set_y_mask = 0x2;
const uchar idx_set_z_mask = 0x4;
const uchar mask_8[8] = {
0x1, 0x2, 0x4, 0x8,
0x10, 0x20, 0x40, 0x80
};
// Mask for counting the previous valid bits
const uchar count_mask_8[8] = {
0x1, 0x3, 0x7, 0xF,
0x1F, 0x3F, 0x7F, 0xFF
};
// uint64_t manipulation masks
const ulong child_pointer_mask = 0x0000000000007fff;
const ulong far_bit_mask = 0x8000;
const ulong valid_mask = 0xFF0000;
const ulong leaf_mask = 0xFF000000;
const ulong contour_pointer_mask = 0xFFFFFF00000000;
const ulong contour_mask = 0xFF00000000000000;
// push the root node to the parent stack
ulong current_index = *settings_buffer;
ulong head = octree_descriptor_buffer[current_index];
uint parent_stack_position = 0;
ulong parent_stack[32];
uchar scale = 0;
uchar idx_stack[32];
ulong current_descriptor = 0;
bool found = false;
parent_stack[parent_stack_position] = head;
// Set our initial dimension and the position at the corner of the oct to keep track of our position
int dimension = 128;
int3 quad_position = (0, 0, 0);
// While we are not at the required resolution
// Traverse down by setting the valid/leaf mask to the subvoxel
// Check to see if it is valid
// Yes?
// Check to see if it is a leaf
// No? Break
// Yes? Scale down to the next hierarchy, push the parent to the stack
//
// No?
// Break
while (dimension > 1) {
// So we can be a little bit tricky here and increment our
// array index that holds our masks as we build the idx.
// Adding 1 for X, 2 for Y, and 4 for Z
int mask_index = 0;
// Do the logic steps to find which sub oct we step down into
if (position.x >= (dimension / 2) + quad_position.x) {
// Set our voxel position to the (0,0) of the correct oct
quad_position.x += (dimension / 2);
// increment the mask index and mentioned above
mask_index += 1;
// Set the idx to represent the move
idx_stack[scale] |= idx_set_x_mask;
}
if (position.y >= (dimension / 2) + quad_position.y) {
quad_position.y |= (dimension / 2);
mask_index += 2;
// TODO What is up with the binary operator on this one?
idx_stack[scale] ^= idx_set_y_mask;
}
if (position.z >= (dimension / 2) + quad_position.z) {
quad_position.z += (dimension / 2);
mask_index += 4;
idx_stack[scale] |= idx_set_z_mask;
}
// Check to see if we are on a valid oct
if ((head >> 16) & mask_8[mask_index]) {
// Check to see if it is a leaf
if ((head >> 24) & mask_8[mask_index]) {
// If it is, then we cannot traverse further as CP's won't have been generated
found = true;
return found;
}
// If all went well and we found a valid non-leaf oct then we will traverse further down the hierarchy
scale++;
dimension /= 2;
// Count the number of valid octs that come before and add it to the index to get the position
// Negate it by one as it counts itself
int count = popcount((uchar)(head >> 16) & count_mask_8[mask_index]) - 1;
// access the element at which head points to and then add the specified number of indices
// to get to the correct child descriptor
current_index = current_index + (head & child_pointer_mask) + count;
head = octree_descriptor_buffer[current_index];
// Increment the parent stack position and put the new oct node as the parent
parent_stack_position++;
parent_stack[parent_stack_position] = head;
}
else {
// If the oct was not valid, then no CP's exists any further
// This implicitly says that if it's non-valid then it must be a leaf!!
// It appears that the traversal is now working but I need
// to focus on how to now take care of the end condition.
