Author Topic: Scene Rotation  (Read 12508 times)

stormp

• byte
• Posts: 38
Scene Rotation
« on: October 28, 2007, 07:02:12 am »
Hi :-)

I'm looking for some pointers on how to mimic scene rotation similar to how GoogleEarth does.  IE: grabbing a point on a (imaginary) sphere and dragging that point around.

I found this article: http://www.mactech.com/articles/mactech/Vol.15/15.03/NaturalObjectRotation/index.html

Is there any similar examples in jPCT?

Appreciate any help!

Thanks.

S.
« Last Edit: October 28, 2007, 07:09:02 am by stormp »

hthth

• byte
• Posts: 13
Re: Scene Rotation
« Reply #1 on: October 28, 2007, 03:12:52 pm »
I don't remember having come across any examples that show it, and I don't have the code at hand. But it should be relatively easy to implement, have a look at the following methods in Object3D:

Quote
rotateX(float w), rotateY(float w) & rotateZ(float w)

Rotates the object's rotation matrix around the x-axis by the given angle w (radian, counter clockwise for positive values).

You'd just calculate the X/Y deltas of the mouse-drags and feed those distances into the rotation methods. The other thing you could do is keep the object still and rotate the camera around the object instead (in that case look at the Camera class documentation for rotation).

stormp

• byte
• Posts: 38
Re: Scene Rotation
« Reply #2 on: October 28, 2007, 04:05:07 pm »
You'd just calculate the X/Y deltas of the mouse-drags and feed those distances into the rotation methods. The other thing you could do is keep the object still and rotate the camera around the object instead (in that case look at the Camera class documentation for rotation).

Hi, thanks for the reply.  I have a regular camera rotation based on the x/y deltas of the mouse-drags.  Unless I've understood you wrong, what you described wouldn't give me the drag point (where the mouse button was first downed) that Google Earth has.  If you grabbed a sphere close to the top and dragged it a little, the rotation would be much more than if you grabbed a sphere from the middle and dragged it the equal distance across the screen.

hthth

• byte
• Posts: 13
Re: Scene Rotation
« Reply #3 on: October 28, 2007, 04:42:16 pm »
I misunderstood your question. So basically, you already have the mousedragging — but you want acceleration depending on the the mouse position relative to the sphere?

How about this:

• Register where the user initially clicks on screen. You can do that using the mousePressed(MouseEvent e) method.
• Find the center of the screen
• Multiply the rotation by the distance from the initial mousePressed coordinate to the center-of-screen coordinate.

Example: rotationX = (dragDelta.X * (centerDelta(mouseOrigin.X, Center.X)))

Where dragDelta is the distance the user dragged. Center is the center of screen and mouseOrigin is where the user initially pressed at the start of the drag. The centerDelta method would return a sensible variable based on the distance from mouseOrigin.X to Center.X. So the further away from the center the user clicks, the rotation gains more force.

If I'm still misunderstanding you, and the actual question is how you get an acceleration effect — where the rotation slowly decelerates over time — then the answer would be the above and additionally decreasing the value of delta from center multiplier over time.

This'd work okay if the sphere is close to you, but might not work so well if the sphere is small on screen. Does this help? Or am I still misunderstanding your question?

EDIT: This, of course, does not depend on the user actually grabbing the sphere. If that's an issue then you could use Interact2D to check if the user is actually clicking on the sphere or not, and then calculating the distance from that point to the center of the sphere (instead of calculating the distance from the click and to the center of the screen).
« Last Edit: October 28, 2007, 04:50:53 pm by hthth »

stormp

• byte
• Posts: 38
Re: Scene Rotation
« Reply #4 on: October 28, 2007, 05:13:26 pm »
EDIT: This, of course, does not depend on the user actually grabbing the sphere. If that's an issue then you could use Interact2D to check if the user is actually clicking on the sphere or not, and then calculating the distance from that point to the center of the sphere (instead of calculating the distance from the click and to the center of the screen).

The sphere is imaginary.     Basically I have a 3D graph chart as a scene and I wanted to implement "natural rotations" around the chart (scene).  The effect is that if the user grabs a point and drags it in any zig zag direction, they can return to the scene to the original view by simply dragging the mouse back to its original mouse down position -  "zero-hysteresis rotation".  With normal rotations you cant really do this. :-S  Also, the original mouse-down point can not be rotated behind the sphere.

