[Bug Report] TypeError in SuperTML: Image.fromarray requires uint8

[wntrsnw130]

Hi,

I encountered a TypeError when using SuperTML with feature_importance=True.

Description:
I think the error occurs because np.zeros creates an array of floats (float64) by default, but PIL.Image.fromarray(..., 'RGB') expects an integer type (usually uint8). This causes a crash during the initialization of the image canvas.

Error Trace:

TypeError: Cannot handle this data type: (1, 1, 3), <f8
File ".../TINTOlib/supertml.py", line 137, in __event2img
    img = Image.fromarray(np.zeros([self.image_pixels, self.image_pixels, 3]), 'RGB')

Proposed Fix:
In TINTOlib/supertml.py, line 137, explicitly specify the dtype as np.uint8:

# Old (Error)
img = Image.fromarray(np.zeros([self.image_pixels, self.image_pixels, 3]), 'RGB')

# New (Fix)
img = Image.fromarray(np.zeros([self.image_pixels, self.image_pixels, 3], dtype=np.uint8), 'RGB')

I manually applied this fix in my local environment and it solved the issue. Thanks!

I am an undergraduate student currently using this library for my thesis research on tabular-to-image algorithms. Thank you for your great work on this tool!