// Currently it adds the last parent on the second to lowest
// oct CP. Not sure if thats correct
found = 0;
return found;
}
}
found = 1;
return found;
}
// =================================== Boolean ray intersection ============================
// =========================================================================================
bool cast_light_intersection_ray(
global char* map,
global int3* map_dim,
float3 ray_dir,
float3 ray_pos,
global float* lights,
global int* light_count
){
float distance_to_light = DistanceBetweenPoints(ray_pos, (float3)(lights[4], lights[5], lights[6]));
//if (distance_to_light > 200.0f){
// return false;
//}
// Setup the voxel step based on what direction the ray is pointing
int3 voxel_step = { 1, 1, 1 };
voxel_step *= (ray_dir > 0) - (ray_dir < 0);
if (any(ray_dir == (0.0f,0.0f,0.0f)))
return false;
// Setup the voxel coords from the camera origin
int3 voxel = convert_int3(ray_pos);
// Delta T is the units a ray must travel along an axis in order to traverse an integer split
float3 delta_t = fabs(1.0f / ray_dir);
// Compute intersection_t and add in the offset
float3 intersection_t = delta_t * ((ray_pos)-floor(ray_pos)) * convert_float3(voxel_step);
// for negative values, wrap around the delta_t
intersection_t += delta_t * -convert_float3(isless(intersection_t, 0));
int3 face_mask = { 0, 0, 0 };
int length_cutoff = 0;
// Andrew Woo's raycasting algo
do {
// Fancy no branch version of the logic step
face_mask = intersection_t.xyz <= min(intersection_t.yzx, intersection_t.zxy);
intersection_t += delta_t * fabs(convert_float3(face_mask.xyz));
voxel.xyz += voxel_step.xyz * face_mask.xyz;
if (any(voxel >= *map_dim) ||
any(voxel < 0)) {
return false;
}
// If we hit a voxel
int voxel_data = map[voxel.x + (*map_dim).x * (voxel.y + (*map_dim).z * (voxel.z))];
if (voxel_data != 0)
return true;
if (++length_cutoff > 300)
return false;
} while (any(isless(intersection_t, (float3)(distance_to_light - 1))));
return false;
}
// ====================================== Raycaster entry point =====================================
// ==================================================================================================
constant float4 fog_color = { 0.73f, 0.81f, 0.89f, 0.8f };
constant float4 overshoot_color = { 0.00f, 0.00f, 0.00f, 0.00f };
constant float4 overshoot_color_2 = { 0.00f, 0.00f, 0.00f, 0.00f };
__kernel void raycaster(
global char* map,
global int3* map_dim,
global int2* resolution,
global float3* projection_matrix,
global float2* cam_dir,
global float3* cam_pos,
global float* lights,
global int* light_count,
__write_only image2d_t image,
//global int* seed_memory,
__read_only image2d_t texture_atlas,
global int2 *atlas_dim,
global int2 *tile_dim,
global ulong *octree_descriptor_buffer,
global uint *octree_attachment_lookup_buffer,
global ulong *octree_attachment_buffer,
global ulong *settings_buffer
){
// int global_id = x * y;
// Get and set the random seed from seed memory
//int seed = seed_memory[global_id];
//int random_number = rand(&seed);
//seed_memory[global_id] = seed;
// Get the pixel on the viewport, and find the view matrix ray that matches it
int2 pixel = (int2)(get_global_id(0), get_global_id(1));
float3 ray_dir = projection_matrix[pixel.x + (*resolution).x * pixel.y];
// Pitch
ray_dir = (float3)(
ray_dir.z * sin((*cam_dir).x) + ray_dir.x * cos((*cam_dir).x),
ray_dir.y,
ray_dir.z * cos((*cam_dir).x) - ray_dir.x * sin((*cam_dir).x)
);
// Yaw
ray_dir = (float3)(
ray_dir.x * cos((*cam_dir).y) - ray_dir.y * sin((*cam_dir).y),
ray_dir.x * sin((*cam_dir).y) + ray_dir.y * cos((*cam_dir).y),
ray_dir.z
);
// Setup the voxel step based on what direction the ray is pointing
int3 voxel_step = {1, 1, 1};
voxel_step *= (ray_dir > 0) - (ray_dir < 0);
// Setup the voxel coords from the camera origin
int3 voxel = convert_int3(*cam_pos);
// Delta T is the units a ray must travel along an axis in order to
// traverse an integer split
if (any(ray_dir == (0.0f,0.0f,0.0f)))
return;
float3 delta_t = fabs(1.0f / ray_dir);
// Intersection T is the collection of the next intersection points
