Sign in to follow this  
halim5127

OpenGL [HELP] Porting SMAA from Direct3D to OpenGL

Recommended Posts

Hi, please help me on this problem...
I am trying to port SMAA demo (which was built on Direct3D and HLSL) to OpenGL and GLSL.
Here's the SMAA link [url="http://www.iryoku.com/smaa/"]http://www.iryoku.com/smaa/[/url]
I use OpenGL 3.3 and GLSL 330.

I've changed nothing in the shaders. All I do is flipping my Input image, the Area texture, and the Search texture horizontally.
I flipped them because Direct3D texture coordinates origin is on the upper left of the image, and OpenGL is on the lower left.

So far the only correct result I got is in the 1st pass (edge detection), and I cannot get the correct result in the 2nd pass.

I've attached the reference images I took from SMAA 1x using unigine02.png as input, and the color edge detection mode was used.
The image results I got have the "ogl" suffix.
The SMAA settings I used: SMAA_PRESET_HIGH, no reprojection, no predication, the image size is 1280x720

And here's the 2nd pass shader I ported, but it basically the same as the Direct3D version.

vertex shader
[CODE]
#define ATTR_POSITION 0
#define ATTR_NORMAL 1
#define ATTR_TEXCOORD0 5
layout(location = ATTR_POSITION) in vec3 in_Position;
layout(location = ATTR_TEXCOORD0) in vec2 in_TexCoord0;
out float2 Pixcoord;
out float2 Texcoord;
out float4 Offset[3];
/**
* Blend Weight Calculation Vertex Shader
*/
void main(void)
{
gl_Position = float4(in_Position, 1.0);
Texcoord = in_TexCoord0;
Pixcoord = Texcoord / SMAA_PIXEL_SIZE;
// We will use these offsets for the searches later on (see @PSEUDO_GATHER4):
Offset[0] = Texcoord.xyxy + SMAA_PIXEL_SIZE.xyxy * float4(-0.25, -0.125, 1.25, -0.125);
Offset[1] = Texcoord.xyxy + SMAA_PIXEL_SIZE.xyxy * float4(-0.125, -0.25, -0.125, 1.25);
// And these for the searches, they indicate the ends of the loops:
Offset[2] = float4(Offset[0].xz, Offset[1].yw) +
float4(-2.0, 2.0, -2.0, 2.0) *
SMAA_PIXEL_SIZE.xxyy * float(SMAA_MAX_SEARCH_STEPS);
}
[/CODE]

