init

app.py

@@ -200,6 +200,7 @@ def generate_image(prompt_scene,
                    roll=0.1,
                    pitch=0.1,
                    fov=1.0,
+                   thinking_gen=False,
                    progress=gr.Progress(track_tqdm=True)):
     # Clear CUDA cache and avoid tracking gradients
     torch.cuda.empty_cache()
@@ -354,7 +355,7 @@ with gr.Blocks(css=custom_css) as demo:
                 fov = gr.Slider(minimum=0.3491, maximum=1.8326, value=1.5000, step=0.1000, label="fov value")
         with gr.Accordion("Settings", open=True):
             with gr.Row():
-                thinking_und = gr.Checkbox(label="Thinking", value=False)
+                thinking_gen = gr.Checkbox(label="Thinking", value=False)
                 seed_input = gr.Number(label="Seed (Optional)", precision=0, value=42)
             
         generation_button = gr.Button("Generate Images")
@@ -377,15 +378,17 @@ with gr.Blocks(css=custom_css) as demo:
     with gr.Tab("Camera Understanding"):
         gr.Markdown(value="## Camera Understanding")
         image_input = gr.Image()
+        
+        with gr.Accordion("Settings", open=True):
+            with gr.Row():
+                thinking_und = gr.Checkbox(label="Thinking", value=False)
+                und_seed_input = gr.Number(label="Seed (Optional)", precision=0, value=42)
 
         understanding_button = gr.Button("Chat")
         understanding_output = gr.Textbox(label="Response")
         
         camera_map = gr.Image(label="Camera maps (up vector and latitude)")
 
-        with gr.Accordion("Advanced options", open=False):
-            und_seed_input = gr.Number(label="Seed", precision=0, value=42)
-
         examples_inpainting = gr.Examples(
             label="Examples",
             examples=[
@@ -401,13 +404,13 @@ with gr.Blocks(css=custom_css) as demo:
 
     generation_button.click(
         fn=generate_image,
-        inputs=[prompt_input, seed_input, roll, pitch, fov, thinking_und],
+        inputs=[prompt_input, seed_input, roll, pitch, fov, thinking_gen],
         outputs=image_output
     )
 
     understanding_button.click(
         camera_understanding,
-        inputs=[image_input, und_seed_input],
+        inputs=[image_input, thinking_und, und_seed_input],
         outputs=[understanding_output, camera_map]
     )