TLDR: I want to rotate camera but render sprites in regards to their world position not camera position.
Howdy,
I'm currently using LibGDX and have come across an issue in regards to camera/object rotation.
Say I have my camera with a rotation of 0 and I have an object(sprite) to the left of the camera's center.
i.e.
Camera Normal (0 degrees rotation)
The sprite renders fine when given a standard world coordinate, however once I rotate my camera, that world coordinate differs from the camera's new (x, y) values.
If I then rotate my camera smoothly 90 degrees to the right(clockwise so that the up direction is facing to the right like the picture below), the object(sprite) that used to be on the left should have simulated a left rotation in regards to the camera (the rotation happens via the camera, the sprite just needs to update position) and now be below the camera's center point.
i.e.
Camera Rotated (90 degrees clockwise)
I'm confused as to how I would calculate the sprite's new locations/positions during the smooth rotation.
Cheers,
Solist.
After looking everywhere for a solution to this problem for 3 weeks it was merely a matter of me needing to call the method
batch.setProjectionMatrix(camera.combined);
in order to update the sprites to their new position in regards to the changing camera rotation.
This link here explains how the Projection Matrix works.
Related
I am creating a 3d text on blender2.8 but I am having a hard time trying to make it appear on the camera view.
I have tried rotating and scaling the camera but it doesn't help.
The camera object is the focal point for the final image, like in real life, if you want to get more in the picture you need to move the camera back or zoom out.
With the camera close -
With the camera far away -
To zoom out, you reduce the camera's focal length.
below you can see the result of the optical flow if a camera makes a translation movement. If the camera makes a roll rotation the result looks like the second picture. Is it possible to retrieve the yaw angle from a camera if its only rotation around the yaw axis?
I think in the optical flow you can recognize if the camera is rotating around the yaw axis (z-axis), but i don't know how to retrieve the information how much the cam has rotated.
I would be gradeful for any hints. Thanks
Translation:
Roll rotation:
Orientation of camera:
If you have a pure rotation of your cam then you can use findhomography. You need four point correspondence in your pictures. For a pure rotation the homography matrix is already a rotation matrix. Otherwise you need to decompose the homograohy matrix. For a camera movement off 6 dof you can use the function find essential matrix and decompose this to translation and rotation.
While looking through the tutorials I've seen the Ogre::Camera::getCameraToViewportRay method being used. I was trying understand what it does.
First I imagine a viewport, being placed somewhere in the 3D scene, let's say on the screen of the TV object. I can easily imagine how to transform the 2D coordinate on the viewport to the 3D coordinate of the scene and then to make a ray from the camera position point through that point on the VP.
But I can not understand how it's done when the VP is on the the RenderWindow(on my monitor). I mean, where is the render window in the scene, where is the point on the renderwindow's VP in the scene? How is the point on the renderwindow's VP transformed into a 3D point of the scene?
Thanks for answer!
The viewport shows what you see through a camera, but the viewport is in front of the camera.
There is a stackoverflow post with information about the relation of camera and viewport and a nice visual illustration: https://stackoverflow.com/a/7125486/2168872
The camera to viewport ray is a worldspace ray, starting from your camera and intersecting the viewport at a certain point, e.g. where your mouse cursor points to.
I've just started with Three.js. Like really just now.
After playing with it for an hour or so and building a tool that helps me understand how the different elements work together (Camera, Light, Objects), I found something strange.
The tool: http://hotblocks.nl/tests/three/cubes.html
This is the current default set up:
the Camera is positioned 210 upwards and
500 backwards and
246 to the right
the Camera is rotated slightly to the left
the light is directly above and shines in all directions
As you can see, the objects are at the very bottom of the viewport. So I want to turn the Camera downward, so I can see more of them.
Try that: turn camera.rotation.x down.
That works, but the angle of rotation is wrong! Instead of the Camera rotating, it's the World rotating around its Z axis.
That's not right, is it?
The Y axis is also wrong. It rotates the World around its Y axis.
Rotating the Camera around its Z axis, works perfectly: the Camera rotates, not the World.
Am I doing it wrong? Or understanding it wrong?
PS Since the Camera rotation is only around its Y axis, the objects' vertical edges should be vertical in the result as well. In the default set up, they are. Rotating the camera around its X axis, shouldn't change that, but it does. Only rotating around its Z axis should change that (and it does). Am I wrong?
PPS I know about Camera.lookAt( THREE.Vector3 target ), but that changes the rotation of the camera, including its Z axis, and that shouldn't be necessary, logically.
Answer received on Github: https://github.com/mrdoob/three.js/issues/1163
I have a sprite and if it is touched the touch should be recognized. I used the coordinates to do so. I took the coordinates (min x, min y, max x , max y)of the sprite image. But The sprite image is not a rectangular shape. So, even if I touch the coordinates outside the sprite and inside the rectangular bounds the sprite is recognized.
But for my application I need only the sprite to be recognized. So, I have to take only the coordinates of the sprite, but it is not regular shape. I am using CCSprite in my program.
So, what can I do to for only the sprite to be selected ? Which classes should use for this?
Thank You.
You could try one of the following...
Create a bounding box smaller than the absolute extents of the sprite image. Yes it will be smaller than the sprite. This will eliminate the dead space click detection of the sprite the trade off being parts of your sprite which look selectable won't be
Use a circular bounding area to detect if the user has clicked on your sprite. Again you will have the dead space problem in my first suggestion but the sphere may give you some better coverage area over the sprite giving you better results on touch detection
This is a standard problem in physics collision detection systems which often end up using circles or rectangles as their collision bodies. I would go with the either a circle or rectangle smaller than the size of your sprite as your bounding area. Going finer detail than that you could generate bounding area polygons. This would however introduce a whole bunch of new issues and concerns.
I am building a Cocos2D game right now and what I am doing is first I step through my sprites and see which sprites the touch hit (they overlap in my app)
Then, for each sprite hit I use [sprite convertTouchToNodeSpace] to get an X,Y co-ordinate inside the sprite, which I can use (although the Y axis is flipped) to reference the CGImage I created the sprite with.
If the pixel at the touch point is 'clear' ie alpha 0, then the sprite was not really touched, and I check the next sprite in the z-order to see if it has color where it was touched.
Sometimes I think I should be using a two color mask image to go along with each sprite, not the sprite image. But, I am mr. make it work, then make it fast.
I realise this is not super efficient, but I do not have very many sprites and I do this only for touches.