make pil and then convert

app.py

@@ -6,6 +6,7 @@ import torchvision
 import torchvision.transforms as T
 #from torchvision.transforms import v2 as T2
 import cv2
+import PIL
 from PIL import Image
 import numpy as np
 
@@ -13,9 +14,9 @@ output_res = (768,768)
 
 conditioning_image_transforms = T.Compose(
     [
-        #T2.ScaleJitter(target_size=output_res, scale_range=(0.5, 3.0)),
+        #T2.ScaleJitter(target_size=output_res, scale_range=(0.5, 3.0))),
         T.RandomCrop(size=output_res, pad_if_needed=True, padding_mode="symmetric"),
-        T.ToTensor(),
+        T.PILToTensor(),
         T.Normalize([0.5], [0.5]),
     ]
 )
@@ -35,8 +36,9 @@ generator = torch.manual_seed(0)
 # inference function takes prompt, negative prompt and image
 def infer(prompt, negative_prompt, image):
     # implement your inference function here
+    inp = PIL.Image.create(image)
 
-    cond_input = conditioning_image_transforms(torch.from_numpy(image))
+    cond_input = conditioning_image_transforms(image)
     #cond_input = T.ToPILImage(cond_input)
     
     output = pipe(