// for all 3 axis XYZ. We take the full positive cardinality when
// subtracting the floor, so we must transfer the sign over from
// the voxel step
float3 intersection_t = delta_t * ((*cam_pos) - ceil(*cam_pos)) * convert_float3(voxel_step);
// When we transfer the sign over, we get the correct direction of
// the offset, but we merely transposed over the value instead of mirroring
// it over the axis like we want. So here, isless returns a boolean if intersection_t
// is less than 0 which dictates whether or not we subtract the delta which in effect
// mirrors the offset
intersection_t -= delta_t * convert_float3(isless(intersection_t, 0));
int distance_traveled = 0;
int max_distance = 700;
uint bounce_count = 0;
int3 face_mask = { 0, 0, 0 };
int voxel_data = 0;
float3 face_position = (0,0,0);
float4 voxel_color= (0,0,0,0);
float2 tile_face_position = (0,0);
float3 sign = (0,0,0);
float4 first_strike = (0,0,0,0);
// Andrew Woo's raycasting algo
while (distance_traveled < max_distance && bounce_count < 2) {
// Fancy no branch version of the logic step
face_mask = intersection_t.xyz <= min(intersection_t.yzx, intersection_t.zxy);
intersection_t += delta_t * fabs(convert_float3(face_mask.xyz));
voxel.xyz += voxel_step.xyz * face_mask.xyz;
// Test for out of bounds contions, add fog
if (any(voxel >= *map_dim)){
write_imagef(image, pixel, white_light(mix(fog_color, overshoot_color, 1.0 - max(distance_traveled / 700.0f, (float)0)), (float3)(lights[7], lights[8], lights[9]), face_mask));
return;
}
if (any(voxel < 0)) {
write_imagef(image, pixel, white_light(mix(fog_color, overshoot_color_2, 1.0 - max(distance_traveled / 700.0f, (float)0)), (float3)(lights[7], lights[8], lights[9]), face_mask));
return;
}
// If we hit a voxel
// if (voxel.x < 128 && voxel.y < 128 && voxel.z < 128){
// if (get_oct_vox(
// voxel,
// octree_descriptor_buffer,
// octree_attachment_lookup_buffer,
// octree_attachment_buffer,
// settings_buffer
// )){
// voxel_data = 1;
// } else {
// voxel_data = 0;
// }
// } else {
voxel_data = map[voxel.x + (*map_dim).x * (voxel.y + (*map_dim).z * (voxel.z))];
//}
if (voxel_data != 0) {
// Determine where on the 2d plane the ray intersected
face_position = (float3)(0);
tile_face_position = (float2)(0);
sign = (float3)(1.0f, 1.0f, 1.0f);
// First determine the percent of the way the ray is towards the next intersection_t
// in relation to the xyz position on the plane
if (face_mask.x == -1) {
sign.x *= -1.0;
// the next intersection for this plane - the last intersection of the passed plane / delta of this plane
// basically finds how far in on the other 2 axis we are when the ray traversed the plane
float z_percent = (intersection_t.z - (intersection_t.x - delta_t.x)) / delta_t.z;
float y_percent = (intersection_t.y - (intersection_t.x - delta_t.x)) / delta_t.y;
// Since we intersected face x, we know that we are at the face (1.0)
// I think the 1.001f rendering bug is the ray thinking it's within the voxel
// even though it's sitting on the very edge
face_position = (float3)(1.0001f, y_percent, z_percent);
tile_face_position = (float2)(y_percent, z_percent);
}
else if (face_mask.y == -1) {
sign.y *= -1.0;
float x_percent = (intersection_t.x - (intersection_t.y - delta_t.y)) / delta_t.x;
float z_percent = (intersection_t.z - (intersection_t.y - delta_t.y)) / delta_t.z;
face_position = (float3)(x_percent, 1.0001f, z_percent);
tile_face_position = (float2)(x_percent, z_percent);
}
else if (face_mask.z == -1) {
sign.z *= -1.0;
float x_percent = (intersection_t.x - (intersection_t.z - delta_t.z)) / delta_t.x;
float y_percent = (intersection_t.y - (intersection_t.z - delta_t.z)) / delta_t.y;
face_position = (float3)(x_percent, y_percent, 1.0001f);
tile_face_position = (float2)(x_percent, y_percent);
}
// Because the raycasting process is agnostic to the quadrant
// it's working in, we need to transpose the sign over to the face positions.
// If we don't it will think that it is always working in the (1, 1, 1) quadrant
// and will just "copy" the quadrant. This includes shadows as they use the face_position
// in order to cast the intersection ray!!
