I'm a beginner to DirectX, so the solution may seem obvious to you. I'm following this tutorial teaching me how to set up a simple directx project with no starting code, and so far the tutorials are working perfectly. The problem is now that I have it clearing the backbuffer in my Render() function, it will not set the background to any color but black.
Here's my Render() function:
//set render target
devcon->OMSetRenderTargets(1, rendertarget.GetAddressOf(), nullptr);
//clear backbuffer to blue
float color[4] = { 0.0f, 0.2f, 0.4f, 1.0f };
devcon->ClearRenderTargetView(rendertarget.Get(), color);
//TODO: Render code here
//swap buffers
swapchain->Present(1, 0);
and my setup in the initialization:
//get a ptr to the backBuffer
ComPtr<ID3D11Texture2D> backbuffer;
swapchain->GetBuffer(0, __uuidof(ID3D11Texture2D), &backbuffer);
//create a render target pointing to the backBuffer
dev->CreateRenderTargetView(backbuffer.Get(), nullptr, &rendertarget);
Your code for clearing the back buffer to blue is correct.
There must be a type or failure somewhere else.
Related
I was playing around with spotLight in libGDX. I used this code to project the light source in -y direction. To my surpise, my floorModel was completely black. I went to docs and found this interesting fact:
Note that the default shader doesn't support spot lights, you'll have to supply your own shader to use this class.
I think this may be the reason the what spotlight isn't working. I have however found 0 examples of setting up spot light properly. Could you provide a simple example on setting up spot light properly?
EDIT for those who read the first question: after exporting custom object for a second time, the light interaction was working as exptected-meaning the object was not visible when only spotlight was used. Not sure why.
public void show() {
float factor=3f;
camera = new PerspectiveCamera(45, Gdx.graphics.getWidth(),Gdx.graphics.getHeight());
camera.position.set(1.4f * factor, 6f, -1f * factor);
camera.lookAt(0f,0f,0f);
camera.near =0.1f;
camera.far = 300f;
modelBatch = new ModelBatch();
modelBuilder=new ModelBuilder();
UBJsonReader jsonReader = new UBJsonReader();
G3dModelLoader modelLoader = new G3dModelLoader(jsonReader);
model = modelLoader.loadModel(Gdx.files.getFileHandle("convertedModel3.g3db", Files.FileType.Internal));
floorModel = modelBuilder.createBox(30,1,30,new Material(ColorAttribute.createDiffuse(Color.BLUE)), VertexAttributes.Usage.Position|VertexAttributes.Usage.Normal);
modelInstance = new ModelInstance(model);
floorInstance=new ModelInstance(floorModel);
Array<Material> materials=modelInstance.materials;
materials.get(0).set(ColorAttribute.createDiffuse(Color.WHITE));
materials.get(0).set(ColorAttribute.createReflection(Color.WHITE));
float intensity = 0.1f;
environment = new Environment();
Vector3 pos = new Vector3(0,10,0);
Vector3 dir = new Vector3(0,-10,0);
environment.add(new SpotLight().set(Color.WHITE,pos,dir,1f,100,1));
}
#Override
public void render(float delta) {
Gdx.gl.glViewport(0, 0, Gdx.graphics.getWidth(), Gdx.graphics.getHeight());
Gdx.gl.glClearColor(0, 0, 0, 1);
Gdx.gl.glClear(GL20.GL_COLOR_BUFFER_BIT | GL20.GL_DEPTH_BUFFER_BIT);
camera.update();
modelBatch.begin(camera);
modelBatch.render(modelInstance, environment);
modelBatch.render(floorInstance,environment);
modelBatch.end();}
As the documentation says, currently spotlight is not supported by the default shader. This is because good generic spotlight requires per fragment lighting and the default shader uses per vertex lighting. Therefor you wont find a simple example on that. You could wait for this pull request to be included. You could use that as an example (although it's probably an overkill for your usage and surely not simple). Or you could provide your own shader that includes spot lighting.
Here's a basic tutorial on how to create a shader. This wiki page contains more information about how to customize the shader. The actual shader implementation is not really libGDX specific (except for the uniform names), you can use any algorithm you like that is compatible with your target device.