fragment shader
[code]#version 330 core
#if SMAA_PRESET_LOW == 1
#define SMAA_THRESHOLD 0.15
#define SMAA_MAX_SEARCH_STEPS 4
#define SMAA_MAX_SEARCH_STEPS_DIAG 0
#define SMAA_CORNER_ROUNDING 100
#elif SMAA_PRESET_MEDIUM == 1
#define SMAA_THRESHOLD 0.1
#define SMAA_MAX_SEARCH_STEPS 8
#define SMAA_MAX_SEARCH_STEPS_DIAG 0
#define SMAA_CORNER_ROUNDING 100
#elif SMAA_PRESET_HIGH == 1
#define SMAA_THRESHOLD 0.1
#define SMAA_MAX_SEARCH_STEPS 16
#define SMAA_MAX_SEARCH_STEPS_DIAG 8
#define SMAA_CORNER_ROUNDING 25
#elif SMAA_PRESET_ULTRA == 1
#define SMAA_THRESHOLD 0.05
#define SMAA_MAX_SEARCH_STEPS 32
#define SMAA_MAX_SEARCH_STEPS_DIAG 16
#define SMAA_CORNER_ROUNDING 25
#endif
/**
* SMAA_THRESHOLD specifies the threshold or sensitivity to edges.
* Lowering this value you will be able to detect more edges at the expense of
* performance.
*
* Range: [0, 0.5]
* 0.1 is a reasonable value, and allows to catch most visible edges.
* 0.05 is a rather overkill value, that allows to catch 'em all.
*
* If temporal supersampling is used, 0.2 could be a reasonable value, as low
* contrast edges are properly filtered by just 2x.
*/
#ifndef SMAA_THRESHOLD
#define SMAA_THRESHOLD 0.1
#endif
/**
* SMAA_DEPTH_THRESHOLD specifies the threshold for depth edge detection.
*
* Range: depends on the depth range of the scene.
*/
#ifndef SMAA_DEPTH_THRESHOLD
#define SMAA_DEPTH_THRESHOLD (0.1 * SMAA_THRESHOLD)
#endif
/**
* SMAA_MAX_SEARCH_STEPS specifies the maximum steps performed in the
* horizontal/vertical pattern searches, at each side of the pixel.
*
* In number of pixels, it's actually the double. So the maximum line length
* perfectly handled by, for example 16, is 64 (by perfectly, we meant that
* longer lines won't look as good, but still antialiased).
*
* Range: [0, 98]
*/
#ifndef SMAA_MAX_SEARCH_STEPS
#define SMAA_MAX_SEARCH_STEPS 16
#endif
/**
* SMAA_MAX_SEARCH_STEPS_DIAG specifies the maximum steps performed in the
* diagonal pattern searches, at each side of the pixel. In this case we jump
* one pixel at time, instead of two.
*
* Range: [0, 20]; set it to 0 to disable diagonal processing.
*
* On high-end machines it is cheap (between a 0.8x and 0.9x slower for 16
* steps), but it can have a significant impact on older machines.
*/
#ifndef SMAA_MAX_SEARCH_STEPS_DIAG
#define SMAA_MAX_SEARCH_STEPS_DIAG 8
#endif
/**
* SMAA_CORNER_ROUNDING specifies how much sharp corners will be rounded.
*
* Range: [0, 100]; set it to 100 to disable corner detection.
*/
#ifndef SMAA_CORNER_ROUNDING
#define SMAA_CORNER_ROUNDING 25
#endif
/**
* Predicated thresholding allows to better preserve texture details and to
* improve performance, by decreasing the number of detected edges using an
* additional buffer like the light accumulation buffer, object ids or even the
* depth buffer (the depth buffer usage may be limited to indoor or short range
* scenes).
*
* It locally decreases the luma or color threshold if an edge is found in an
* additional buffer (so the global threshold can be higher).
*
* This method was developed by Playstation EDGE MLAA team, and used in
* Killzone 3, by using the light accumulation buffer. More information here:
* http://iryoku.com/aacourse/downloads/06-MLAA-on-PS3.pptx
*/
#ifndef SMAA_PREDICATION
#define SMAA_PREDICATION 0
#endif
/**
* Threshold to be used in the additional predication buffer.
*
* Range: depends on the input, so you'll have to find the magic number that
