How to update pyqtgraph image data when zooming, translating, resizing

Question

I'm using PySide. Inside a QMainWindow, I have an ImageItem displayed inside a PlotWidget.

Suppose I have a function generateImage(r, np) that can dynamically generate an image. The function returns a np x np array of pixels, for the image located inside the (-r, -r)-(r, r) square (in real coordinates). I want the axis to display the real coordinates, not the pixel coordinates.

1. If I zoom in / zoom out, I'd like to generate the same number of pixels, but within a smaller / larger square (in real coordinates);
2. If I translate the image, I need to generate an image with a larger square and with a larger number of pixels, to display the region of interrest using the same resolution;
3. If I resize the window, I need to generate a different number of pixels, to keep the resolution of the displayed image the same.

The code I have actually looks like this:

graph = PlotWidget()
image = ImageItem()
graph.addItem(image)
graph.setAspectLocked(True)  # to keep the pixels square

r = 10e-6  # 10 µm
np = 101  # but should be related to number of pixels in the window
image.setImage(generateImage(r, np))
image.setRect(QRectF(-r, -r, 2*r, 2*r))

My guess is to use the sigRangeChanged signal to detect a change in the view. However, I'm not sure how to detect the size and the resolution of the window, and how to correctly update the image.

I naively tried to do something like this:

graph.sigRangeChanged.connect(updateRange)

def updateRange(view, rng):
    r = max(abs(rng[0][0]), abs(rng[0][1]),
            abs(rng[1][0]), abs(rng[1][1]))
    image.setImage(generateImage(r, np))
    image.setRect(QRectF(-r, -r, 2*r, 2*r))

However, it does not work because the range returned by the signal is slightly larger than r itself. Therefore, I see the image becoming smaller and smaller...

• How can I determine the number of pixels I need to fill the window?
• How can I determine the size of the image I need to generate (in real coordinates)
• How can I update what is displayed without recursively updating the image range?
• How can I display only a part of the generated image? Do I need to slice the array before giving it to setImage(), or can I give the entire array and let pyqtgraph display only the region I need?
• More generally, what is the relation between the coordinates of the different elements and transformations (graph.viewRect(), image.viewRect(), image.boundingRect(), image.sceneBoundingRect(), ...)