 |
Author |
Topic: map an image to a mesh (Read 2802 times) |
|
_C
|
map an image to a mesh
« on: Nov 14th, 2004, 7:49pm » |
|
hi everybody,
i want to map an image onto a polygon mesh
while creating the mesh following a [][]
pointarray. has someone done that so far?
this would help me a lot further!
thanks for any hints
c
|
|
|
|
|
_C
|
Re: map an image to a mesh
« Reply #2 on: Nov 14th, 2004, 11:34pm » |
|
hi fjen,
thank you, i knew this one, nevertheless it remains
hard to imagine what the u, v thing does.
it works to map an image onto a mesh, problems remai-
ning are: how to get the positions of an images` points with certain colors into an [][]array, not
using a Vector for that since it is hard to handle
with the iteration things.
bye
c
|
|
|
|
|
fjen
|
Re: map an image to a mesh
« Reply #3 on: Nov 15th, 2004, 1:29am » |
|
think of them as being x and y coordinates in the texture-2d-space. so every point in your mesh represents/holds one point in the texture.
how exactly is the data arranged in you [][]array?
/F
|
|
|
|
|
_C
|
Re: map an image to a mesh
« Reply #4 on: Nov 15th, 2004, 11:59am » |
|
so, what i exactly wanted to do is:
reading an transparent image, and put every position
in the image with an alpha higher than zero into
an array. the problem is: how can i get the positions
into a [][]array?
ido: for(...img.width;i++) {
for(...img.height; j++) {
if(alpha(pixels[j][i])>0) {
//and then i need j, i in an [][]
}
}
}
pixels beeing an array with the pixels, i only
want to have the indices. strange thing.
c
|
|
|
|
|
fjen
|
Re: map an image to a mesh
« Reply #5 on: Nov 15th, 2004, 5:42pm » |
|
i don't seem to get the big picture.
why is it you need the [][] to hold the same indices as in pixels? this is just more or less duplicating pixels[][] in the end ... because if you'd have alpha==0 at j = 10 and i =100 then the_array[][] would be of size 10*100. get it?
to read this value you'd have to run thru the_array[][] the same way as you do right now with pixels[][], so in the end you could just run thru pixels itself and test for alpha==0 again.
maybe you could give a little more insight of what you are trying to do? some code maybe?
i just read your original post, and i think you're just on the wrong way in creating these points(array) (from the image ?) ... using a [][] array for a mesh makes great sense if all the rows are of same length, but that's not what you are gonna end up with doing it the way you've tried.
/F
|
|
|
|
|
_C
|
Re: map an image to a mesh
« Reply #6 on: Nov 15th, 2004, 7:56pm » |
|
hi fjen,
you are right, this doesnt really make sense.
the point was just thut later on, i maybe wanted
to change the z values of the points i got out
of the picture, besides, it was the aim to have
an array with 'non alpha coordinates', so that
i could later on draw the image with what i want,
eg boxes, rects and so on.
so, if i wanted to create the array, i would do it
like this
for(int i=0; i<pixels.length; i++) {
for(int j<pixels.width; j++) {
if(alpha(pixels[j][i]>0) {
newarray[j][i] = j*i;
}
}
}
?