* works for you.
*/
#ifndef SMAA_PREDICATION_THRESHOLD
#define SMAA_PREDICATION_THRESHOLD 0.01
#endif
/**
* How much to scale the global threshold used for luma or color edge
* detection when using predication.
*
* Range: [1, 5]
*/
#ifndef SMAA_PREDICATION_SCALE
#define SMAA_PREDICATION_SCALE 2.0
#endif
/**
* How much to locally decrease the threshold.
*
* Range: [0, 1]
*/
#ifndef SMAA_PREDICATION_STRENGTH
#define SMAA_PREDICATION_STRENGTH 0.4
#endif
/**
* Temporal reprojection allows to remove ghosting artifacts when using
* temporal supersampling. We use the CryEngine 3 method which also introduces
* velocity weighting. This feature is of extreme importance for totally
* removing ghosting. More information here:
* http://iryoku.com/aacourse/downloads/13-Anti-Aliasing-Methods-in-CryENGINE-3.pdf
*
* Note that you'll need to setup a velocity buffer for enabling reprojection.
* For static geometry, saving the previous depth buffer is a viable
* alternative.
*/
#ifndef SMAA_REPROJECTION
#define SMAA_REPROJECTION 0
#endif
/**
* SMAA_REPROJECTION_WEIGHT_SCALE controls the velocity weighting. It allows to
* remove ghosting trails behind the moving object, which are not removed by
* just using reprojection. Using low values will exhibit ghosting, while using
* high values will disable temporal supersampling under motion.
*
* Behind the scenes, velocity weighting removes temporal supersampling when
* the velocity of the subsamples differs (meaning they are different objects).
*
* Range: [0, 80]
*/
#define SMAA_REPROJECTION_WEIGHT_SCALE 30.0
//-----------------------------------------------------------------------------
// Non-Configurable Defines
#ifndef SMAA_AREATEX_MAX_DISTANCE
#define SMAA_AREATEX_MAX_DISTANCE 16
#endif
#ifndef SMAA_AREATEX_MAX_DISTANCE_DIAG
#define SMAA_AREATEX_MAX_DISTANCE_DIAG 20
#endif
#define SMAA_AREATEX_PIXEL_SIZE (1.0 / float2(160.0, 560.0))
#define SMAA_AREATEX_SUBTEX_SIZE (1.0 / 7.0)
#if SMAA_GLSL_3 == 1 || SMAA_GLSL_4 == 1
#define SMAATexture2D sampler2D
#define SMAASampleLevelZero(tex, coord) textureLod(tex, coord, 0.0)
#define SMAASampleLevelZeroPoint(tex, coord) textureLod(tex, coord, 0.0)
#define SMAASample(tex, coord) texture(tex, coord)
#define SMAASamplePoint(tex, coord) texture(tex, coord)
#define SMAASampleLevelZeroOffset(tex, coord, Offset) textureLodOffset(tex, coord, 0.0, Offset)
#define SMAASampleOffset(tex, coord, Offset) texture(tex, coord, Offset)
#define SMAALerp(a, b, t) mix(a, b, t)
#define SMAASaturate(a) clamp(a, 0.0, 1.0)
#define SMAA_FLATTEN
#define SMAA_BRANCH
#define float2 vec2
#define float3 vec3
#define float4 vec4
#define int2 ivec2
#define int3 ivec3
#define int4 ivec4
#endif
#if SMAA_GLSL_3 == 1
#define SMAAMad(a, b, c) (a * b + c)
#endif
#if SMAA_GLSL_4 == 1
#define SMAAMad(a, b, c) fma(a, b, c)
#define SMAAGather(tex, coord) textureGather(tex, coord)
#endif
in float2 Texcoord;
in float2 Pixcoord;
in float4 Offset[3];
out float4 out_Color0;
uniform SMAATexture2D EdgesTex;
uniform SMAATexture2D AreaTex;
uniform SMAATexture2D SearchTex;
uniform int4 SubsampleIndices;
//-----------------------------------------------------------------------------
// Diagonal Search Functions
#if SMAA_MAX_SEARCH_STEPS_DIAG > 0 || SMAA_FORCE_DIAGONAL_DETECTION == 1
/**
* These functions allows to perform diagonal pattern searches.
*/
float SMAASearchDiag1(SMAATexture2D edgesTex, float2 texcoord, float2 dir, float c) {
