What I want to do:
In PySide, one could override the paintEvent() method of a QWidget to draw a custom widget. The bounding rect can be customized by overriding the boundingRect() method.
How does one do this in JavaFX? My goal is to create a custom Rectangle object, that draws itself smaller than it's bounding rect.
In context:
I'm creating an MS Paint clone in JavaFX. I'm working on the selection box that you use to select/move pixels around. I want the cursor to change to the appropriate resizing cursor when it is near the outside of the selection box.
However, the bounding rect is the same size as the selection box drawn on the screen, so the cursor only changes when it is on top of the box, but not when near. My solution is to set the bounding rect to larger than the actual selection box is, so the cursor change will occur. Then, override it's paintEvent() equivalent to draw a smaller selection box.
Thanks for your help.
I have a similar use case to you, and asked a similiar question here : Drawing transform independent layout bounds in JavaFX.
The JavaFX API is much higher level though than Java2D or PySide (I am assuming from your snippets, because I actually never heard of it ;) ), it does not allow you to override painting of Nodes, nor can you stop a Node from inheriting its parents transform.
This means that you need a seperate Group parallel to your content where you create the selection box and update it according to your needs (content changes etc.).
Example SceneGraph:
Scene
contentGroup
someShapeFromUser
selectionBoxGroup
selectionBoxOfSomeShapeFromUser
Related
In graphics view m setting scene in that adding objects (by dropping ), i can move these items by mouse, when i moved one object on another object moved object should be transparent. how can i make it?
I don't believe you actually want full transparency since it will make it impossible to visually recognize the transparent object later on. Reduced opacity - yes.
As for your question: each item inside your scene has a bounding rectangle (or other type of bounding area). You can easily get it by calling boundingRect() of your item. The returned QRectF has (just like QRect) has the bool QRect::intersects(const QRect &rectangle) const function, which takes another rectangle and checks if a collision is present.
Whenever you move your mouse while dragging an item you need to iterate either through all or just a subset of all items in your scene (by subset I mean just the items in a specific region to increase the performance) and check for collision. If a collision is detected, you can alter either the item you are dragging or the item underneath it.
Of course to make sure that one item covers another one you also need to check the Z value. The easiest way to do that is if you keep all currently not being dragged items at the same Z level and then, whenever you drag one, increase it's Z level by one so that it is "above" the others.
I am trying to make a simple program in which I have added a qgraphics scene and in this I have added a QGraphicsRectItem. I have implemented mouse press event, paint event, bounding rect. Now I have drawn a point on one side of rectangle because there can be multiple rectangle I can drop on screen so just to differentiate between them of different color. Now I can move my rectangle inside graphics seen and can increase the size of rectangle by moving it's one side at a time. The problem that I am facing is when I trying to draw point on one side of rectangle at the time of moving it, it leaves traces on graphics scene. can I remove the ghost lines?
This happens either because your boundingRect method isn't correct, or because you forgot to call prepareGeometryChange before making changes that affect the boundingRect result. Your boundingRect needs to include space for line widths, for example; that's a common mistake.
I have a QGraphicsRectItem over a scene. I intend to drag and drop this window over the scene. When the rect item reaches the left boundary end I have to show it appearing from the right end. Currently I am using two objects and hiding and showing them by calculating the boundary of scene which involves lot of calculations.
Is there any better way to achieve the same effect using just a single object?
Thank You
You could use a single item that spans the entire scene, and draw the rectangle (or 2 parts of it) in it's paint method.
But then you would lose the optimization of the BSP tree, your item would redraw even if some unrelated area repaints. If this is just 1 item, I guess it would not have much impact.
You would need to implement your own dragging with mousemove event and the like, though this is not much code, you just need to get it right.
I have a full size QGLWidget which paints the application background using QPainter (might change to native openGL commands in the future).
On top of this QGLWidget I use QWidgets (non-GL) for the user interface elements. These are, for example, QLineEdits and QPushButtons. I put them into a custom painted QWidget which uses semi-transparent background painting. The paintEvents of the QLineEdit and QPushButton are overwritten and use semi-transparent backgrounds, too.
