DirtyScreen doesnt work and crashes jupyterlabs

[nkemter]

I want to use DirtyScreen as the description sounds very useful. However the example picture in the documentation doesn't show any effect. And when using it in my Jupyterlabs it doesnt work there either. It also crashes jupyterlabs (it opens a "DirtyScreen" subprocess/window that freezes instantly. After that I need to restart the kernel). Im using python 3.11.8

im basically only using different provided examples of you

dirtyscreen = DirtyScreen(n_clusters = (50,100),
                                                      n_samples = (2,20),
                                                      std_range = (1,5),
                                                      value_range = (150,250),
                                                      p=1.0
                          )

image_augmented = dirtyscreen(image)
cv2.imshow("dirtyscreen", image_augmented)
# create borders
ysize, xsize = image.shape[:2]
image[0,:] = 0
image[ysize-1,:] = 0
image[:, 0] = 0
image[:, xsize-1] = 0

# create borders
ysize, xsize = image_augmented.shape[:2]
image_augmented[0,:] = 0
image_augmented[ysize-1,:] = 0
image_augmented[:, 0] = 0
image_augmented[:, xsize-1] = 0

plt.figure()
plt.subplot(1,2,1)
plt.imshow(image, cmap="gray")
plt.title("input")
plt.axis('off')

plt.subplot(1,2,2)
plt.imshow(image_augmented, cmap="gray")
plt.title("augmented")
plt.axis('off')

fname = "../Data/output/augraphy_punkte.png"
if os.path.isfile(fname):
    os.remove(fname)
plt.imsave(fname, image_augmented, cmap="gray")

[kwcckw]

Hi, the dirty screen effect may not be very distinct, but if you zoom it, you will be able to see those tiny blobs. Here's the zoomed image from the documentation output:

Also i tried your similar code with JupyterLab 4.0.11 and python 3.11, looks like i didn't getting any problem.

Have you try with the other function? Is it crash on all functions or just dirty screen?

[nkemter]

Thanks for the fast answer. After wasting quite a lot of time, I figured out that the crash was completely my fault and not related to dirtyscreen. I misunderstood "cv2.imshow()"

You can close it :)