I believe this sort of explains it better:
http://www.mactech.com/articles/mactech/Vol.15/15.03/NaturalObjectRotation/index.html
Quote
The main idea is to think in terms of arc intervals. If we have two arbitrary points A and B on the surface of a unit sphere, the most natural way to get from A to B is to rotate the sphere so that A follows the shortest path (or geodesic) from A to B. Thus the rotation occurs in the plane of the geodesic. If a = (xa, ya, za) and b = (xb,yb,zb) are the position vectors of the points, the axis of rotation is given by a 5 b. The angle of the rotation can be obtained from cos-1 (a × b).

How do we get from 2D mouse coordinates to 3D rotations? We construct a pseudo-3D coordinate space as follows. (This is essentially the method of Shoemake in Graphic Gems IV, p. 176.) Superimpose an imaginary sphere on our 3D object such that the center of the sphere is at the position vector c = (screen_x, screen_y, 0), where screen_x and screen_y are the local screen coordinates of the center of the object. We assume that any mouse-downs will happen on the surface of our imaginary sphere, which has a radius of r pixels.

I'd prefer to rotate the camera rather than the scene ^_^  To be honest my math is rusty to achieve this :-(

EgonOlsen

• Administrator
• quad
• Posts: 12037
Re: Scene Rotation
« Reply #5 on: October 30, 2007, 09:28:15 am »
The tutorial that you've linked to looks quite easy to implement (judging from a quick look). I would go with the vector based approach. If you want to use the quaternion-based, look at the skeletal-animation-framework in the download section. I think it contains some quaternion classes that may be helpful.

stormp

• byte
• Posts: 38
Re: Scene Rotation
« Reply #6 on: October 30, 2007, 11:05:02 am »
I think my main problem is that I haven't fully sorted out in my head the transition from mouse (screen space) to camera space to world space and back.  Are there any examples that show mouse interaction with objects?  That might enlighten me more :-)

In the example I posted, it appears everything is done in world space and the object is moving, not a free moving camera that does arbitrary rotations on mouse drags.
« Last Edit: October 30, 2007, 11:08:03 am by stormp »

EgonOlsen

• Administrator
• quad
• Posts: 12037
Re: Scene Rotation
« Reply #7 on: October 30, 2007, 09:06:48 pm »
There are some examples of mouse/object interaction in the forums IIRC, but the search function still sucks (= it finds threads that are totally unrelated to the search phrase), so they are a bit hard to find. I've found this, not sure if it is helpful: http://www.jpct.net/forum2/index.php/topic,226
Anyway, the transition from screen to camera space is quite easy because Interact2D offers methods for it. Going back from camera to world space requires to apply the invers camera transformation, which is not that difficult either. If you are still stuck, i'll try to write an example that shows how to do it, but my time is a little limited right now..

stormp

• byte
• Posts: 38
Re: Scene Rotation
« Reply #8 on: December 10, 2007, 03:44:17 pm »
Hi Egon,

Thanks for the code!  I tried messaging you before but I think my browser was acting up.. the replies never end up in the outbox. :-S

Your example seems to be similar to my Euler Angle Rotation example here:

http://testcod07.fortunecity.com/rotate/EulerRotateCamTest/

It's not really what I'm trying to accomplish. I'm trying to get the same effect that google earth has.  The use clicks on the screen, the code calculates the closet point on an imaginary sphere as the drag point. As the user drags the cursor around the screen, the camera moves such that the original point grabbed is rotated to where the current mouse position is.  If the user drags outside the sphere area and around it would rotate around the Z axis (i believe).

Does that make sense?   I think the term is called Trackball Rotation? or Virtual Trackball Rotation.

Very grateful for your time.

Storm.

« Last Edit: December 11, 2007, 10:21:20 am by stormp »

EgonOlsen

• Administrator
• quad
• Posts: 12037
Re: Scene Rotation
« Reply #9 on: December 12, 2007, 12:14:48 am »
I see...however, i don't have a solution ATM and not much time to invent one. If i would start to code one, i would try this approach:

• see if the mouse hits the globe
• if so, determine the intersection point using calcMinDistance() or something
• getting the rotation matrix from the resulting vertex (center of the globe->intersection point)
• get the current rotation matrix and interpolate between the two
• set the interpolated matrix as new rotation matrix
• repeat the last two steps until both matrices match (almost)

That's just an idea, it's untested...it may work, it may produce complete nonsense. But it's the best i can come up with at the moment.
If it doesn't help, i'll try to implement that solution myself, but i can't say when...

stormp

• byte
• Posts: 38
Re: Scene Rotation
« Reply #10 on: December 12, 2007, 01:26:06 am »
Thanks Egon,

I also found this site that should solve part of the problem but it implements Quaternion math.

http://viewport3d.com/trackball.htm

Storm.