texcoord += dir * SMAA_PIXEL_SIZE;
float2 e = float2(0.0, 0.0);
float i;
for (i = 0.0; i < float(SMAA_MAX_SEARCH_STEPS_DIAG); i++) {
e.rg = SMAASampleLevelZero(edgesTex, texcoord).rg;
SMAA_FLATTEN if (dot(e, float2(1.0, 1.0)) < 1.9) break;
texcoord += dir * SMAA_PIXEL_SIZE;
}
return i + float(e.g > 0.9) * c;
}
float SMAASearchDiag2(SMAATexture2D edgesTex, float2 texcoord, float2 dir, float c) {
texcoord += dir * SMAA_PIXEL_SIZE;
float2 e = float2(0.0, 0.0);
float i;
for (i = 0.0; i < float(SMAA_MAX_SEARCH_STEPS_DIAG); i++) {
e.g = SMAASampleLevelZero(edgesTex, texcoord).g;
e.r = SMAASampleLevelZeroOffset(edgesTex, texcoord, int2(1, 0)).r;
SMAA_FLATTEN if (dot(e, float2(1.0, 1.0)) < 1.9) break;
texcoord += dir * SMAA_PIXEL_SIZE;
}
return i + float(e.g > 0.9) * c;
}
/**
* Similar to SMAAArea, this calculates the area corresponding to a certain
* diagonal distance and crossing edges 'e'.
*/
float2 SMAAAreaDiag(SMAATexture2D areaTex, float2 dist, float2 e, float offset) {
float2 texcoord = float(SMAA_AREATEX_MAX_DISTANCE_DIAG) * e + dist;
// We do a scale and bias for mapping to texel space:
texcoord = SMAA_AREATEX_PIXEL_SIZE * texcoord + (0.5 * SMAA_AREATEX_PIXEL_SIZE);
// Diagonal areas are on the second half of the texture:
texcoord.x += 0.5;
// Move to proper place, according to the subpixel offset:
texcoord.y += SMAA_AREATEX_SUBTEX_SIZE * offset;
// Do it!
#if SMAA_HLSL_3 == 1
return SMAASampleLevelZero(areaTex, texcoord).ra;
#else
return SMAASampleLevelZero(areaTex, texcoord).rg;
#endif
}
/**
* This searches for diagonal patterns and returns the corresponding weights.
*/
float2 SMAACalculateDiagWeights(SMAATexture2D edgesTex, SMAATexture2D areaTex, float2 texcoord, float2 e, int4 subsampleIndices) {
float2 weights = float2(0.0, 0.0);
float2 d;
d.x = e.r > 0.0? SMAASearchDiag1(edgesTex, texcoord, float2(-1.0, 1.0), 1.0) : 0.0;
d.y = SMAASearchDiag1(edgesTex, texcoord, float2(1.0, -1.0), 0.0);
SMAA_BRANCH
if (d.r + d.g > 2.0) { // d.r + d.g + 1 > 3
float4 coords = SMAAMad(float4(-d.r, d.r, d.g, -d.g), SMAA_PIXEL_SIZE.xyxy, texcoord.xyxy);
float4 c;
c.x = SMAASampleLevelZeroOffset(edgesTex, coords.xy, int2(-1, 0)).g;
c.y = SMAASampleLevelZeroOffset(edgesTex, coords.xy, int2( 0, 0)).r;
c.z = SMAASampleLevelZeroOffset(edgesTex, coords.zw, int2( 1, 0)).g;
c.w = SMAASampleLevelZeroOffset(edgesTex, coords.zw, int2( 1, -1)).r;
float2 e = 2.0 * c.xz + c.yw;
float t = float(SMAA_MAX_SEARCH_STEPS_DIAG) - 1.0;
e *= step(d.rg, float2(t, t));
weights += SMAAAreaDiag(areaTex, d, e, float(subsampleIndices.z));
}
d.x = SMAASearchDiag2(edgesTex, texcoord, float2(-1.0, -1.0), 0.0);
float right = SMAASampleLevelZeroOffset(edgesTex, texcoord, int2(1, 0)).r;
d.y = right > 0.0? SMAASearchDiag2(edgesTex, texcoord, float2(1.0, 1.0), 1.0) : 0.0;
SMAA_BRANCH
if (d.r + d.g > 2.0) { // d.r + d.g + 1 > 3
float4 coords = SMAAMad(float4(-d.r, -d.r, d.g, d.g), SMAA_PIXEL_SIZE.xyxy, texcoord.xyxy);
float4 c;
c.x = SMAASampleLevelZeroOffset(edgesTex, coords.xy, int2(-1, 0)).g;
c.y = SMAASampleLevelZeroOffset(edgesTex, coords.xy, int2( 0, -1)).r;
c.zw = SMAASampleLevelZeroOffset(edgesTex, coords.zw, int2( 1, 0)).gr;
float2 e = 2.0 * c.xz + c.yw;
float t = float(SMAA_MAX_SEARCH_STEPS_DIAG) - 1.0;
e *= step(d.rg, float2(t, t));
weights += SMAAAreaDiag(areaTex, d, e, float(subsampleIndices.w)).gr;
}
return weights;
}
#endif
//-----------------------------------------------------------------------------
