Update modeling.py

modeling.py

@@ -8,7 +8,7 @@ import torch
 from .dino_wrapper2 import DinoWrapper
 from .transformer import TriplaneTransformer
 from .synthesizer_part import TriplaneSynthesizer
-from .processor import LRMImageProcessor
+# from .processor import LRMImageProcessor
 
 class CameraEmbedder(nn.Module):
     def __init__(self, raw_dim: int, embed_dim: int):
@@ -69,12 +69,13 @@ class LRMGenerator(PreTrainedModel):
             triplane_dim=config.triplane_dim, samples_per_ray=config.rendering_samples_per_ray,
         )
 
-    def forward(self, image):
+    def forward(self, image, camera):
 
         # we use image processor directly in the forward pass
-        processed_image, source_camera = self.image_processor(image)
-        image = processed_image
-        camera = source_camera
+        #TODO: we should have the following:
+        # processor = AutoProcessor.from_pretrained("jadechoghari/vfusion3d")
+        # processed_image, source_camera = processor(image)
+        # 
         
         assert image.shape[0] == camera.shape[0], "Batch size mismatch"
         N = image.shape[0]