Jump to content
  • Advertisement
Sign in to follow this  
simonj2

OpenGL The zooming in blues

This topic is 2999 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

Hi there,

I'm building a 2d application (of a graph if anyone's interested!) using OpenGl in C++. In it, users can drag and also zoom. There is a main function that is called everytime there is any action, the relevant parts are:


//Only scale Y with timeline (not X):
glScaled(1, currentScale, 1);

//Draw timeline (NB, this will always stay in the same place):
//cout << "scale is: " << currentScale << endl;
glTranslated(-totalMovedX, 0, 0);
graph->timeline->draw();
glTranslated(totalMovedX, 0, 0);

//Now, scale the rest:
glScaled(currentScale, 1, 1);
graph->draw();

..(not important)...

glScaled(1/currentScale, 1/currentScale, 1);
//Now, move it a bit if dragging:
if (parent->leftButtonDown){
//cout << "dragging by " << -(mousex-curx) << " " << -(mousey-cury) << endl;
glTranslated(-(mousex-curx), -(mousey-cury), 0);
totalMovedX += (-(mousex-curx));
totalMovedY += (-(mousey-cury));


}



Now, what this does is draws a timeline on the left (along the Y axis), only scales that (if there's a zoom, currentScale starts off as 1), and allows the user to drag the canvas to view different parts of the graph.

What I've been stuck on for an embarrassing number of hours is the zoom. I've got the scale as you've seen above, and I have a zoomIn function:


void RefMapGLScene::zoomIn(bool toMouse){
if (currentScale > 3){
return;
}
pthread_mutex_lock(&parent->gtkLock);
currentScale = currentScale+zoomFactor;

//now zoom in the centre:
double xAdjust = (get_width())*zoomFactor/2;
double yAdjust = get_height()*zoomFactor/2;
glTranslated(-xAdjust, -yAdjust, 0);
totalMovedX += -xAdjust;
totalMovedY += -yAdjust;


Zoomfactor is 0.1, there is an accompanying zoomOut function, but it's just the same, except we subtract zoomFactor.

Now, the problem comes with zooming into the center. Normally scale will appear to zoom towards 0,0 (ie, the bottom left), so I need to do a translate to move the zoom towards the centre. This translate works fine so long as the user does not drag inside the application to see a different part of the graph. If the user does drag to see another part of the graph, it doesn't zoom into the area that the user is looking, but rather, if they dragged to see what's on the right (ie, dragged mouse left), then the app zooms into the left, and vice versa for if they dragged to see what's on the left, the zoom zooms into the right.

To represent this problem visually, imagine you start looking in the centre (where 0 is the origin):
----0----
Then zooming in and out works fine.

But if you look to the right
-0-------
Then zooming in will make everything bigger, but the translate will not leave you zooming into the correct area infront of you, ie, you will head towards the origin:
--0------
---0-----

I hope that kind of makes sense to somebody... :)

Thank you for reading all that. If it doesn't leave you a little bit confused, then you're a bright person! I'd greatly appreciate any help. I've tried using glOrtho to do the zoom in and that's just not working. I just don't see why this translation isn't doing it right... I've played around with it lots, and it still doesn't work... Thank you for any replies in advance!

Share this post


Link to post
Share on other sites
Advertisement
Here is some code from a C# application which I think does what you need:

ret.Translate(Display.ClientSize.Width / 2, Display.ClientSize.Height / 2);
ret.Scale(m_zoom, m_zoom);
ret.Translate(-m_page_bounds.Width / 2 - Hscroll.Value, -m_page_bounds.Height / 2 - Vscroll.Value);

The first line places the origin in the center of the drawing area, the second zooms, and the third moves to the center of the page bounds (your graph maybe)
HScroll and VScroll are scrollbars.

In my application, this is a function which returns a matrix which is needed to properly retrieve the right mouse coordinates (when you zoom out your drawing, you will zoom in your mouse coordinates) in which case you need the inverse o that matrix to retrieve the correct mouse coordinates.

Share this post


Link to post
Share on other sites
Hey,

Thanks very much for that reply.

I've reworked my algorithm a little, to try to make it simpler and closer to your one, but it's still not working unfortunantly.