EgonOlsen

• Administrator
• quad
• Posts: 12037
Re: Scene Rotation
« Reply #11 on: December 12, 2007, 10:50:45 am »
Whatever can be done with quaternions can be done with matrices as the concepts are mathematically equivalent (albeit some tutorials may tell you otherwise). However, it may be difficult to do the transfer, so maybe you should check out the skeletal animation package in the download section which includes some classes for quaternions.

stormp

• byte
• Posts: 38
Re: Scene Rotation
« Reply #12 on: December 12, 2007, 11:40:40 am »
Will do Egon.  Thanks!

stormp

• byte
• Posts: 38
Trackball Rotation
« Reply #13 on: January 23, 2008, 05:24:08 pm »
Here is a great example of what I would like to do:

http://www.vizmo.com/rgizmo.html

EgonOlsen

• Administrator
• quad
• Posts: 12037
Re: Scene Rotation
« Reply #14 on: January 23, 2008, 06:29:21 pm »
Maybe i don't get it, but isn't that basically the same as the code that i had given to you with the difference that the mouse cursor is visible and the sphere's position below the cursor rotates with the cursor? If so, maybe this slightly modified version is a starting point. I don't think that it is mathematically correct and it only works for the given sphere- and window size, but it should be possible to make the d=w/200f-thing depend on those dimensions so that it works for all sizes. Or am i not getting it again?

Code: [Select]
`import com.threed.jpct.*;import com.threed.jpct.util.*;import org.lwjgl.opengl.*;import org.lwjgl.input.*;/** * A Simple demo for mouse-powered object viewing. * It offers two modes: Camera around object (default) and object "around" camera. */public class GoogleCube {    private static float distance=50;    // If true, the object transforms related to the camera. If false, the camera to the object.    private static boolean affectObject=false;    public static void main(String[] args) throws Exception {        World world=new World();        FrameBuffer buffer=new FrameBuffer(640, 480, FrameBuffer.SAMPLINGMODE_HARDWARE_ONLY);        buffer.disableRenderer(IRenderer.RENDERER_SOFTWARE);        buffer.enableRenderer(IRenderer.RENDERER_OPENGL);        Object3D cube=Primitives.getSphere(15);        world.addObject(cube);        cube.build();        Camera cam=world.getCamera();        cam.moveCamera(Camera.CAMERA_MOVEUP, distance);        cam.rotateCameraX((float)Math.PI/2f);        world.setAmbientLight(100, 100, 100);        Light light=new Light(world);        light.setIntensity(0, 0, 255);        light.setPosition(new SimpleVector(-100, 0, 0));        Mouse.create();        while (!Display.isCloseRequested() && !Mouse.isButtonDown(1)) {            buffer.clear();            world.renderScene(buffer);            world.draw(buffer);            buffer.update();            buffer.displayGLOnly();            int x=Mouse.getDX();            int y=Mouse.getDY();            int w=Mouse.getDWheel();            if (Mouse.isButtonDown(0)) {                SimpleVector line=new SimpleVector(x, 0, y);                Matrix m=line.normalize().getRotationMatrix();                if (affectObject) {                    // Object "around" camera                    cube.rotateAxis(m.getXAxis(), line.length() / 200f);                } else {                    // Camera around object                    m.rotateAxis(m.getXAxis(), (float) -Math.PI/2f);                    cam.moveCamera(Camera.CAMERA_MOVEIN, distance);                    cam.rotateAxis(m.invert3x3().getXAxis(), line.length() / 200f);                    cam.moveCamera(Camera.CAMERA_MOVEOUT, distance);                }            }            if (w!=0) {            float d=w/200f;                if (affectObject) {                    cube.translate(0, -d, 0);                } else {                    distance-=d;                    cam.moveCamera(Camera.CAMERA_MOVEIN, d);                }            }            Thread.sleep(10);        }        Mouse.setGrabbed(false);        Mouse.destroy();        buffer.disableRenderer(IRenderer.RENDERER_OPENGL);        buffer.dispose();    }}`