// Horizontal/Vertical Search Functions
/**
* This allows to determine how much length should we add in the last step
* of the searches. It takes the bilinearly interpolated edge (see
* @PSEUDO_GATHER4), and adds 0, 1 or 2, depending on which edges and
* crossing edges are active.
*/
float SMAASearchLength(SMAATexture2D searchTex, float2 e, float bias, float scale) {
// Not required if searchTex accesses are set to point:
// float2 SEARCH_TEX_PIXEL_SIZE = 1.0 / float2(66.0, 33.0);
// e = float2(bias, 0.0) + 0.5 * SEARCH_TEX_PIXEL_SIZE +
// e * float2(scale, 1.0) * float2(64.0, 32.0) * SEARCH_TEX_PIXEL_SIZE;
e.r = bias + e.r * scale;
return 255.0 * SMAASampleLevelZeroPoint(searchTex, e).r;
}
/**
* Horizontal/vertical search functions for the 2nd pass.
*/
float SMAASearchXLeft(SMAATexture2D edgesTex, SMAATexture2D searchTex, float2 texcoord, float end) {
/**
* @PSEUDO_GATHER4
* This texcoord has been offset by (-0.25, -0.125) in the vertex shader to
* sample between edge, thus fetching four edges in a row.
* Sampling with different offsets in each direction allows to disambiguate
* which edges are active from the four fetched ones.
*/
float2 e = float2(0.0, 1.0);
while (texcoord.x > end &&
e.g > 0.8281 && // Is there some edge not activated?
e.r == 0.0) { // Or is there a crossing edge that breaks the line?
e = SMAASampleLevelZero(edgesTex, texcoord).rg;
texcoord -= float2(2.0, 0.0) * SMAA_PIXEL_SIZE;
}
// We correct the previous (-0.25, -0.125) offset we applied:
texcoord.x += 0.25 * SMAA_PIXEL_SIZE.x;
// The searches are bias by 1, so adjust the coords accordingly:
texcoord.x += SMAA_PIXEL_SIZE.x;
// Disambiguate the length added by the last step:
texcoord.x += 2.0 * SMAA_PIXEL_SIZE.x; // Undo last step
texcoord.x -= SMAA_PIXEL_SIZE.x * SMAASearchLength(searchTex, e, 0.0, 0.5);
return texcoord.x;
}
float SMAASearchXRight(SMAATexture2D edgesTex, SMAATexture2D searchTex, float2 texcoord, float end) {
float2 e = float2(0.0, 1.0);
while (texcoord.x < end &&
e.g > 0.8281 && // Is there some edge not activated?
e.r == 0.0) { // Or is there a crossing edge that breaks the line?
e = SMAASampleLevelZero(edgesTex, texcoord).rg;
texcoord += float2(2.0, 0.0) * SMAA_PIXEL_SIZE;
}
texcoord.x -= 0.25 * SMAA_PIXEL_SIZE.x;
texcoord.x -= SMAA_PIXEL_SIZE.x;
texcoord.x -= 2.0 * SMAA_PIXEL_SIZE.x;
texcoord.x += SMAA_PIXEL_SIZE.x * SMAASearchLength(searchTex, e, 0.5, 0.5);
return texcoord.x;
}
float SMAASearchYUp(SMAATexture2D edgesTex, SMAATexture2D searchTex, float2 texcoord, float end) {
float2 e = float2(1.0, 0.0);
while (texcoord.y > end &&
e.r > 0.8281 && // Is there some edge not activated?
e.g == 0.0) { // Or is there a crossing edge that breaks the line?
e = SMAASampleLevelZero(edgesTex, texcoord).rg;
texcoord -= float2(0.0, 2.0) * SMAA_PIXEL_SIZE;
}
texcoord.y += 0.25 * SMAA_PIXEL_SIZE.y;
texcoord.y += SMAA_PIXEL_SIZE.y;
texcoord.y += 2.0 * SMAA_PIXEL_SIZE.y;
texcoord.y -= SMAA_PIXEL_SIZE.y * SMAASearchLength(searchTex, e.gr, 0.0, 0.5);
return texcoord.y;
}
float SMAASearchYDown(SMAATexture2D edgesTex, SMAATexture2D searchTex, float2 texcoord, float end) {
float2 e = float2(1.0, 0.0);
while (texcoord.y < end &&
e.r > 0.8281 && // Is there some edge not activated?
e.g == 0.0) { // Or is there a crossing edge that breaks the line?
e = SMAASampleLevelZero(edgesTex, texcoord).rg;
texcoord += float2(0.0, 2.0) * SMAA_PIXEL_SIZE;
}

