[mtl canvas] fix bias computation for clockwise triangles

This commit is contained in:
Martin Fouilleul 2023-03-15 10:04:51 +01:00
parent 5c1a220f1f
commit f2cb96826c
1 changed files with 35 additions and 42 deletions

View File

@ -31,6 +31,27 @@ bool is_top_left(float2 a, float2 b)
return( (a.y == b.y && b.x < a.x) return( (a.y == b.y && b.x < a.x)
||(b.y < a.y)); ||(b.y < a.y));
} }
/*
bool is_top_left(int2 a, int2 b)
{
return( (a.y == b.y && b.x < a.x)
||(b.y < a.y));
}
*/
//////////////////////////////////////////////////////////////////////////////
//TODO: we should do these computations on 64bits, because otherwise
// we might overflow for values > 2048.
// Unfortunately this is costly.
// Another way is to precompute triangle edges (b - a) in full precision
// once to avoid doing it all the time...
//////////////////////////////////////////////////////////////////////////////
int orient2d(int2 a, int2 b, int2 c)
{
return((b.x-a.x)*(c.y-a.y) - (b.y-a.y)*(c.x-a.x));
}
kernel void TileKernel(constant mg_vertex* vertexBuffer [[buffer(0)]], kernel void TileKernel(constant mg_vertex* vertexBuffer [[buffer(0)]],
constant uint* indexBuffer [[buffer(1)]], constant uint* indexBuffer [[buffer(1)]],
@ -67,13 +88,6 @@ kernel void TileKernel(constant mg_vertex* vertexBuffer [[buffer(0)]],
boxMin = max(boxMin, clip.xy); boxMin = max(boxMin, clip.xy);
boxMax = min(boxMax, clip.zw); boxMax = min(boxMax, clip.zw);
//NOTE(martin): compute triangle orientation and bias for each edge
int cw = ((p1 - p0).x*(p2 - p0).y - (p1 - p0).y*(p2 - p0).x) > 0 ? 1 : -1;
int bias0 = is_top_left(p1, p2) ? 0 : -1;
int bias1 = is_top_left(p2, p0) ? 0 : -1;
int bias2 = is_top_left(p0, p1) ? 0 : -1;
//NOTE(martin): fill triangle data //NOTE(martin): fill triangle data
const float subPixelFactor = 16; const float subPixelFactor = 16;
@ -93,20 +107,26 @@ kernel void TileKernel(constant mg_vertex* vertexBuffer [[buffer(0)]],
triangleArray[gid].cubic1 = vertexBuffer[i1].cubic; triangleArray[gid].cubic1 = vertexBuffer[i1].cubic;
triangleArray[gid].cubic2 = vertexBuffer[i2].cubic; triangleArray[gid].cubic2 = vertexBuffer[i2].cubic;
triangleArray[gid].p0 = int2(p0 * subPixelFactor); int2 ip0 = int2(p0 * subPixelFactor);
triangleArray[gid].p1 = int2(p1 * subPixelFactor); int2 ip1 = int2(p1 * subPixelFactor);
triangleArray[gid].p2 = int2(p2 * subPixelFactor); int2 ip2 = int2(p2 * subPixelFactor);
triangleArray[gid].bias0 = bias0; triangleArray[gid].p0 = ip0;
triangleArray[gid].bias1 = bias1; triangleArray[gid].p1 = ip1;
triangleArray[gid].bias2 = bias2; triangleArray[gid].p2 = ip2;
//NOTE(martin): compute triangle orientation and bias for each edge
int cw = orient2d(ip0, ip1, ip2) > 0 ? 1 : -1;
triangleArray[gid].cw = cw; triangleArray[gid].cw = cw;
triangleArray[gid].bias0 = is_top_left(p1, p2) ? -(1-cw)/2 : -(1+cw)/2;
int4 box = int4(floor(fbox))/RENDERER_TILE_SIZE; triangleArray[gid].bias1 = is_top_left(p2, p0) ? -(1-cw)/2 : -(1+cw)/2;
triangleArray[gid].bias2 = is_top_left(p0, p1) ? -(1-cw)/2 : -(1+cw)/2;
//NOTE(martin): it's important to do the computation with signed int, so that we can have negative xMax/yMax //NOTE(martin): it's important to do the computation with signed int, so that we can have negative xMax/yMax
// otherwise all triangles on the left or below the x/y axis are attributed to tiles on row/column 0. // otherwise all triangles on the left or below the x/y axis are attributed to tiles on row/column 0.
int4 box = int4(floor(fbox))/RENDERER_TILE_SIZE;
int xMin = max(0, box.x); int xMin = max(0, box.x);
int yMin = max(0, box.y); int yMin = max(0, box.y);
int xMax = min(box.z, nTilesX-1); int xMax = min(box.z, nTilesX-1);
@ -153,33 +173,6 @@ kernel void SortKernel(constant mg_triangle_data* triangleArray [[buffer(0)]],
} }
} }
bool is_top_left(int2 a, int2 b)
{
return( (a.y == b.y && b.x < a.x)
||(b.y < a.y));
}
//////////////////////////////////////////////////////////////////////////////
//TODO: we should do these computations on 64bits, because otherwise
// we might overflow for values > 2048.
// Unfortunately this is costly.
// Another way is to precompute triangle edges (b - a) in full precision
// once to avoid doing it all the time...
//////////////////////////////////////////////////////////////////////////////
//TODO: coalesce
int orient2d(int2 a, int2 b, int2 c)
{
return((b.x-a.x)*(c.y-a.y) - (b.y-a.y)*(c.x-a.x));
}
int is_clockwise(int2 p0, int2 p1, int2 p2)
{
return((p1 - p0).x*(p2 - p0).y - (p1 - p0).y*(p2 - p0).x);
}
kernel void RenderKernel(const device uint* tileCounters [[buffer(0)]], kernel void RenderKernel(const device uint* tileCounters [[buffer(0)]],
const device uint* tileArrayBuffer [[buffer(1)]], const device uint* tileArrayBuffer [[buffer(1)]],
const device mg_triangle_data* triangleArray [[buffer(2)]], const device mg_triangle_data* triangleArray [[buffer(2)]],