The whole UI should look like the following (This is a screenshot where I disabled OpenGL and used QWidget instead of QGLWidget for the background. Note the semi-transparent top bar which also draws a shadow (within its own region)):
When the QLineEdit has the focus, it should have a higher opacity but still not fully opaque:
So now, with OpenGL enabled (The background then is a QGLWidget), the semi-transparent widgets above don't paint on top of the background but on (it seems to be) uninitialized data. The image shining through the top bar is sometimes the whole window itself and sometimes other windows currently being on my desktop.
This currently looks like the following (In this screenshot, the data on which the semi-transparent painting operations are painted on seems to be an image of the widget itself, having an offset.):
When I wrote text into the line edit (here: "This is some text which has been there before!"), removed it and set the focus back to the background widget (so the magnifier icon and the placeholder text appear), the previously painted things still shine through (Note that the visible border should not be visible anymore, but also still shines through):
So the problem is: Instead of being painted on top of the underlying widgets, the semi-transparent widget is painted on top of the old results, initially being something like "uninitialized memory".
Why does this happen? How can I solve the problem?
Things you should know before answering:
The background scene is a composition of tiles which are rendered off-screen. So it can be painted very fast and repainting of the background for every little change of the overlay isn't problematic.
The top bar is a custom QWidget with manual painting and arrangement of the contained two widgets (the button and line edit).
The two widgets overwrite the paintEvent, only draw their own (semi-transparent) background when they have focus and don't use frames already provided by Qt. So the white border in the second screenshot is drawn in my custom paintEvent.
I want the background and the composition of overlay widgets to be separately implementable. The background is an AbstractMapView which has some concrete map view classes. The whole window is an AbstractView (there are multiple concrete views, too), which contains both a concrete map view and the overlay widgets, composed in a way itself decides. Therefore, I don't want the logic of the overlay widgets to be part of the underlying map view. (I hope you understood this, as it is a bit complicated.)
This sounds like an issue where the GL content (i.e. your background aka the QGLWidget) is not in the Qt context. While I'm not a pro on GL painting with Qt, you may want to look at this discussion regarding GL painting and a QLabel for some direction/potential hints.
http://www.qtcentre.org/threads/40335-QLabel-on-top-of-a-QGLWidget-background-issue
In short, we here at the office use OpenGL painting and offscreen rendering of maps and it's very important to make sure Qt is aware of the pixels so your foreground widgets can have the semi-transparency applied to their backgrounds.
The particular product we use also renders the map in tiles and supports providing the GL output in a buffer (i.e. it's call a snapshot and is provided as a bitmap) at which point we use the paintEvent of a regular QWidget to paint the buffer so that the painted pixels are in Qt context.
You can define a Qframe with Qt::SplashScreen flag as the search box and set its opacity. Put your widgets inside it such as the search textbox and positon it where it should appear on the mainwindow. It will also be a good idea to reposition it as the mainwindow is moved or resized overriding its moveEvent.
Hi i'm trying to get a photoshop-like behaviour for my QGraphicsScene
The grid in the background should not resize with the call of scale. And I must be able to save the picture with QPixmap::grabWidget(view) but without the background grid. I can probably do it with removing the background layer just before saving the picture, but i'm not sure if its cleanest way to do it.
Any ideas ?
thx.
Question 1
The grid in the background should not resize with the call of scale.
Use the QGraphicsItem::ItemIgnoresTransformations flag.
The item ignores inherited transformations (i.e., its position is
still anchored to its parent, but the parent or view rotation, zoom or
shear transformations are ignored). This flag is useful for keeping
text label items horizontal and unscaled, so they will still be
readable if the view is transformed. When set, the item's view
geometry and scene geometry will be maintained separately.
In order to set this flag use the setFlag function when creating the grid item.
Question 2
I must be able to save the picture with QPixmap::grabWidget(view) but without the
background grid.
Call the hide function on the grid item before calling the grabWidget. After you have grabbed it you show it again by calling the show function.