texcoord.y -= 0.25 * SMAA_PIXEL_SIZE.y;
texcoord.y -= SMAA_PIXEL_SIZE.y;
texcoord.y -= 2.0 * SMAA_PIXEL_SIZE.y;
texcoord.y += SMAA_PIXEL_SIZE.y * SMAASearchLength(searchTex, e.gr, 0.5, 0.5);
return texcoord.y;
}
/**
* Ok, we have the distance and both crossing edges. So, what are the areas
* at each side of current edge?
*/
float2 SMAAArea(SMAATexture2D areaTex, float2 dist, float e1, float e2, float offset) {
// Rounding prevents precision errors of bilinear filtering:
float2 texcoord = float(SMAA_AREATEX_MAX_DISTANCE) * round(4.0 * float2(e1, e2)) + dist;

// We do a scale and bias for mapping to texel space:
texcoord = SMAA_AREATEX_PIXEL_SIZE * texcoord + (0.5 * SMAA_AREATEX_PIXEL_SIZE);
// Move to proper place, according to the subpixel offset:
texcoord.y += SMAA_AREATEX_SUBTEX_SIZE * offset;
// Do it!
return SMAASampleLevelZero(areaTex, texcoord).rg;
}
//-----------------------------------------------------------------------------
// Corner Detection Functions
void SMAADetectHorizontalCornerPattern(SMAATexture2D edgesTex, inout float2 weights, float2 texcoord, float2 d) {
#if SMAA_CORNER_ROUNDING < 100 || SMAA_FORCE_CORNER_DETECTION == 1
float4 coords = SMAAMad(float4(d.x, 0.0, d.y, 0.0),
SMAA_PIXEL_SIZE.xyxy, texcoord.xyxy);
float2 e;
e.r = SMAASampleLevelZeroOffset(edgesTex, coords.xy, int2(0.0, 1.0)).r;
bool left = abs(d.x) < abs(d.y);
e.g = SMAASampleLevelZeroOffset(edgesTex, coords.xy, int2(0.0, -2.0)).r;
if (left) weights *= SMAASaturate(float(SMAA_CORNER_ROUNDING) / 100.0 + 1.0 - e);
e.r = SMAASampleLevelZeroOffset(edgesTex, coords.zw, int2(1.0, 1.0)).r;
e.g = SMAASampleLevelZeroOffset(edgesTex, coords.zw, int2(1.0, -2.0)).r;
if (!left) weights *= SMAASaturate(float(SMAA_CORNER_ROUNDING) / 100.0 + 1.0 - e);
#endif
}
void SMAADetectVerticalCornerPattern(SMAATexture2D edgesTex, inout float2 weights, float2 texcoord, float2 d) {
#if SMAA_CORNER_ROUNDING < 100 || SMAA_FORCE_CORNER_DETECTION == 1
float4 coords = SMAAMad(float4(0.0, d.x, 0.0, d.y),
SMAA_PIXEL_SIZE.xyxy, texcoord.xyxy);
float2 e;
e.r = SMAASampleLevelZeroOffset(edgesTex, coords.xy, int2( 1.0, 0.0)).g;
bool left = abs(d.x) < abs(d.y);
e.g = SMAASampleLevelZeroOffset(edgesTex, coords.xy, int2(-2.0, 0.0)).g;
if (left) weights *= SMAASaturate(float(SMAA_CORNER_ROUNDING) / 100.0 + 1.0 - e);
e.r = SMAASampleLevelZeroOffset(edgesTex, coords.zw, int2( 1.0, 1.0)).g;
e.g = SMAASampleLevelZeroOffset(edgesTex, coords.zw, int2(-2.0, 1.0)).g;
if (!left) weights *= SMAASaturate(float(SMAA_CORNER_ROUNDING) / 100.0 + 1.0 - e);
#endif
}
//-----------------------------------------------------------------------------
// Blending Weight Calculation Pixel Shader (Second Pass)
float4 SMAABlendingWeightCalculationPS(float2 texcoord,
float2 pixcoord,
float4 offset[3],
SMAATexture2D edgesTex,
SMAATexture2D areaTex,
SMAATexture2D searchTex,
int4 subsampleIndices) { // Just pass zero for SMAA 1x, see @SUBSAMPLE_INDICES.
float4 weights = float4(0.0, 0.0, 0.0, 0.0);
float2 e = SMAASample(edgesTex, texcoord).rg;
SMAA_BRANCH
if (e.g > 0.0) { // Edge at north
#if SMAA_MAX_SEARCH_STEPS_DIAG > 0 || SMAA_FORCE_DIAGONAL_DETECTION == 1
// Diagonals have both north and west edges, so searching for them in
// one of the boundaries is enough.
weights.rg = SMAACalculateDiagWeights(edgesTex, areaTex, texcoord, e, subsampleIndices);
// We give priority to diagonals, so if we find a diagonal we skip
// horizontal/vertical processing.
SMAA_BRANCH
if (dot(weights.rg, float2(1.0, 1.0)) == 0.0) {
#endif
float2 d;
// Find the distance to the left:
float2 coords;
coords.x = SMAASearchXLeft(edgesTex, searchTex, offset[0].xy, offset[2].x);
coords.y = offset[1].y; // offset[1].y = texcoord.y - 0.25 * SMAA_PIXEL_SIZE.y (@CROSSING_OFFSET)
d.x = coords.x;
// Now fetch the left crossing edges, two at a time using bilinear
// filtering. Sampling at -0.25 (see @CROSSING_OFFSET) enables to
// discern what value each edge has:
float e1 = SMAASampleLevelZero(edgesTex, coords).r;
// Find the distance to the right:
coords.x = SMAASearchXRight(edgesTex, searchTex, offset[0].zw, offset[2].y);
d.y = coords.x;
// We want the distances to be in pixel units (doing this here allow to
// better interleave arithmetic and memory accesses):
d = d / SMAA_PIXEL_SIZE.x - pixcoord.x;
// SMAAArea below needs a sqrt, as the areas texture is compressed
// quadratically:
float2 sqrt_d = sqrt(abs(d));
// Fetch the right crossing edges:
float e2 = SMAASampleLevelZeroOffset(edgesTex, coords, int2(1, 0)).r;
// Ok, we know how this pattern looks like, now it is time for getting
// the actual area:
weights.rg = SMAAArea(areaTex, sqrt_d, e1, e2, float(subsampleIndices.y));
// Fix corners:
SMAADetectHorizontalCornerPattern(edgesTex, weights.rg, texcoord, d);
#if SMAA_MAX_SEARCH_STEPS_DIAG > 0 || SMAA_FORCE_DIAGONAL_DETECTION == 1
} else
e.r = 0.0; // Skip vertical processing.
#endif
}
SMAA_BRANCH
if (e.r > 0.0) { // Edge at west
float2 d;
// Find the distance to the top:
float2 coords;
coords.y = SMAASearchYUp(edgesTex, searchTex, offset[1].xy, offset[2].z);
coords.x = offset[0].x; // offset[1].x = texcoord.x - 0.25 * SMAA_PIXEL_SIZE.x;
d.x = coords.y;
// Fetch the top crossing edges:
float e1 = SMAASampleLevelZero(edgesTex, coords).g;
// Find the distance to the bottom:
coords.y = SMAASearchYDown(edgesTex, searchTex, offset[1].zw, offset[2].w);
d.y = coords.y;
// We want the distances to be in pixel units:
d = d / SMAA_PIXEL_SIZE.y - pixcoord.y;
// SMAAArea below needs a sqrt, as the areas texture is compressed
// quadratically:
float2 sqrt_d = sqrt(abs(d));
// Fetch the bottom crossing edges:
float e2 = SMAASampleLevelZeroOffset(edgesTex, coords, int2(0, 1)).g;
// Get the area for this direction:
weights.ba = SMAAArea(areaTex, sqrt_d, e1, e2, float(subsampleIndices.x));
// Fix corners:
SMAADetectVerticalCornerPattern(edgesTex, weights.ba, texcoord, d);
}
return weights;
}
void main(void)
{
out_Color0 = SMAABlendingWeightCalculationPS(Texcoord, Pixcoord, Offset, EdgesTex, AreaTex, SearchTex, SubsampleIndices);
}[/code]


