How to Crop single image before calling torchvision.utils.save_image, If I am using PIL lib Image.crop(....) method then image quality degrade.

 vutils.save_image(fixed_fake.data,outputpath , normalize=True)
print("output path",outputpath)
img = Image.open(outputpath)
noOfRow = 5
noOfColumn = 8
x1 = 2
y1 = 2
x2 = 130
y2 = 130
folder = file_batch

for i in range(0, noOfColumn):
    dest_dir = file_batch[i].split("/")[7]
    if not os.path.exists(outf+"/"+dest_dir):
        os.mkdir(outf+"/"+dest_dir)
    for j in range(1, noOfRow + 1):
        area = (x1, y1, x2, y2)
        cropped_img = img.crop(area)
        imgName = "{}{}".format(i, j)
        cropped_img.save(os.path.join(outf+dest_dir,filename))
        y1 = y1 + 130
        y2 = y2 + 130
    x1 = x1 + 130
    x2 = x2 + 130
   y1 = 2
    y2 = 130

Start by reproducing the provided workflow with torchvision.utils.save_image, PIL Image.open, Image.crop, and cropped_img.save. Compare the saved image before and after cropping to determine where quality changes; done means the source of the degradation is identified and the expected image quality or handling is documented.