Here you can find an example on how to adjust the default (per vertex lighting) shader to somewhat support spotlight (the quality depends on your models). Note that the modification to the DefaultShader class are already included, you'd only have to focus on the actual shader (GLSL) programs in that case.
I have a simple canvas animation, as follows:
function animate(){
var canvas = document.getElementById("canvas");
canvas.width = $(window).width();
canvas.addEventListener("click", replay, false);
var context = canvas.getContext("2d");
//animation code
function replay(e){
animate();
}
}
So my expectation is when the user clicks the canvas the animation will replay because I am reassigning the width:
canvas.width = $(window).width();
which will reset the entire context state ( read it here http://diveintohtml5.info/canvas.html)
It works the first time you click the canvas but after that it remembers the context transformation state and shows some weird animation.
I tried adding this:
context.setTransform( 1, 0, 0, 1, 0, 0 );
to explicitly reset the transformation matrix to identity matrix before re-drawing but has no effect.
It is different from this How to clear the canvas for redrawing
because it is about how to clear the display of context but I want to clear everything the display plus the context state.
I tried logging the state of each variable during the animation I get something weird with the ff code.
function moveHorizontal(x){
if(x < 100 ){
context.translate(1, 0);
console.log("x:"+x);
drawImage();
x += 1;
setTimeout(moveHorizontal,10,x);
}
}
I am calling this function initially as moveHorizontal(0).
so the log is as expected the first time:
x:0
x:1
x:2
.
.
x:99
when I click "replay" I get same thing as above:
x:0
x:1
x:2
.
.
x:99
which is correct but when I click "replay" the second time
I am getting this:
x:0
x:0
x:1
x:1
.
.
.
x:99
x:99
which triples the translation leading to unexpected effect.
any explanation?
Update
With the new information this may be the cause of the problem (there is not enough code to be 100% sure but it should give a good pointer as to where the problem is):
As the code is asynchronous and seem to share a global/parent variable x, what happens is that when, at some point, the replay is invoked before one earlier has stopped, the x is reset to 0 causing both (or n number of loops) to continue.
To prevent this you either need to isolate the x variable for each run, or using a simpler approach blocking the loop from running if it's already running using x from a previous invocation. You can use a flag to do this:
var x = 0,
isBusy = false;
function animation(x) {
if (isBusy) return;
if (x < 100) {
isBusy = true;
// ...
setTimeout(...); // etc.
}
else {
isBusy = false;
}
}
optionally, store the timeout ID from setTimeout (reqID = setTimeout(...)) and use clearTimeout(reqID) when you invoke a new loop.
Old answer
The only way to reset the context state entirely is to:
Set a new dimension
Replace it with a new canvas (which technically isn't resetting)
The latter should be unnecessary though.
According to the standard:
When the user agent is to set bitmap dimensions to width and height,
it must run the following steps:
Reset the rendering context to its default state.
...
In the example code in your post you are using a mix of jQuery and vanilla JavaScript. When working with canvas (or video element) I always recommend using vanilla JavaScript only.
Since you are not showing more of the code it is not possible for us to pinpoint to any specific error. But as neither setting an identity matrix nor setting a size for canvas seem to work, it indicates that the problem is in a different part of the code. Perhaps you have persistent offsets involved for your objects that are not reset.
Try to eliminate any possible errors coming from using jQuery to set width as well:
canvas.width = window.innerWidth; // and the same for height
Two other things I notice is that you set the size of canvas inside a function that appears to be part of a loop - if this is the case it should be avoided. The browser will have to, in essence, build a new canvas every time this happens. Second, this line:
canvas.addEventListener("click", replay, false);
can simply be replaced with (based on the code as shown):
canvas.addEventListener("click", animate, false);
You will need to post more code to get a deeper going answer though. But as a conclusion in regards to context: setting a new size will reset it. The problem is likely somewhere else in the code.
I'm attempting to use LibGDX Stage for my GUI elements, but I'm having loads of difficulty getting it to render properly. Right now, I'm attempting to render a chat window in the lower left corner of the screen.
