How to hide ITextSceneNode

[aungsithu]

Hi,

I've some objects with some text displayed on top.
I'm using ITextSceneNode and place the text above the object's pos.

What happen is since these objects are moving in the scene, sometimes they are behind some other objects and cannot be seen. But the problem is the text above that object is always visible.

How can I hide my text if the object is not visible in the scene?

Thanks in advance

[JP]

Hmmm you'd have to come up with some system yourself i think... somehow determine whether the object is visible to the player and if it's not set the text node to invisible. Not entirely sure how you'd go about doing that though!

Alternatively you could create a new type of text scene node which doesn't render on top of everything and instead renders only when visible. Not sure if you could achieve that with the current text scene node or not by setting material params to do with the z buffer.

[Acki]

if I remember correctly the default z-buffer flag is false for text scene nodes...
try to enable z-buffer for it:

Code: Select all

txtNode->setMaterialFlag(EMF_ZBUFFER, true);

[B@z]

acki: i dont think that works, i tried that too.
didn't work for me. aungsithu try it too xD

well i think you should send a ray from the camera, to the model, and just set visibility to false if there is an obstacle

[Acki]

acki: i dont think that works, i tried that too.
didn't work for me. aungsithu try it too xD

yes, you're right...
I had the billboard in mind, where I had this problem some time ago, but its default was changed to true meanwhile...

I'm not sure why the text scene node doesn't react on the zbuffer flag though, maybe there is a reason for this ???
or maybe it's simply a bug ???

well i think you should send a ray from the camera, to the model, and just set visibility to false if there is an obstacle

ok, but if you have many text scene nodes then this would cause some overheat, doesn't it ???
well, if it is the only way to do this...

[Eigen]

No need for casting a ray every frame. You could do it like once a second or so.

[B@z]

yeah it would be cool if the zbuffer flag work..

well ray casting... didn't cause problem to me xD yet... xD
but the little problem with that is if you have the collision manager collide your camera, then it won't show the text node if youre near to the wall.
so you have to move the ray a little.

[aungsithu]

@Acki:

Code: Select all

txtNode->setMaterialFlag(EMF_ZBUFFER, true);

This doesn't work for me also.

I want to try that ray casting too.
Can anybody give a basic idea of implementing ray casting in Irrlicht?

Thanks

[Frank Dodd]

I use a couple of lines like: -

Code: Select all

   line3d< f32 > ray(  vectorStart.x, vectorStart.y, vectorStart.z,
                        vectorEnd.x, vectorEnd.y, vectorEnd.z );
    ISceneNode * collisionNode = smgr->getSceneCollisionManager()->getSceneNodeFromRayBB( ray );

in my FreeBasic Irrlicht wrapper. I think Eigens comment is certainly noteworthy too, if you have ten labeled objects in your scene you could test one a frame or something.

[rogerborg]

CTextSceneNode use uses Font->draw() to render the text. This is a purely 2D method, which doesn't take into account z depth.

There's also addBillboardTextSceneNode(), which renders the text onto a billboard. This does use z depth, but the billboard then scales with distance, and doesn't look very nice anyway.

An occlusion check might be good enough: this is the trivial case.

Code: Select all

#include "irrlicht.h"
#include <math.h>

using namespace irr;
using namespace core;
using namespace scene;
using namespace video;

#ifdef _IRR_WINDOWS_
#pragma comment(lib, "Irrlicht.lib")
#endif

int main()
{
	IrrlichtDevice *device = createDevice( EDT_OPENGL,
								 dimension2d<s32>(800, 600),
								 32);
	ISceneManager* smgr  = device->getSceneManager();
	IVideoDriver* driver = device->getVideoDriver();

	ICameraSceneNode * camera = smgr->addCameraSceneNode();

	ISceneNode * cube1 = smgr->addCubeSceneNode();
	cube1->setMaterialTexture(0, driver->getTexture("../../media/wall.bmp"));
	cube1->setMaterialFlag(video::EMF_LIGHTING, false);
	cube1->setPosition(vector3df(0, 0, 40));


	ISceneNode * cube2 = smgr->addCubeSceneNode();
	cube2->setMaterialTexture(0, driver->getTexture("../../media/wall.bmp"));
	cube2->setMaterialFlag(video::EMF_LIGHTING, false);

	ITextSceneNode * text = smgr->addTextSceneNode(smgr->getGUIEnvironment()->getBuiltInFont(), stringw("Test").c_str());
	text->setParent(cube2);
	text->setPosition(vector3df(0, 8, 0));

	const u32 start = device->getTimer()->getTime();
	while (device->run())
	{
		const u32 now = device->getTimer()->getTime();
		const f32 depth = 40.f + 20.f * sin((f32)((now - start) % 2000) / 2000.f * 2.f * PI);
		cube2->setPosition(vector3df(5, 5, depth));
		cube2->updateAbsolutePosition();

		line3df rayToCube(camera->getAbsolutePosition(), cube2->getAbsolutePosition());
		if(cube2 == smgr->getSceneCollisionManager()->getSceneNodeFromRayBB(rayToCube))
			text->setVisible(true);
		else
			text->setVisible(false);

		driver->beginScene(true, true, video::SColor(0,100,100,100));
		smgr->drawAll();
		driver->endScene();
	}

	device->drop();
} 

If you're working inside (e.g.) a level mesh, then have a look in example 07.Collision/main.cpp. This uses getCollisionPoint() to test collision of a ray with a level mesh.

If you do that as well using the same camera-to-object ray (I'd do it after the bounding box check, since it's more expensive) then if you get a hit with the level mesh, you can assume that it's occluding the object of interest.

Both of these tests are very gross, and only test on a single ray to the centre of the object of interest. It's really up to you whether this is good enough, or whether you have to do more ray tests, or find another method of testing for occlusion.

[B@z]

my code:

Code: Select all

			vector3df point;
			triangle3df tri;
			if (!smgr->getSceneCollisionManager()->getCollisionPoint(line3df(smgr->getActiveCamera()->getPosition()+MoveVector(vector3df(0,player->getRotation(),0),20), player[i].getPosition()), cMap.getTriangleSelector(),point,tri) &&
				player->getPosition().getDistanceFrom(players[i].getPosition()) < 650)
			{
                            players[i].textNode->setVisible(false);
			}
                        else players[i].textNode->setVisible(true);

oh MoveRotation has two parameters, first is the rotation, and the second is the length, and gives back a vector.
just a simple function what i wrote xD

[aungsithu]

Thank you everyone.
I'll try it out and will let you know the results.

Cheers

[aungsithu]

OK. I gonna use that occlusion check.
That looks fit for me.
Thank you everybody..