Update app.py

app.py

@@ -139,8 +139,8 @@ def generate_image(prompt, negative_prompt, seed, randomize_seed, width, height,
 
 # Gradio Interface
 @spaces.GPU
-def process_workflow(image, text_prompt, use_vlm, use_enhancer, model_choice, vlm_model_choice, negative_prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps):
-    if use_vlm and image is not None:
+def process_workflow(image, text_prompt, vlm_model_choice, use_enhancer, model_choice, negative_prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps):
+    if image is not None:
         if vlm_model_choice == "Long Captioner":
             prompt = create_captions_rich(image)
         else:  # Florence
@@ -182,23 +182,19 @@ title = """<h1 align="center">VLM Captioner + Prompt Enhancer + Kolors Image Gen
 </center></p>
 """
 
-# Gradio Interface
 with gr.Blocks(css=custom_css, theme=gr.themes.Soft(primary_hue="blue", secondary_hue="gray")) as demo:
     gr.HTML(title)
     
     with gr.Row():
         with gr.Column(scale=1):
             with gr.Group(elem_classes="input-group"):
-                input_image = gr.Image(label="Input Image for VLM")
-                use_vlm = gr.Checkbox(label="Use VLM Captioner", value=False)
+                input_image = gr.Image(label="Input Image (VLM Captioner)")
                 vlm_model_choice = gr.Radio(["Long Captioner", "Florence"], label="VLM Model", value="Long Captioner")
             
-            with gr.Group(elem_classes="input-group"):
-                text_prompt = gr.Textbox(label="Text Prompt")
+            with gr.Accordion("Advanced Settings", open=False):
+                text_prompt = gr.Textbox(label="Text Prompt (optional, used if no image is uploaded)")
                 use_enhancer = gr.Checkbox(label="Use Prompt Enhancer", value=False)
                 model_choice = gr.Radio(["Medium", "Long"], label="Enhancer Model", value="Long")
-            
-            with gr.Accordion("Advanced Settings", open=False):
                 negative_prompt = gr.Textbox(label="Negative Prompt")
                 seed = gr.Slider(label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=0)
                 randomize_seed = gr.Checkbox(label="Randomize Seed", value=True)
@@ -218,7 +214,7 @@ with gr.Blocks(css=custom_css, theme=gr.themes.Soft(primary_hue="blue", secondar
     generate_btn.click(
         fn=process_workflow,
         inputs=[
-            input_image, text_prompt, use_vlm, use_enhancer, model_choice, vlm_model_choice,
+            input_image, text_prompt, vlm_model_choice, use_enhancer, model_choice,
             negative_prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps
         ],
         outputs=[output_image, final_prompt, used_seed]