Anyone please help me.. I've spent many days to solve this problem but couldn't get it works.
Thanks in advance and I'm sorry for my bad english..

Share this post


Link to post
Share on other sites
in one shader file your doing this :
#define float4 vec4
#define float3 vec3
etc....

which allows you to use float4 for example.. But the other shader you dont difine any of them..
Could that be the problem or are you not posting the full source?

Share this post


Link to post
Share on other sites
Hi,

I've just started working on implementing this :)
In the SMAA.h file it says that they have GLSL implementation, so you should check it out.

Share this post


Link to post
Share on other sites
Hi,

I've completed the edge detection pass, it was just a kind of copy-paste...
Now I implemented the blending weights pass, it was the same thing. However it doesn't work for me neither.
Loading the area texture works perfectly, but it is the opposite with loading the search texture.
So I started to track the issue in gDEBugger. When I started my app (from gdeb) it paused at an opengl error:
[CODE]
glTexImage2D(GL_TEXTURE_2D , 0 , GL_R8 , 66 , 33 , 0 , WGL_DRAW_TO_BITMAP_ARB , GL_UNSIGNED_BYTE , 0x01319778);
[/CODE]

this call caused it. Now this call supposed to load in the search texture. The original call from the source looks like this:
[CODE]
glActiveTexture(GL_TEXTURE2);
search_tex.create();
search_tex.valid = true;
search_tex.bind();
glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE );
glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE );
glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST );
glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST );
glTexImage2D( GL_TEXTURE_2D, 0, GL_R8, ( GLsizei )SEARCHTEX_WIDTH, ( GLsizei )SEARCHTEX_HEIGHT, 0, GL_R, GL_UNSIGNED_BYTE, &searchTexBytes );
search_tex.width = SEARCHTEX_WIDTH;
search_tex.height = SEARCHTEX_HEIGHT;
glActiveTexture(GL_TEXTURE0);
[/CODE]

as you can see gDEBugger reported that I used WGL_DRAW_TO_BITMAP_ARB instead of GL_R. As it turns out both GL_R and WGL_DRAW_TO_BITMAP_ARB are defined as 0x2002, so this might have been the issue, but this left me curious. So I looked up glTexImage2D and as it turns out you have to use GL_RED instead of GL_R, which would be more logical...

but after loading the images this way, and flipping them, still I don't get the desired results...

