Difference between two method for save tensor to image

1

I try to save the output of model with two method.

First one is using ‘from torchvision.utils import save_image’.
Second is using ‘torchvision.transforms.ToPILImage()’.

But the results of two method was different.

Is there any big difference between two method?

2

How large are the differences?
I just tested both methods and they look at least visually the same.
Based on the code from save_image and to_pil_image I would also expect to get the same result (I’m not sure, if the addition of 0.5 will be visible or not).