up

Author: sayakpaul

src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit.py

@@ -308,7 +308,6 @@ def encode_prompt(
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
         prompt_embeds_mask = prompt_embeds_mask.repeat(1, num_images_per_prompt, 1)
         prompt_embeds_mask = prompt_embeds_mask.view(batch_size * num_images_per_prompt, seq_len)
-
         return prompt_embeds, prompt_embeds_mask
 
     def check_inputs(

src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit_plus.py

@@ -309,6 +309,7 @@ def encode_prompt(
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
         """
+        print(f"{image[0].size=}")
         device = device or self._execution_device
 
         prompt = [prompt] if isinstance(prompt, str) else prompt
@@ -322,7 +323,7 @@ def encode_prompt(
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
         prompt_embeds_mask = prompt_embeds_mask.repeat(1, num_images_per_prompt, 1)
         prompt_embeds_mask = prompt_embeds_mask.view(batch_size * num_images_per_prompt, seq_len)
-
+        print(f"{prompt_embeds.shape=}, {prompt_embeds_mask.shape=}")
         return prompt_embeds, prompt_embeds_mask
 
     # Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage_edit.QwenImageEditPipeline.check_inputs

tests/pipelines/qwenimage/test_qwenimage_edit.py

@@ -133,15 +133,17 @@ def get_dummy_inputs(self, device, seed=0):
         else:
             generator = torch.Generator(device=device).manual_seed(seed)
 
+        # Even if we specify smaller dimensions for the images, it won't work because of how
+        # the internal implementation enforces a minimal resolution of 1024x1024.
         inputs = {
             "prompt": "dance monkey",
-            "image": Image.new("RGB", (32, 32)),
+            "image": Image.new("RGB", (1024, 1024)),
             "negative_prompt": "bad quality",
             "generator": generator,
             "num_inference_steps": 2,
             "true_cfg_scale": 1.0,
-            "height": 32,
-            "width": 32,
+            "height": 1024,
+            "width": 1024,
             "max_sequence_length": 16,
             "output_type": "pt",
         }
@@ -240,5 +242,8 @@ def test_vae_tiling(self, expected_diff_max: float = 0.2):
     def test_encode_prompt_works_in_isolation(
         self, extra_required_param_value_dict=None, keep_params=None, atol=1e-4, rtol=1e-4
     ):
-        keep_params = ["image"]
+        # We include `image` because it's needed in both `encode_prompt` and some other subsequent calculations.
+        # `max_sequence_length` to maintain parity between its value during all invokations of `encode_prompt`
+        # in the following test.
+        keep_params = ["image", "max_sequence_length"]
         super().test_encode_prompt_works_in_isolation(extra_required_param_value_dict, keep_params, atol, rtol)

tests/pipelines/qwenimage/test_qwenimage_edit_plus.py

@@ -134,16 +134,18 @@ def get_dummy_inputs(self, device, seed=0):
         else:
             generator = torch.Generator(device=device).manual_seed(seed)
 
-        image = Image.new("RGB", (32, 32))
+        # Even if we specify smaller dimensions for the images, it won't work because of how
+        # the internal implementation enforces a minimal resolution of 384*384.
+        image = Image.new("RGB", (384, 384))
         inputs = {
             "prompt": "dance monkey",
             "image": [image, image],
             "negative_prompt": "bad quality",
             "generator": generator,
             "num_inference_steps": 2,
             "true_cfg_scale": 1.0,
-            "height": 32,
-            "width": 32,
+            "height": 384,
+            "width": 384,
             "max_sequence_length": 16,
             "output_type": "pt",
         }
@@ -239,7 +241,10 @@ def test_vae_tiling(self, expected_diff_max: float = 0.2):
     def test_encode_prompt_works_in_isolation(
         self, extra_required_param_value_dict=None, keep_params=None, atol=1e-4, rtol=1e-4
     ):
-        keep_params = ["image"]
+        # We include `image` because it's needed in both `encode_prompt` and some other subsequent calculations.
+        # `max_sequence_length` to maintain parity between its value during all invokations of `encode_prompt`
+        # in the following test.
+        keep_params = ["image", "max_sequence_length"]
         super().test_encode_prompt_works_in_isolation(extra_required_param_value_dict, keep_params, atol, rtol)
 
     @pytest.mark.xfail(condition=True, reason="Batch of multiple images needs to be revisited", strict=True)