EDIT: I attached some images of my edge detection and blending weights results and the smaa demo's results.
you can see that the edge detection works but the when the area texture and the search texture gets involved it just fails to reproduce the correct results.
I attached the input file as well

Share this post


Link to post
Share on other sites
Hi,

I hope this doesn't count as resurrecting old topic :)
SMAA got finally fixed, it now works in OGL/GLSL, you can find the source code here: https://github.com/scrawl/smaa-opengl

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

Sign in to follow this  

  • Forum Statistics

    • Total Topics
      627733
    • Total Posts
      2978839
  • Similar Content

    • By DelicateTreeFrog
      Hello! As an exercise for delving into modern OpenGL, I'm creating a simple .obj renderer. I want to support things like varying degrees of specularity, geometry opacity, things like that, on a per-material basis. Different materials can also have different textures. Basic .obj necessities. I've done this in old school OpenGL, but modern OpenGL has its own thing going on, and I'd like to conform as closely to the standards as possible so as to keep the program running correctly, and I'm hoping to avoid picking up bad habits this early on.
      Reading around on the OpenGL Wiki, one tip in particular really stands out to me on this page:
      For something like a renderer for .obj files, this sort of thing seems almost ideal, but according to the wiki, it's a bad idea. Interesting to note!
      So, here's what the plan is so far as far as loading goes:
      Set up a type for materials so that materials can be created and destroyed. They will contain things like diffuse color, diffuse texture, geometry opacity, and so on, for each material in the .mtl file. Since .obj files are conveniently split up by material, I can load different groups of vertices/normals/UVs and triangles into different blocks of data for different models. When it comes to the rendering, I get a bit lost. I can either:
      Between drawing triangle groups, call glUseProgram to use a different shader for that particular geometry (so a unique shader just for the material that is shared by this triangle group). or
      Between drawing triangle groups, call glUniform a few times to adjust different parameters within the "master shader", such as specularity, diffuse color, and geometry opacity. In both cases, I still have to call glBindTexture between drawing triangle groups in order to bind the diffuse texture used by the material, so there doesn't seem to be a way around having the CPU do *something* during the rendering process instead of letting the GPU do everything all at once.
      The second option here seems less cluttered, however. There are less shaders to keep up with while one "master shader" handles it all. I don't have to duplicate any code or compile multiple shaders. Arguably, I could always have the shader program for each material be embedded in the material itself, and be auto-generated upon loading the material from the .mtl file. But this still leads to constantly calling glUseProgram, much more than is probably necessary in order to properly render the .obj. There seem to be a number of differing opinions on if it's okay to use hundreds of shaders or if it's best to just use tens of shaders.
      So, ultimately, what is the "right" way to do this? Does using a "master shader" (or a few variants of one) bog down the system compared to using hundreds of shader programs each dedicated to their own corresponding materials? Keeping in mind that the "master shaders" would have to track these additional uniforms and potentially have numerous branches of ifs, it may be possible that the ifs will lead to additional and unnecessary processing. But would that more expensive than constantly calling glUseProgram to switch shaders, or storing the shaders to begin with?
      With all these angles to consider, it's difficult to come to a conclusion. Both possible methods work, and both seem rather convenient for their own reasons, but which is the most performant? Please help this beginner/dummy understand. Thank you!
    • By JJCDeveloper
      I want to make professional java 3d game with server program and database,packet handling for multiplayer and client-server communicating,maps rendering,models,and stuffs Which aspect of java can I learn and where can I learn java Lwjgl OpenGL rendering Like minecraft and world of tanks
    • By AyeRonTarpas
      A friend of mine and I are making a 2D game engine as a learning experience and to hopefully build upon the experience in the long run.

      -What I'm using:
          C++;. Since im learning this language while in college and its one of the popular language to make games with why not.     Visual Studios; Im using a windows so yea.     SDL or GLFW; was thinking about SDL since i do some research on it where it is catching my interest but i hear SDL is a huge package compared to GLFW, so i may do GLFW to start with as learning since i may get overwhelmed with SDL.  
      -Questions
      Knowing what we want in the engine what should our main focus be in terms of learning. File managements, with headers, functions ect. How can i properly manage files with out confusing myself and my friend when sharing code. Alternative to Visual studios: My friend has a mac and cant properly use Vis studios, is there another alternative to it?  
    • By ferreiradaselva
      Both functions are available since 3.0, and I'm currently using `glMapBuffer()`, which works fine.
      But, I was wondering if anyone has experienced advantage in using `glMapBufferRange()`, which allows to specify the range of the mapped buffer. Could this be only a safety measure or does it improve performance?
      Note: I'm not asking about glBufferSubData()/glBufferData. Those two are irrelevant in this case.
    • By xhcao
      Before using void glBindImageTexture(    GLuint unit, GLuint texture, GLint level, GLboolean layered, GLint layer, GLenum access, GLenum format), does need to make sure that texture is completeness. 
  • Popular Now