Update app.py

app.py

@@ -25,10 +25,11 @@ def predict_clothing(images):
                      {"type": "text", "text": """Define only the clothing from this image in 1-3 words, always include the colour of the clothing too.
                                                  Your response should be only the definition.
                                                  Examples:
-                                                 - Black Tshirt
-                                                 - Blue jeans
-                                                 - Olive Cargo pants
-                                                 - WHite shirt and Brown trousers"""}
+                                                 Black Tshirt
+                                                 Blue jeans
+                                                 Olive Cargo pants
+                                                 If there are topwear and bottomwear to be seen in the image, define both. Example: White shirt and Brown trousers
+                                                 """}
                  ]}
                ]
     input_text = processor.apply_chat_template(messages, add_generation_prompt=True)
@@ -63,10 +64,10 @@ def generate_image(category, img1, img2, img3, height, width, img_guidance_scale
     wears = predict_clothing(input_images[1:])
     if len(wears)==1:
         dress = wears[0]
-        text = f"""A {category} wearing {dress}, with a completely white background. The {category} is <img><|image_1|></img>. The {dress} is in <img><|image_2|></img>."""
+        text = f"""A {category} wearing {dress}, with a completely white background. The {category} is in <img><|image_1|></img>. The {dress} is in <img><|image_2|></img>."""
     elif len(wears)==2:
         topwear, bottomwear = wears[0], wears[1]
-        text = f"""A {category} wearing {topwear} and {bottomwear}, with a completely white background. The {category} is <img><|image_1|></img>. The {topwear} is in <img><|image_2|></img>. The {bottomwear} is in <img><|image_3|></img>."""
+        text = f"""A {category} wearing {topwear} and {bottomwear}, with a completely white background. The {category} is in <img><|image_1|></img>. The {topwear} is in <img><|image_2|></img>. The {bottomwear} is in <img><|image_3|></img>."""
     else:
         input_images = None
 
@@ -126,7 +127,7 @@ description = """
 
 Credits = """**Credits**
 
-Made using [OmniGen](https://huggingface.co/Shitao/OmniGen-v1): Unified Image Generation [paper](https://arxiv.org/abs/2409.11340) [code](https://github.com/VectorSpaceLab/OmniGen)
+Made using [Llava 3.2 Vision](https://huggingface.co/meta-llama/Llama-3.2-11B-Vision-Instruct) and [OmniGen](https://huggingface.co/Shitao/OmniGen-v1): Unified Image Generation
 """
 
 
@@ -183,16 +184,14 @@ with gr.Blocks() as demo:
         generate_image,
         inputs=[category, image_input_1, image_input_2, image_input_3, height_input, width_input, img_guidance_scale_input, num_inference_steps,
                 seed_input, separate_cfg_infer, offload_model, use_input_image_size_as_output, max_input_image_size, randomize_seed, guidance_scale_input],
-        outputs=output_image,
-    )
+        outputs=output_image,)
 
     gr.Examples(
         examples=get_example(),
         fn=run_for_examples,
         inputs=[image_input_1, image_input_2, image_input_3, height_input, width_input, img_guidance_scale_input, seed_input,
                 max_input_image_size, randomize_seed, use_input_image_size_as_output, guidance_scale_input],
-        outputs=output_image,
-    )
+        outputs=output_image,)
 
     gr.Markdown(Credits)