eessppeelllloo/processing4
1
0
Fork 0
mirror of https://github.com/processing/processing4.git synced 2026-02-12 01:50:44 +01:00
Code Issues Packages Projects Releases Wiki Activity

changes from ewjordan for several accuracy issues

Browse Source
This commit is contained in:
benfry
2007-06-23 23:56:25 +00:00
parent dca0bf9909
commit eb5fc6d87f
5 changed files with 1494 additions and 100 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
core/src/processing/core/PConstants.java
View File

@@ -309,13 +309,14 @@ public interface PConstants {
//static final int SCALE_STROKE_WIDTH = 0;
//static final int LIGHTING_AFFECTS_STROKE = 1;
static final int ENABLE_NATIVE_FONTS = 2;
static final int DISABLE_TEXT_SMOOTH = 3;
static final int ENABLE_NATIVE_FONTS = 2;
static final int DISABLE_TEXT_SMOOTH = 3;
//static final int DISABLE_SMOOTH_HACK = 4;
static final int DISABLE_DEPTH_TEST = 5;
static final int NO_FLYING_POO = 6;
static final int ENABLE_DEPTH_SORT = 7;
static final int DISABLE_ERROR_REPORT = 8;
static final int DISABLE_DEPTH_TEST = 5;
static final int NO_FLYING_POO = 6;
static final int ENABLE_DEPTH_SORT = 7;
static final int DISABLE_ERROR_REPORT = 8;
static final int ENABLE_ACCURATE_TEXTURES = 9;
static final int HINT_COUNT = 9;
static final int HINT_COUNT = 10;
}