With every zoom:


double xAdjust = (get_width())*zoomFactor/2;
double yAdjust = get_height()*zoomFactor/2;

zoomTranslateX -= xAdjust;
zoomTranslateY -= yAdjust;


And then in the draw loop:


//So far, we have no translates
glTranslated(0, -zoomTranslateY, 0);

//Only scale Y with timeline (not X):
glScaled(1, currentScale, 1);
glTranslated(0, totalMovedY/currentScale, 0);
//Draw timeline (NB, this will always stay in the same place):
graph->timeline->draw();
glTranslated(0, -totalMovedY/currentScale, 0);

glScaled(1, 1/currentScale, 1);

//Now we are back without any translates or scales
//So move to where you think you need to be:
glTranslated(-zoomTranslateX, 0, 0);


glScaled(currentScale, currentScale, 1);
glTranslated(totalMovedX/currentScale, totalMovedY/currentScale, 0);

do other drawing here...


//now, end the drawing:
glTranslated(-totalMovedX/currentScale, -totalMovedY/currentScale, 0);
glScaled(1/currentScale, 1/currentScale, 1);

//Now we're back without any translates or scales:
glTranslated(zoomTranslateX, zoomTranslateY, 0);



And I'm having exactly the same problem as before I'm afraid! The code is perhaps a bit simpler now, but the same thing's happening, ie, the zoom in always go towards a centre point where we start, whether or not I've moved to the side.. So, just to restate, if I've dragged my canvas so I'm seeing the area to the right, the zoom in zooms to the left (and vice versa with the other side)!

Thanks again for your reply though, it made it simpler in my head...

Share this post


Link to post
Share on other sites
Sorry to double post, I've been looking at: http://www.opengl.org/discussion_boards/ubbthreads.php?ubb=showflat&Number=108344

I shouldn't have to call GlOrtho again after I change the scaling factor, or after I translate, should I? I've tried it, but didn't have any luck... but perhaps I was missing something if I was meant to call that :P

Share this post


Link to post
Share on other sites
I would have to look through your code carefully to be sure (and even then there might not be enough context to be certain), but I have a feeling you might be going about this the wrong way.

First, minor detail, but this:
glTranslated(-totalMovedX, 0, 0);
graph->timeline->draw();
glTranslated(totalMovedX, 0, 0);
Should probably be implemented using glPushMatrix() and glPopMatrix() (since you appear to be using the fixed-function matrix stack).

Let me ask this: are you resetting the modelview matrix to identity at the beginning of each frame, or are you just letting transforms accumulate from frame to frame, and using (implicit) inverse transforms in order to 'cancel out' previous transforms?

If the latter, I don't think you want to be doing it that way (unless there's some good reason for doing it that way that I'm not aware of). It's more typical to track whatever information you need (e.g. zoom factor, camera position, etc.) in your own code, and then build transforms for rendering 'from scratch' (that is, starting with identity) each frame.

Also, I'd think it would be more sensible to implement zoom in terms of the parameters of the orthographic projection rather than via scaling. You can zoom in and out by adjusting the projection, and move the camera by adjusting the camera position (which is then used to set up the 'view' part of the modelview matrix, typically by inversion/negation).

Maybe I'm wrong and the way you're doing it is fine, but I get the sense from glancing over your code that something may be amiss.

Share this post


Link to post
Share on other sites
Hey,

Thanks for that jyk. I'll move it all onto the matrix stack using push/pop, and will call Glortho every time I do the draw loop. But that will have to wait until I'm home :) Will post how it goes, thanks again.

Simon

Share this post


Link to post
Share on other sites
Thanks Jyk, that really helped clean up the code and make everything straight in my mind. It turned out that by doing this, I was forced to rethink the code a little and came across my bug.

I'll post it here incase anyone else has my issue

When zooming I needed to do this:
double xAdjust = (((get_width())*zoomFactor/2)+(-totalMovedX*zoomFactor));

(ie, I needed to account for the area that I had already moved by dragging the mouse, and add the difference to that due to the scale - that's the area following the +, this is what I didn't have before).

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

We are the game development community.

Whether you are an indie, hobbyist, AAA developer, or just trying to learn, GameDev.net is the place for you to learn, share, and connect with the games industry. Learn more About Us or sign up!

Sign me up!