Here is the construction of the GUI objects:
stage = new Stage();
stage.setCamera(controller.getCamera());
uiSkin = new Skin(Gdx.files.internal("res/gui/skin/uiskin.json"));
Table table = new Table(uiSkin);
stage.addActor(table);
table.setSize(300, 260);
table.setPosition(0, 0);
table.setFillParent(false);
table.bottom();
table.left();
table.pad(10);
lblChatLabel = new Label("", uiSkin);
lblChatLabel.setWrap(true);
final ScrollPane scroll = new ScrollPane(lblChatLabel, uiSkin);
txtChatBar = new TextField("do a chat", uiSkin);
txtChatBar.setName("txtChatBar");
table.add(txtChatBar).width(300f);
table.row();
table.add(scroll).expand().fill().colspan(4);
txtChatBar.addListener(new InputListener() {
public boolean keyDown(InputEvent event, int keycode) {
if (keycode == Input.Keys.ENTER) {
sendMessage();
// Close the chat bar
showChat = false;
return true;
}
return false;
}
});
And here is my render() method:
Log.debug("void", "x: " + stage.getCamera().position.x + ", y: " + stage.getCamera().position.y);
stage.act();
for (Actor a : stage.getActors()) {
a.draw(spriteBatch, 1);
}
In another section of code elsewhere, the game's camera object is translated to center on the player:
camera.translate(targetPosition.x, targetPosition.y);
camera.update();
So that all said, the chat window renders properly, but it only does so at 0, 0. I've also changed called stage.draw() rather than iterate through the Actors individually, but that causes the same issue. Here is a screenshot illustrating the issue:
http://i.imgur.com/8uz5lV6.jpg
Finally, I've tried to translate the stage manually by setting the viewport, but that causes an even weirder issue.
float cx = controller.getCamera().getX();
float cy = controller.getCamera().getY();
float width = controller.getCamera().viewportWidth;
float height = controller.getCamera().viewportHeight;
stage.act();
stage.setViewport(width, height, true, cx, cy, width, height);
stage.draw();
Image here:
http://i.imgur.com/JpSLq1s.jpg
Certainly I am doing something wrong, but I have no idea at this point. I would have assumed that the stage follows the Camera translation, but that doesn't appear to be the case. Any suggestions are welcome! Thanks!
I believe the issue is this line of code:
stage.setCamera(controller.getCamera());
If I'm reading correctly, you want the chat window to always render from (0,0), no matter where the camera is on the screen. If that's the case, then the stage shouldn't have any relation with the camera, which moves around and just further complicates getting the stage in the right position to be rendered properly.
Without that line of code, you should be able to just call
stage.act();
stage.draw();
and it should work fine.
I'm new to the canvas tag and am playing around with some animation. Basically, I'm trying to setup a "ground" section composed of multiple images (similar to an 8bit side scroller game like Mario Brothers). The ground will be composed of multiple images, which I've built a constructor function to load these and tile them across the bottom.
function Ground(context,ImageName,ImgX,ImgY,ImgW,ImgH){
this.width=ImgW;
this.height=ImgH;
this.x=ImgX;
this.y=ImgY;
img=new Image();
img.onload=function(){context.drawImage(img,ImgX,ImgY,ImgW,ImgH);};
img.src='images/'+ImageName;};
This seems to work out just fine. I've then setup the rest of the animation, including a basic setup for Key Left/Right events, like so:
window.onload=function(){
var canvas=document.getElementById('canvas'),
context=canvas.getContext('2d'),
Grounds=[],
ImgX=-150; // INITIAL STARTING X FOR FIRST GROUND TILE
// INSERT GROUND ELEMENTS
for(var i=0,l=8; i<l; i++){
var ImgX+=150;
Grounds[i]=new Ground(context,"ground.png",ImgX,650,150,150);
};
// ASSIGN LEFT/RIGHT KEYS
window.addEventListener('keyup',function(event){
switch(event.keyCode){
case 37:
for(var i=0,l=Grounds.length; i<l; i++){
Grounds[i].x+=10;
};
break;
case 39:break;
};
});
// ANIMATION LOOP
(function drawFrame(){
window.mozRequestAnimationFrame(drawFrame,canvas);
context.clearRect(0, 0, canvas.width, canvas.height);
}());
};
I know exactly what my problem is, but don't know how to solve it. The animation loop is clearing the canvas every frame, but not redrawing the updated position (if any) when the user presses the left arrow key.
