Test stand astig removal creates some poor images.

[githubdoe]

I see the first image is squashed left to right creating a non round contour plot. I asked AI to find the cause. It is helping. I will no more later.

[githubdoe]

If AI is correct the problem is with the new make contour square code . Here is what it said:

The culprit is the re-entrancy guard and aspect ratio logic inside ContourPlot.

The Problem
In your ContourPlot::setSurface and ContourPlot::updateAspectRatio methods:

When you call plot->resize(Width, Width), it triggers a QEvent::Resize.

Your eventFilter catches this and calls updateAspectRatio().

updateAspectRatio() uses canvas()->width() and canvas()->height() to calculate the new scales.

Crucially: In Qt, canvas()->width() does not update instantly. When you resize the parent plot, the internal canvas size often lags by one event loop cycle.

On the first image, the canvas still thinks it has its old default "rectangular" dimensions (e.g., 640x480). It calculates the aspect ratio based on those old dimensions, resulting in a "squashed" scale.

On the second image, the canvas has finally caught up to being square from the previous iteration, so the math works out.

[githubdoe]

If AI is correct here is it's fix. I will let Julian decide if AI is correct. I'm suspect that it is not correct. But....

Additional Improvement in ContourPlot.cpp
To make your class more robust for future use, you should modify updateAspectRatio to check if the canvas is actually valid before trusting its dimensions:

C++

void ContourPlot::updateAspectRatio()
{
// ... existing guards ...

QWidget *c = canvas();
// If the plot is not yet visible or just resized, width() might be stale
int canvas_w = c->width();
int canvas_h = c->height();

// If we are in a "minimal" or "batch" mode, we should perhaps 
// just default to a 1:1 ratio if the canvas hasn't initialized.
if (canvas_w <= 1 || canvas_h <= 1) { 
    setAxisScale(QwtPlot::xBottom, 0, m_wf->data.cols);
    setAxisScale(QwtPlot::yLeft, 0, m_wf->data.rows);
    isReentering = false;
    return;
}

// ... rest of existing logic ...

}

[atsju]

I will try it out when I have some time. Most of the discussions I had with AI on this topic where useful to better understand QWT and propose some related functions to read more documentation. However it has never been accurate on actual fixes so far.