heres how i did it:
Code:
int imgwidth, imgheight;
int res=30;
float d=1;
float zValues[][];
BImage img;
float rtx, rtz;
int imgArray[][];
int t, tt;
float noiseScale, noiseScale2;
color strokeColors[][];
void setup() {
img=loadImage("9045welt_outlined.gif"); imageMode(CENTER_DIAMETER);
imgArray = new int[img.height][img.width];
for(int i=0; i<img.width; i++) {
for(int j=0; j<img.height; j++) {
imgArray[j][i] = img.pixels[j*img.height+i];
}
}
imgwidth = img.width;
imgheight= img.height;
zValues = new float[imgheight][imgwidth];
for(int i=0; i<zValues.length; i++) {
for(int j=0; j<zValues[0].length; j++) {
noiseScale+=0.0015; noiseScale2+=0.003;
zValues[i][j] = noise(noiseScale, noiseScale2)*50 ;//noise(noiseScale, noiseScale2);
}
}
zValues[10*res][15*res] = 200;
zValues[16*res][15*res] = -200;
strokeColors = new color[imgheight][imgwidth];
getStrokeColor();
size(1024, 768);
background(150);
}
void rotation() {
rtx+=((mouseY-height/2)*0.5-rtx)*0.001;
rtz+=((mouseX-width/2)*0.5-rtz)*0.001;
rotateX(rtx); rotateZ(rtz);
}
void loop() {
clear();
mesh();
}
void mesh() {
push();
translate(width/2, height/2, 0);
scale(0.8);
rotation();
for(int i=0; i<imgwidth-res; i+=res) {
beginShape(TRIANGLE_STRIP);
for(int j=0; j<imgheight; j+=res) {
t = int(zValues[j][i] * d);
tt = int(zValues[j][i+res] * d);
stroke(255,255,255,12);
texture(img);
vertex(i-imgwidth/2, j-imgheight/2,t,i,j);
vertex(i+res-imgwidth/2, j-imgheight/2, tt, i+res, j);
}
endShape();
}
pop();
}
void getStrokeColor() {
for(int i=0; i<img.width-res; i++) {
for(int j=0; j<img.height-res; j++) {
if(alpha(imgArray[j][i])>0) {
strokeColors[j][i] = color(100,100,100);
}
else strokeColors[j][i] = color(255,255,255,0);
}
}
}
|
|
|
|
|
|
|
fjen
|
Re: map an image to a mesh
« Reply #7 on: Nov 15th, 2004, 9:36pm » |
|
what i'd do is:
create a simple class to hold you coordinates & alpha:
class Point3D
{
float x, y, z, u, v;
int alpha;
Point3D( float x, float y, float z, float u, float v, int alpha)
{
this.x = x; this.y = y; ....
}
}
create an array to store the points:
int wres = img.width/res;
int hres = img.height/res;
Point3D points = new Point3D[wres*hres];
for (int wx=0; wx <wres; wx++)
{
for (int hy=0; hy <hres; hy++)
{
// calculate x, y, noise and u, v here ...
int a = alpha( img.pixels[wx*res + (hy*res) * img.width] );
points[wx+hy*wres] = new Point3D(wx, hy, your_noise, u, v, a);
}
}
now you don't have to recalculate eveything in the loop ..
to create/draw your mesh simply use somehting like:
vertex(points[0].x, points[0].y, points[0].z, points[0].u, points[0].v);
i have not tested the code ... just for illustration.
/F
|
|
|
|
|
_C
|
Re: map an image to a mesh
« Reply #8 on: Nov 16th, 2004, 1:58am » |
|
thank you! that was fine. didnt try the code,
but i will when i can keep my eyes opened again 
c
|
|
|
|
|
_C
|
Re: map an image to a mesh
« Reply #9 on: Nov 16th, 2004, 2:02am » |
|
..the reason why i tried to store it in a matrix
like [][] is that later on, i may would have to
define exact positions on a mesh, and it was easy
for me to imagine where the points were when i can
imagine them at a specific point on the matrix, for
your consideration.
|
|
|
|
|
fjen
|
Re: map an image to a mesh
« Reply #10 on: Nov 16th, 2004, 3:18am » |
|
you still can store the points3d in your 2d-array. multi-dimensional arrays can be pretty not-fun ... but in your case it's gonna be static(*) i think ...
so as a little guten-morgen-gutsele:
Point3D[][] points = new Point3D[wres][hres];
for (int wx=0; wx <wres; wx++)
{
for (int hy=0; hy <hres; hy++)
{
// calculate x, y, noise and u, v here ...
int a = alpha( img.pixels[wx*res + (hy*res) * img.width] );
points[wx][hx] = new Point3D(wx, hy, your_noise, u, v, a);
}
}
sleep well,
/F
(*) static not in a java sense ... not to be resized later i ment.
|
| « Last Edit: Nov 16th, 2004, 3:19am by fjen » |
|
|
|
|
_C
|
Re: map an image to a mesh
« Reply #11 on: Nov 16th, 2004, 8:45pm » |
|
thats great. but could you explain what you meant
with 'guten-morgen-gutsele'?
|
|
|
|
|
fjen
|
Re: map an image to a mesh
« Reply #12 on: Nov 16th, 2004, 10:04pm » |
|
good-morning-cookie ... thought you were from germany because the image you used in your code was called welt-something ... /F
|
| « Last Edit: Nov 16th, 2004, 10:04pm by fjen » |
|
|
|
|
_C
|
Re: map an image to a mesh
« Reply #13 on: Nov 17th, 2004, 2:59pm » |
|
ahh, that feeds the dog in me
|
|
|
|
|
fjen
|
Re: map an image to a mesh
« Reply #14 on: Nov 17th, 2004, 5:49pm » |
|
tips & treats that is ...
|
|
|
|
|
|
