问题
I am trying to implement picking in WebGL. I have a lot of objects (around 500), and I'd like each one to be allowed to be picked. In order to do that, I did a loop which assigns a unique colour to each object (cf. picking principle):
for (var i = 0, len = objects.length; i < len; i++) {
framecolors[count++] = i % 256 / 256; //Red
framecolors[count++] = Math.floor(i/256) / 256; //Green
framecolors[count++] = Math.floor(i/(256*256)) / 256; //Blue
}
framecolors
was then used in a classical buffer to check whether each object had a different shade of red. It worked.
Now, I want to use my objects' original colours, AND a framebuffer with the shades of red in the background. I've been through some code, and I am a bit confused.
Here's what I have tried so far.
Function called before picking:
//Creates texture
colorTexture = gl.createTexture();
gl.bindTexture(gl.TEXTURE_2D, colorTexture);
gl.texParameteri(gl.TEXTURE_2D, gl.TEXTURE_MAG_FILTER, gl.NEAREST);
gl.texParameteri(gl.TEXTURE_2D, gl.TEXTURE_MIN_FILTER, gl.NEAREST);
gl.texParameteri(gl.TEXTURE_2D, gl.TEXTURE_WRAP_S, gl.CLAMP_TO_EDGE);
gl.texParameteri(gl.TEXTURE_2D, gl.TEXTURE_WRAP_T, gl.CLAMP_TO_EDGE);
gl.texImage2D(gl.TEXTURE_2D, 0, gl.RGBA, 400, 400, 0, gl.RGBA, gl.UNSIGNED_BYTE, null);
//Creates framebuffer
fb = gl.createFramebuffer();
gl.bindFramebuffer(gl.FRAMEBUFFER, fb);
gl.framebufferTexture2D(gl.FRAMEBUFFER, gl.COLOR_ATTACHMENT0, gl.TEXTURE_2D, colorTexture, 0);
gl.bindTexture(gl.TEXTURE_2D, colorTexture);
gl.enable(gl.DEPTH_TEST);
gl.bindFramebuffer(gl.FRAMEBUFFER, fb);
gl.clear(gl.DEPTH_BUFFER_BIT);
gl.drawArrays(gl.POINTS, 0, vertexPositionBuffer.numItems);
Function called after:
gl.bindFramebuffer(gl.FRAMEBUFFER, null);
gl.bindTexture(gl.TEXTURE_2D, colorTexture);
gl.drawArrays(gl.POINTS, 0, vertexPositionBuffer.numItems);
As you can probably understand, I am not very comfortable with framebuffers, I just don't really get how they work, even though I read a lot about them. I have no idea how I can link framecolors
to the framebuffer. Is there a way?
Thanks, R.
回答1:
A framebuffer is a collection of attachments (renderbuffers and/or textures). It works just like rendering without a framebuffer. (in fact the browser is internally using a framebuffer to implement WebGL's canvas)
In your case you're missing a few things. You most likely need a depth buffer attached otherwise when you render your scene you won't get zBuffering and the wrong objects will appear in front.
// create renderbuffer
depthBuffer = gl.createRenderbuffer();
gl.bindRenderbuffer(gl.RENDERBUFFER, depthBuffer);
// allocate renderbuffer
gl.renderbufferStorage(
gl.RENDERBUFFER, gl.DEPTH_COMPONENT16, width, height);
// attach renderebuffer
gl.framebufferRenderbuffer(
gl.FRAMEBUFFER,
gl.DEPTH_ATTACHMENT,
gl.RENDERBUFFER,
depthBuffer);
In general you should also check your framebuffer works. After attaching all the attachments you call gl.checkFramebufferStatus
if (gl.checkFramebufferStatus(gl.FRAMEBUFFER) != gl.FRAMEBUFFER_COMPLETE) {
alert("this combination of attachments does not work");
return;
}
A framebuffer can be incomplete for any number of reasons. The most common is the attachments are not same size or the GPU doesn't support that combination of attachments. Note: In WebGL certain combinations are required to work but you since you might change the code later to use different formats it's probably still a good idea to check.
You also need to set the viewport by calling gl.viewport
whenever you switch framebuffers.
gl.bindFramebuffer(gl.FRAMEBUFFER, someFramebuffer);
gl.viewport(0, 0, someFramebufferWidth, someFramebufferHeight);
That includes putting it back when setting things back to the canvas
gl.bindFramebuffer(gl.FRAMEBUFFER, null); // render to canvas
gl.viewport(0, 0, gl.canvas.width, gl.canvas.height);
Finally there's a bug in the code above in that you're only clearing the depth buffer of the framebuffer where you call gl.clear
. You want to call
gl.clear(gl.COLOR_BUFFER_BIT | gl.DEPTH_BUFFER_BIT);
Since you're going to read the colors later otherwise old colors will be left over.
Finally, and I guess you know this. You figure out which pixel corresponds to the mouse click and call
var colorPicked = new Uint8Array(4);
gl.readPixels(pickX, pickY, 1, 1, gl.RGBA, gl.UNSIGNED_BYTE, colorPicked);
Note that when you call gl.readPixels
you must have your framebuffer bound with gl.bindFramebuffer
or gl.readPixels
will read from the canvas.
回答2:
Thanks a lot gman. Just to complete your answer, about how to use framecolors
with my framebuffer, it's quite simple. In the shaders, I added a variable dedicated to the framebuffer's colors:
if (!offscreen) {
gl_FragColor = normalColors;
}
else {
gl_FragColor = frameColors;
}
Now, before determing which object I selected thanks to the readPixels()
function, I switch back to the framebuffer using:
function renderFrame() {
gl.bindFramebuffer(gl.FRAMEBUFFER, fb); //fb = framebuffer
gl.uniform1i(shaderProgram.uOffscreen, true); //uOffscreen = uniform boolean in shader
draw(); //function called to draw objects (contains gl.clear and gl.viewport)
}
and similarly, I switch back to the usual buffer after calling the function.
I am trying to display the framebuffer instead of the usual buffer when the mouse is clicked. I am having troubles, but I'll post the solution later if I ever find it.
EDIT : Solution : just remove the depth-buffer (render buffer) associated with the framebuffer. Then display the buffer as explained in : WebGL display framebuffer?
来源:https://stackoverflow.com/questions/21841483/webgl-using-framebuffers-for-picking-multiple-objects