QGraphicsView.fitInView令人费解的行为

Question

我正在尝试制作一个带有两个QGraphicsView的小部件，以并排比较两个图像。当用户放大其中一个视图时，另一个视图应放大到同一区域。我的zoomIn函数发送一个带有可见场景矩形的信号，另一个视图中的一个插槽捕获该信号并调用QGraphicsView.fitInView(rectangle)使相同的图像区域可见。 令我惊讶的是，fitInView似乎将矩形（x，y，宽度，高度）视为（x，y，x+宽度，y+高度）。 这是一个错误还是我遗漏了什么

以下是我的代码示例：

QGraphicsView子类

import logging
from PyQt5 import (
    QtWidgets as qw,
    QtCore as qc,
    QtGui as qg
)

class Display(qw.QGraphicsView):
    sigViewportAreaChanged = qc.pyqtSignal(qc.QRectF)

    def __init__(self, *args, **kwargs):
        super(Display, self).__init__(*args, **kwargs)
        ...
        self.zoomInAction = qw.QAction('Zoom in')
        self.zoomInAction.triggered.connect(self.zoomIn)

    @qc.pyqtSlot()
    def zoomIn(self):
        self.scale(1.2, 1.2)
        pos = self.viewport().pos()
        size = self.viewport().size()
        logging.debug(f'{self.objectName()}: viewport:: pos: {pos}, size: {size}')
        pos = self.mapToScene(pos)
        size = self.mapToScene(size.width(), size.height())
        logging.debug(f'{self.objectName()}: scene:: pos: {pos}, size: {size}')
        view_area = qc.QRectF(pos.x(), pos.y(), size.x(), size.y())
        logging.debug(
            f'Emitting changed viewport area, {view_area},\n{mat}')
        self.sigViewportAreaChanged.emit(view_area)

    def setViewportRect(self, rect: qc.QRectF) -> None:
        logging.debug(f'{self.objectName()} <- {self.sender().objectName()}\nFit view to {rect}')
        pos = self.viewport().pos()
        size = self.viewport().size()
        logging.debug(f'{self.objectName()}: before fitting area: viewport:: pos: {pos}, size: {size}')
        pos = self.mapToScene(pos)
        size = self.mapToScene(size.width(), size.height())
        logging.debug(f'{self.objectName()}: before fitting area: scene:: pos: {pos}, size: {size}')
        # self.fitInView(rect)  # Incorrectly sets the visible area
        self.fitInView(rect.x(), rect.y(), rect.width() - rect.x(), rect.height() - rect.y())  # this works
        pos = self.viewport().pos()
        size = self.viewport().size()
        logging.debug(f'{self.objectName()}: after fitting area: viewport:: pos: {pos}, size: {size}')
        pos = self.mapToScene(pos)
        size = self.mapToScene(size.width(), size.height())
        logging.debug(f'{self.objectName()}: after fitting area: scene:: pos: {pos}, size: {size}')  # this suggests that the new viewport is (x, y, x+width, y+height) instead of (x, y, width, height)

容器小部件：

class ReviewWidget(qw.QWidget):
    """A widget with two panes for comparing images"""

    def __init__(self, *args, **kwargs):
        super(ReviewWidget, self).__init__(*args, **kwargs)
        layout = qw.QVBoxLayout()
        self.before = Display()
        self.before.setObjectName('Left')
        self.after = Display()
        self.after.setObjectName('Right')
        self.before.setHorizontalScrollBarPolicy(qc.Qt.ScrollBarAlwaysOn)
        self.before.setVerticalScrollBarPolicy(qc.Qt.ScrollBarAlwaysOn)
        self.after.setHorizontalScrollBarPolicy(qc.Qt.ScrollBarAlwaysOn)
        self.after.setVerticalScrollBarPolicy(qc.Qt.ScrollBarAlwaysOn)

        panes_layout = qw.QHBoxLayout()
        panes_layout.addWidget(self.before)
        panes_layout.addWidget(self.after)
        layout.addLayout(panes_layout)
        self.setLayout(panes_layout)
        self.make_actions()

    def tieViews(self, tie):
        if tie:
            self.before.sigViewportAreaChanged.connect(self.after.setViewportRect)
            self.after.sigViewportAreaChanged.connect(self.before.setViewportRect)
       else:
            self.before.disconnect(self.before.sigViewportAreaChanged)
            self.after.disconnect(self.after.sigViewportAreaChanged)

    def make_actions(self):
        self.tieViewsAction = qw.QAction('Zoom views together')
        self.tieViewsAction.setCheckable(True)
        self.tieViewsAction.triggered.connect(self.tieViews)

Answer 1

我相信你计算size的方法是错误的。您应该相对于pos的转换值来计算它。因为您知道宽度和高度的值，所以可以基于pos的场景值计算view_area的右下角

viewport_pos = self.viewport().pos()
scene_pos = self.mapToScene(viewport_pos)

width = self.viewport().size().width()
height = self.viewport().size().height()

lower_right_corner = QPointF(width - scene_pos.x(), height() - scene_pos.y())

如果没有减法，则假定view_area的原点在(0, 0)，而实际上它是pos

Answer 2

好吧，在读了@Roney Gomes的答案后，我发现了我的错误

为了将大小缩放到场景坐标系，我将大小作为一个点传递到mapToScene

但点和大小在概念上是不同的，虽然大小不依赖于坐标系的原点，但点却依赖于坐标系的原点。 mapToScene将其参数视为一个点，将其转换以反映新的坐标系

因此，我需要减去新坐标系的原点，它实际上是视口的左上角，以恢复大小

更简单的解决方案是直接映射视口矩形：

rect = self.mapToScene(self.viewport().rect())  
rect = rect.boundingRect() # mapToScene(rect) returns QPolygonF
self.sigViewportAreaChanged.emit(rect)