0

I am working on updating an application for a client.

They use Qt and currently use a QGLWidget to display a full-screen view of 1 of 4 possible cameras selected by clicking the appropriate radio button. They then use OpenGL to draw on the image being displayed. This works great, but they want to update the UI to include a quad-split view of all 4 cameras.

My first thought on how to accomplish this was to keep the one QGLWidget for the full-screen display, and have 4 small QGLWidgets for the quad-split. From the documentation I found that you can't overlap QGLWidgets or QOpenGLWidgets because they don't handle z-order appropriately, but that this can be accomplished by using QOpenGLWindows and QWidget::createWindowContainer.

So, I coded up an application that uses a QOpenGLWidget (trying to bring them up to date) for the full-screen view, and 4 smaller QOpenGLWindows using QWidget::createWindowContainer, but this isn't working either.

The widgets built from QOpenGLWindows are always on top even if I use lower() to try to get them behind the full screen QOpenGLWidget. I've also tried using hide() on the widgets built from QOpenGLWindows, however, this has had no effect.

genpfault
  • 51,148
  • 11
  • 85
  • 139
muench
  • 3
  • 1

1 Answers1

0

Do this at a lower level. Keep the one QGLWidget -- in fact don't touch your Qt objects. Instead, change the lower-level rendering so that it makes 4 calls to glViewport.

After each call to glViewport, update the modelview and projection and matrices according to the camera of interest, then draw the 3D scene.

This is simple and performant, because the driver only needs to deal with a single OpenGL context. You might have some extra work to adjust mouse input, but I think it'll be worthwhile.

prideout
  • 2,895
  • 1
  • 23
  • 25