163
core/src/processing/core/PGraphics3D.java
View File

@@ -87,6 +87,17 @@ public class PGraphics3D extends PGraphics {
protected PMatrix forwardTransform;
protected PMatrix reverseTransform;
// Added by ewjordan for accurate texturing purposes. Screen plane is
// not scaled to pixel-size, so these manually keep track of its size
// from frustum() calls. Sorry to add public vars, is there a way
// to compute these from something publicly available without matrix ops?
// (used once per triangle in PTriangle with ENABLE_ACCURATE_TEXTURES)
protected float leftScreen;
protected float rightScreen;
protected float topScreen;
protected float bottomScreen;
protected float nearPlane; //depth of near clipping plane
// ........................................................
// pos of first vertex of current shape in vertices array
@@ -1341,6 +1352,24 @@ public class PGraphics3D extends PGraphics {
int tex = triangles[i][TEXTURE_INDEX];
int index = triangles[i][INDEX];
float shift = 0.49f;
boolean shifted = false;
if (drawing2D() && (a[MZ] == 0)) {
shifted = true;
a[X] += shift;
a[Y] += shift;
a[VX] += shift*a[VW];
a[VY] += shift*a[VW];
b[X] += shift;
b[Y] += shift;
b[VX] += shift*b[VW];
b[VY] += shift*b[VW];
c[X] += shift;
c[Y] += shift;
c[VX] += shift*b[VW];
c[VY] += shift*b[VW];
}
triangle.reset();
// This is only true when not textured. We really should pass SPECULAR
@@ -1377,6 +1406,14 @@ public class PGraphics3D extends PGraphics {
b[X], b[Y], b[Z],
c[X], c[Y], c[Z]);
// Need to pass camera-space coordinates to triangle renderer
// in order to compute true texture coordinates, else skip it
if (hints[ENABLE_ACCURATE_TEXTURES]){
triangle.setCamVertices(a[VX], a[VY], a[VZ],
b[VX], b[VY], b[VZ],
c[VX], c[VY], c[VZ]);
}
triangle.setIndex(index);
triangle.render();
@@ -1401,6 +1438,21 @@ public class PGraphics3D extends PGraphics {
raw.vertex(c[X], c[Y]);
}
}
if (drawing2D() && shifted){
a[X] -= shift;
a[Y] -= shift;
a[VX] -= shift*a[VW];
a[VY] -= shift*a[VW];
b[X] -= shift;
b[Y] -= shift;
b[VX] -= shift*b[VW];
b[VY] -= shift*b[VW];
c[X] -= shift;
c[Y] -= shift;
c[VX] -= shift*b[VW];
c[VY] -= shift*b[VW];
}
}
if (raw != null) {
@@ -1409,29 +1461,6 @@ public class PGraphics3D extends PGraphics {
}
/*
public void triangleCallback(float x1, float y1, float z1,
float r1, float g1, float b1, float a1,
float u1, float v1, boolean e1,
float x2, float y2, float z2,
float r2, float g2, float b2, float a2,
float u2, float v2, boolean e2,
float x3, float y3, float z3,
float r3, float g3, float b3, float a3,
float u3, float v3, boolean e3,
PImage texture) {
}
public void lineCallback(float x1, float y1, float z1,
float r1, float g1, float b1, float a1,
float x2, float y2, float z2,
float r2, float g2, float b2, float a2,
float weight, int cap, int join) {
}
*/
protected void depth_sort_lines() {
}
@@ -1448,6 +1477,35 @@ public class PGraphics3D extends PGraphics {
float b[] = vertices[lines[i][VERTEX2]];
int index = lines[i][INDEX];
// 2D hack added by ewjordan 6/13/07
// Offset coordinates by a little bit if drawing 2D graphics.
// http://dev.processing.org/bugs/show_bug.cgi?id=95
// This hack fixes a bug caused by numerical precision issues when
// applying the 3D transformations to coordinates in the screen plane
// that should actually not be altered under said transformations.
// It will not be applied if any transformations other than translations
// are active, nor should it apply in OpenGL mode (PGraphicsOpenGL
// overrides render_lines(), so this should be fine).
// This fix exposes a last-pixel bug in the lineClipCode() function
// of PLine.java, so that fix must remain in place if this one is used.
// Note: the "true" fix for this bug is to change the pixel coverage
// model so that the threshold for display does not lie on an integer
// boundary. Search "diamond exit rule" for info the OpenGL approach.
if (drawing2D() && a[MZ] == 0) {
a[X] += 0.01;
a[Y] += 0.01;
a[VX] += 0.01*a[VW];
a[VY] += 0.01*a[VW];
b[X] += 0.01;
b[Y] += 0.01;
b[VX] += 0.01*b[VW];
b[VY] += 0.01*b[VW];
}
// end 2d-hack
line.reset();
line.setIntensities(a[SR], a[SG], a[SB], a[SA],
@@ -1519,19 +1577,21 @@ public class PGraphics3D extends PGraphics {
// figure out which dimension is the perpendicular axis
boolean foundValidX = false;
boolean foundValidY = false;
for (int i = vertex_start; i < vertex_end; i++) {
if (vertices[i][MX] != 0) foundValidX = true;
if (vertices[i][MY] != 0) foundValidY = true;
for (int j = i; j < vertex_end; j++){
if ( vertices[i][MX] != vertices[j][MX] ) foundValidX = true;
if ( vertices[i][MY] != vertices[j][MY] ) foundValidY = true;
}
}
if (foundValidX && !foundValidY) {
if (foundValidX) {
//d1 = MX; // already the case
d2 = MZ;
} else if (!foundValidX && foundValidY) {
} else if (foundValidY) {
// ermm.. which is the proper order for cw/ccw here?
d1 = MY;
d2 = MZ;
} else {
// screw it, this polygon is just f-ed up
return;
@@ -3055,6 +3115,16 @@ public class PGraphics3D extends PGraphics {
*/
public void frustum(float left, float right, float bottom,
float top, float znear, float zfar) {
//if (hints[ENABLE_ACCURATE_TEXTURES]){
//These vars are only needed if accurate textures are used, however,
//there is the possibility that accurate texturing will only be turned
//on after the perspective matrix has already been set, so we might as
//well store these no matter what since it's not much overhead.
leftScreen = left; rightScreen = right; bottomScreen = bottom; topScreen = top;
nearPlane = znear;
//}
//System.out.println(projection);
projection.set((2*znear)/(right-left), 0, (right+left)/(right-left), 0,
0, (2*znear)/(top-bottom), (top+bottom)/(top-bottom), 0,
@@ -3071,6 +3141,41 @@ public class PGraphics3D extends PGraphics {
}
/*
* This function checks if the modelview matrix is set up to likely be
* drawing in 2D. It merely checks if the non-translational piece of the
* matrix is unity. If this is to be used, it should be coupled with a
* check that the raw vertex coordinates lie in the z=0 plane.
* Mainly useful for applying sub-pixel shifts to avoid 2d artifacts
* in the screen plane.
* Added by ewjordan 6/13/07
*
* TODO need to invert the logic here so that we can simply return
* the value, rather than calculating true/false and returning it.
*/
private boolean drawing2D() {
if (modelview.m00 != 1.0f ||
modelview.m11 != 1.0f ||
modelview.m22 != 1.0f || // check scale
modelview.m01 != 0.0f ||
modelview.m02 != 0.0f || // check rotational pieces
modelview.m10 != 0.0f ||
modelview.m12 != 0.0f ||
modelview.m20 != 0.0f ||
modelview.m21 != 0.0f ||
!((camera.m23-modelview.m23) <= EPSILON &&
(camera.m23-modelview.m23) >= -EPSILON)) { // check for z-translation
// Something about the modelview matrix indicates 3d drawing
// (or rotated 2d, in which case 2d subpixel fixes probably aren't needed)
return false;
} else {
//The matrix is mapping z=0 vertices to the screen plane,
// which means it's likely that 2D drawing is happening.
return true;
}
}
//////////////////////////////////////////////////////////////

