How can I convert the output png Image coordinate to dxf coordinate?

[c-chaitanya]

I have gone over the approach in #219 which shows how to convert dxf co-ordinates to image co-ordinates. I have been converting dxf to png using the matplotlib backend approach.

doc = ezdxf.readfile(path_to_dxf)
layout = doc.modelspace()
fig = plt.figure()
ax = fig.add_axes([0, 0, 1, 1])
ctx = RenderContext(doc)
out = MatplotlibBackend(ax)
CustomFrontend(ctx, out).draw_layout(layout, finalize=True)
fig.savefig(path_to_output_png, dpi=300)
plt.close(fig)

Once I have the image, My requirement is the exact opposite of issue #219 i.e getting dxf co-ordinates given its location on the output png. Is something of this sort possible @mbway @mozman

Any help would be greatly appreciated.

[mbway]

the get_wcs_to_image_transform function produces a matrix which should be invertible, so as a least effort thing you could implement:

def get_image_to_wcs_transform(ax: plt.Axes, image_size: Tuple[int, int]) -> np.ndarray:
    return np.linalg.inv(get_wcs_to_image_transform(ax, image_size))

If you won't have access to the axes (eg if you are reading in the image in a separate script) then you could generate the matrix when you save the image and write the resulting matrix to a file kind of like a world file which is literally just a matrix stored column-major used to georeference images (transform them into some geo coordinate system).
The easiest way to save and load the matrix would be with np.savetxt() and np.loadtxt() respectively.

[c-chaitanya]

Thanks @mbway for the (very) quick reply. Will try it out.