I'm missing the redraw part here and I'm not exactly sure how to handle this or if I'm approaching this entirely wrong. Any help is very appreciated! Thanks!
First of all you're incrementing the property x of the ground tiles but that property is not even used anywhere in your code. Modify your code so that the onload event of those image objects draws the image according to their own x property so changes to it will actually affect what is drawn. Also add the image object as a property of the Ground object so you can access it later on from outside.
Your approach is really not so good but if you want to do it without going back to 0 do it as follows:
function Ground(context,ImageName,ImgX,ImgY,ImgW,ImgH){
this.width=ImgW;
this.height=ImgH;
this.x=ImgX;
this.y=ImgY;
var self = this; // Add local reference to this Ground instance
this.img=new Image(); // img is now a property too
this.img.onload=function(){context.drawImage(this, self.x, self.y,self.width,self.height);};
this.img.src='images/'+ImageName;};
Ok so now you can change the property x of the ground tiles and call the draw function of it again (which is the onload event).
Grounds[i].x+=10;
Grounds[i].img.dispatchEvent(new Event("load"));
Please note that you should really make the updates of all the values first and then all the draw calls separately.
Can you not just add a draw method? You usually so something like this:
init -> update -> clear, redraw -> update -> clear, redraw -> ...
// ANIMATION LOOP
(function drawFrame(){
window.mozRequestAnimationFrame(drawFrame,canvas);
context.clearRect(0, 0, canvas.width, canvas.height);
contect.drawImage(...);
}());
I have a problem and I have potential solution. But I wanted to confirm if there is an easy and simple way to solve my problem.
App type:
Isometric Game
Problem statement:
I am loading images in my flash app and have mouse events attached to them.
The images I load are prop images like vehicles, trees, buildings etc., and all of them are transparent.
Example: Red ball asset (please ignore the yellow background which I applied to describe the problem)
If I click on the actual image area (colored in red), then every thing works perfect
I don't want to trigger mouseevent when I click on empty image part (or transparent area, which I have shown in yellow color)
There is one way I know by creating masks in flash. I don't want to do it unless that is the final option left because I load image assets instead of flash assets and I don't want to create a new mask asset for all the assets
There is another method I was going to adopt by using getPixel method of Bitmap. Which is discussed here.
But there is another problem with this method.
I might be able to ignore the click event when I click on the empty part of the asset but if there is some other asset is behind the image in the same location, then I need to process the click event for the occluded image.
Well, thinking of solution to this problem takes me to the getObjectsUnderPoint where I can scan the occluded assets
Well, what you proposed as a solution is 100% valid. Just move the logic of determining what game object is clicked outside of that object.
Listen for MOUSE_DOWN/MOUSE_UP events at container which contains your game objects.
Catch an event
Check if the game object which is the target of this event is transparent at this point using BitmapData.getPixel32
If it is use getObjectsUnderPoint to find out all other game objects at this point
Find in a loop the first object which is not transparent at this point
Now you got the actual object which is hit.
One interesting solution is to use Sprite objects with the individual non-transparent pixels burnt onto them.
Suppose this is your Loader "complete" handler:
private function loaderCompleteHandler(event:Event):void
{
// Loader is not our child, we use a Sprite instead (below).
var loader:Loader = Loader(event.target);
var sprite:Sprite = new Sprite();
addChild(sprite);
var w:Number = loader.content.width;
var h:Number = loader.content.height;
// Use transparent bitmap.
var bitmapData:BitmapData = new BitmapData(w, h, true, 0);
bitmapData.draw(loader.content);
// Now burn the image onto the Sprite object, ignoring
// the transparent pixels.
for (var xPos:int = 0; xPos < w; xPos++) {
for (var yPos:int = 0; yPos < h; yPos++) {
var pixel32:uint = bitmapData.getPixel32(xPos, yPos);
var alpha:int = pixel32 >>> 24;
if (alpha != 0) {
sprite.graphics.beginFill(pixel32 & 0xFFFFFF, alpha / 0xFF);
sprite.graphics.drawRect(xPos, yPos, 1, 1);
sprite.graphics.endFill();
}
}
}
}
Essentially you want "empty" pixels that aren't clickable, and fully transparent pixels aren't quite the same thing. With this solution you get empty pixels.
Only problem is that this might be slow. Give it a shot.