9
core/src/processing/core/PLine.java
View File

@@ -484,8 +484,13 @@ public class PLine implements PConstants
int xmax = SCREEN_WIDTH1;
int ymax = SCREEN_HEIGHT1;
return ((yi < ymin ? 8 : 0) | (yi > ymax ? 4 : 0) |
(xi < xmin ? 2 : 0) | (xi > xmax ? 1 : 0));
//return ((yi < ymin ? 8 : 0) | (yi > ymax ? 4 : 0) |
// (xi < xmin ? 2 : 0) | (xi > xmax ? 1 : 0));
//(int) added by ewjordan 6/13/07 because otherwise we sometimes clip last pixel when it should actually be displayed.
//Currently the min values are okay because values less than 0 should not be rendered; however, bear in mind that
//(int) casts towards zero, so without this clipping, values between -1+eps and +1-eps would all be rendered as 0.
return ((yi < ymin ? 8 : 0) | ((int)yi > ymax ? 4 : 0) |
(xi < xmin ? 2 : 0) | ((int)xi > xmax ? 1 : 0));
}

1387
core/src/processing/core/PTriangle.java
View File

File diff suppressed because it is too large Load Diff

20
core/todo.txt
View File

@@ -91,6 +91,16 @@ X http://processing.org/discourse/yabb_beta/YaBB.cgi?board=Syntax;action=displ
X significant improvement to text and images in opengl
X now using mipmaps to interpolate large and small images
ewjordan stuff (changes checked in)
_ rect() changes size as it changes position
_ http://dev.processing.org/bugs/show_bug.cgi?id=95
_ strange texture warping in P3D
_ http://dev.processing.org/bugs/show_bug.cgi?id=103
_ lines skip on 200x200 surface because of fixed point rounding error
_ http://dev.processing.org/bugs/show_bug.cgi?id=267
_ this may be same as 95
_ Polygons parallel to z-axis not always filling with nonzero x or y
_ http://dev.processing.org/bugs/show_bug.cgi?id=547
_ PGraphics problem with fillColor
_ http://dev.processing.org/bugs/show_bug.cgi?id=468
@@ -101,12 +111,6 @@ _ http://dev.processing.org/bugs/show_bug.cgi?id=560
_ replaceAll() not supported by 1.1
_ http://dev.processing.org/bugs/show_bug.cgi?id=561
ewjordan stuff
_ Polygons parallel to z-axis not always filling with nonzero x or y
_ http://dev.processing.org/bugs/show_bug.cgi?id=547
_ strange texture warping in P3D
_ http://dev.processing.org/bugs/show_bug.cgi?id=103
_ loadBytes() doesn't do auto gunzip? but loadStrings and createReader do?
_ this might be a good solution for dealing with the differences
_ with loadBytes(), they can use gzipInput (or loadBytesGZ?)
@@ -777,10 +781,6 @@ CORE / PGraphics3D
_ images are losing pixels at the edges
_ http://dev.processing.org/bugs/show_bug.cgi?id=102
_ rect() changes size as it changes position
_ http://dev.processing.org/bugs/show_bug.cgi?id=95
_ lines skip on 200x200 surface because of fixed point rounding error
_ http://dev.processing.org/bugs/show_bug.cgi?id=267
_ P3D not doing bilinear interpolation in text and images
_ because smooth() has to be set (and smooth throws an error in P3D)
_ how should this be handled? a hint? allowing smooth()?