Changing Camera Position does not change renderd image

[SimonHawe]

Hi to all,
when I move my camera around, and changing its position, target vector up vector etc. nothing happens, the output remains the same. I placed the camera on different initial positions and get different initial views but nothing changes when i move it around, that's pretty weird. Can you image why it does not work?
Simon

[JP]

How do you render your scene? Do you have the while loop constantly rendering?

Are you sure the render is being updated?

Are you sure the camera is being moved? How do you know the camera's being moved if you don't see a change on screen?

If you code isn't too big you could post it here.

[SimonHawe]

I render my scene to texture and the drawAll is called constantly in the while loop.
I know my camera has been moved from camera->getAbsolutePosition(). What do you mean be render is being updated? Perhaps i miss something at this point. But it is really strange because I have moved around cameras before.

[JP]

Presumably you repeatedly render it back to the texture yes? Not just once..?

If you show us your render code where you render to the texture and how you render that texture onto the screen then we can probably help more

[SimonHawe]

Code: Select all

ITexture * renderTarget = driver->createRenderTargetTexture(dimension2di(16,16));

	while(device->run())
		{
			
			driver->beginScene(false, true, video::SColor(0,200,200,200));
			m_lookVector = vector3df(0, 1, 0);
			for(int outerLoop = 0; outerLoop < m_nbrOfFacets; ++outerLoop)
			{
				for(int innerLoop = 0; innerLoop <= m_nbrOfFacetsV; innerLoop ++)
				{
					if(innerLoop != 0){
						m_lookVector.rotateXZBy(RADTODEG*(sign*m_angle),m_turnCenter);
						m_upVector.rotateXZBy(RADTODEG*(sign*m_angle),m_turnCenter);
						m_currentX	+=	(sign*m_stepping);
						accumulator	+=	m_angle;
					}


					camera->setTarget(camera->getAbsolutePosition() - m_lookVector);
					driver->setRenderTarget(renderTarget, true, true);
					smgr->drawAll();

					driver->setRenderTarget(0, false, false);
					rect<s32> sourceRect(0, 0, m_stepping-1, m_stepping-1);
					rect<s32> destRect(m_currentX-1, m_currentY-1,m_currentX+m_stepping-1, m_currentY+m_stepping-1);
					driver->draw2DImage(renderTarget, destRect, sourceRect);

				}

				m_lookVector.rotateXZBy(RADTODEG*(sign*(core::PI - accumulator)),m_turnCenter);
				m_lookVector.rotateXYBy(RADTODEG*(sign*(m_facetFov)),m_turnCenter);
				m_currentY	   -=	m_stepping;
				accumulator		=	0;
				sign		    =	sign*(-1);
				m_radius		=	sqrt(pow(m_lookVector.X,2)+pow(m_lookVector.Z,2));
				m_angle			=	acos(1-pow(m_sphereStep,2)/(2*pow(m_radius,2)));
				m_nbrOfFacetsV	=	round(core::PI/m_angle);
				if(m_nbrOfFacetsV == 32)
					m_nbrOfFacetsV = 31;
				m_currentX		=	256 - sign*(m_stepping*(m_nbrOfFacetsV+1))/2 - ((sign>0)? 0 : m_stepping);
			}

			if(receiver.IsKeyDown(KEY_LEFT))
				m_rotangle -= 30.f;
			if(receiver.IsKeyDown(KEY_RIGHT)){
				m_rotangle += 30.f;
			}
			
			camera->setPosition(core::vector3df(20+m_rotangle,20,10)); 
			camera->updateAbsolutePosition();
			driver->endScene();
		}

		

I left out not important things not used for rendering.

[rogerborg]

Code: Select all

m_lookVector = vector3df(0, 1, 0);

//...
m_lookVector.rotateXZBy(RADTODEG*(sign*m_angle),m_turnCenter);
m_upVector.rotateXZBy(RADTODEG*(sign*m_angle),m_turnCenter);

At a quick glance, it looks like you're rotating a Y vector around the Y axis. In other words, it's going to remain (0,1,0).

I'd imagine that you want to start with a (0, 0, 1) look vector.

In fact, I'm pretty sure.

[SimonHawe]

yes, but that is only the point once because i will then rotate this vector around z and this turned vector will be turned around y, but you are right it is once turend without being turned But anyway I can not move the camera

[rogerborg]

Ah, right, I didn't see the m_lookVector.rotateXYBy().

What value does m_upVector start at? I don't see it being re-initialised to anything, let alone to a different value than (0,1,0).

Sorry, I'm not seeing much else wrong with this. I'm far more useful with a complete compilable, debuggable example, hint, hint.

[SimonHawe]

solved the problem was just a dumb mistake, i reckon i have been working to long today

[rogerborg]

I need closure on that anecdote!

[Taymo]

I feel rogerBorg's pain, and why not just use Irrlicht's awesome addCameraSceneNodeFPS() function and forget about the camera for now. You're doing the same thing it's doing anyways..

edit* even with the render target stuff you could add an FPS camera and grab it's values after you've moved.

[rogerborg]

Joking aside, you'll get more help from the community if you engage a bit more. A simple acknowledgement of the render to texture implementation that I took some time to provide would have been appreciated. Not demanded, just appreciated.

[SimonHawe]

@rogerBorg, yes sorry for not saying thank you, even I do not really use your code anymore but it showed me that render to texture is much faster and without your help I reckon i would not have mentioned that and I hope you will keep on helping me. Thanks a lot.

@Taymon, what are you talking about? I had a completly stupid error in my code which had nothing to do with fps camera and i do not need an fps camera, it was just an reinitialization problem and as i already said I did not want to waist your time but after coding 15 hours sometimes you start missing things.[/list]