face_position.x = select((float)(face_position.x), (float)(-face_position.x + 1.0f), (int)(ray_dir.x > 0));
tile_face_position.x = select((float)(tile_face_position.x), (float)(-tile_face_position.x + 1.0f), (int)(ray_dir.x < 0));
if (ray_dir.y > 0){
face_position.y = - face_position.y + 1;
} else {
tile_face_position.x = 1.0 - tile_face_position.x;
// We run into the Hairy ball problem, so we need to define
// a special case for the zmask
if (face_mask.z == -1) {
tile_face_position.x = 1.0 - tile_face_position.x;
tile_face_position.y = 1.0 - tile_face_position.y;
}
}
face_position.z = select((float)(face_position.z), (float)(-face_position.z + 1.0f), (int)(ray_dir.z > 0));
tile_face_position.y = select((float)(tile_face_position.y), (float)(-tile_face_position.y + 1.0f), (int)(ray_dir.z < 0));
// Now either use the face position to retrieve a texture sample, or
// just a plain color for the voxel color. Notice the JANK -1 after the
// conditionals in the select statement. That's because select works on negs
// and pos's. So a false equality will still eval as true as it is technically
// a positive result (0)
// voxel_color = select(
// (float4)(0.25f, 0.64f, 0.87f, 0.0f),
// (float4)voxel_color,
// (int4)((voxel_data == 5) - 1)
// );
if (bounce_count == 0){
voxel_color = (float4)read_imagef(
texture_atlas,
convert_int2(tile_face_position * convert_float2(*atlas_dim / *tile_dim)) +
convert_int2((float2)(3, 0) * convert_float2(*atlas_dim / *tile_dim))
);
voxel_color.w = 0.0f;
first_strike = view_light(
voxel_color,
(convert_float3(voxel) + face_position) - (float3)(lights[4], lights[5], lights[6]),
(float4)(lights[0], lights[1], lights[2], lights[3]),
(convert_float3(voxel) + face_position) - (*cam_pos),
face_mask * voxel_step
);
max_distance = 10;
distance_traveled = 0;
} else {
// voxel_color = (float4)read_imagef(
// texture_atlas,
// convert_int2(tile_face_position * convert_float2(*atlas_dim / *tile_dim)) +
// convert_int2((float2)(3, 0) * convert_float2(*atlas_dim / *tile_dim))
// );
voxel_color.w = 0.2f;
first_strike = white_light(voxel_color, (float3)(1.0f, 1.0f, 1.0f), face_mask);
}
//
// voxel_color = select(
// (float4)(0.0f, 0.239f, 0.419f, 0.0f),
// (float4)read_imagef(
// texture_atlas,
// convert_int2(tile_face_position * convert_float2(*atlas_dim / *tile_dim)) +
// convert_int2((float2)(3, 0) * convert_float2(*atlas_dim / *tile_dim))
// ),
// (int4)((voxel_data == 6) - 1)
// );
// The new direction of the ray
// The list of lights and their distances
// a way to accumulate color and light intensity through multiple bounces??
// Only for indirect lighting and refraction
//
float3 hit_pos = convert_float3(voxel) + face_position;
ray_dir = normalize((float3)(lights[4], lights[5], lights[6]) - hit_pos);
if (any(ray_dir == (0.0f,0.0f,0.0f)))
return;
voxel -= voxel_step * face_mask;
voxel_step = ( 1, 1, 1 );
voxel_step *= (ray_dir > 0) - (ray_dir < 0);
//voxel = convert_int3(hit_pos);
delta_t = fabs(1.0f / ray_dir);
intersection_t = delta_t * ((hit_pos)-floor(hit_pos)) * convert_float3(voxel_step);
intersection_t += delta_t * -convert_float3(isless(intersection_t, 0));
bounce_count += 1;
// max_distance = DistanceBetweenPoints(convert_float3(voxel) + face_position, (float3)(lights[4], lights[5], lights[6]));
// // This has a very large performance hit, I assume CL doesn't really
// like calling into other functions with lots of state.
// if (cast_light_intersection_ray(
// map,
// map_dim,
// normalize((float3)(lights[4], lights[5], lights[6]) - (convert_float3(voxel) + face_position)),
// (convert_float3(voxel) + face_position),
// lights,
// light_count
// )) {
//
// // If the light ray intersected an object on the way to the light point
// write_imagef(image, pixel, white_light(voxel_color, (float3)(1.0f, 1.0f, 1.0f), face_mask));
// return;
// }
//return;
}
distance_traveled++;
}
write_imagef(
image,
pixel,
first_strike
);
// write_imagef(
// image,
// pixel,
// view_light(
// voxel_color,
// (convert_float3(voxel) + face_position) - (float3)(lights[4], lights[5], lights[6]),
// (float4)(lights[0], lights[1], lights[2], lights[3]),
// (convert_float3(voxel) + face_position) - (*cam_pos),
// face_mask * voxel_step
// )
// );
return;
}