Update app.py

app.py

@@ -17,7 +17,7 @@ def randomize_seed_fn(seed: int, randomize_seed: bool) -> int:
         seed = random.randint(0, MAX_SEED)
     return seed
 
-def get_instantID(portrait_in, condition_pose, prompt, style):
+def get_instantID(portrait_in, condition_pose, controlnet, prompt, style):
     
     negative_prompt = "(lowres, low quality, worst quality:1.2), (text:1.2), watermark, (frame:1.2), deformed, ugly, deformed eyes, blur, out of focus, blurry, deformed cat, deformed, photo, anthropomorphic cat, monochrome, pet collar, gun, weapon, blue, 3d, drones, drone, buildings in background, green"
 
@@ -37,7 +37,7 @@ def get_instantID(portrait_in, condition_pose, prompt, style):
 		0.4,	# float (numeric value between 0 and 1.5) in 'Pose strength' Slider component
 		0.4,	# float (numeric value between 0 and 1.5) in 'Canny strength' Slider component
 		0.4,	# float (numeric value between 0 and 1.5) in 'Depth strength' Slider component
-		["pose"],	# List[Literal['pose', 'canny', 'depth']]  in 'Controlnet' Checkboxgroup component
+		controlnet,	# List[Literal['pose', 'canny', 'depth']]  in 'Controlnet' Checkboxgroup component
 		1.5,	# float (numeric value between 0.1 and 20.0) in 'Guidance scale' Slider component
 		seed,	# float (numeric value between 0 and 2147483647) in 'Seed' Slider component
 		"EulerDiscreteScheduler",	# Literal['DEISMultistepScheduler', 'HeunDiscreteScheduler', 'EulerDiscreteScheduler', 'DPMSolverMultistepScheduler', 'DPMSolverMultistepScheduler-Karras', 'DPMSolverMultistepScheduler-Karras-SDE']  in 'Schedulers' Dropdown component
@@ -105,7 +105,7 @@ def get_short_caption(image_in):
     print(result)
     return result
 
-def infer(image_in, camera_shot, conditional_pose, prompt, style, chosen_model):
+def infer(image_in, camera_shot, conditional_pose, controlnet_selection, prompt, style, chosen_model):
     
     if camera_shot == "custom":
         if conditional_pose != None:
@@ -114,7 +114,7 @@ def infer(image_in, camera_shot, conditional_pose, prompt, style, chosen_model):
             raise gr.Error("No custom conditional shot found !")
     
     
-    iid_img = get_instantID(image_in, conditional_pose, prompt, style)
+    iid_img = get_instantID(image_in, conditional_pose, controlnet_selection, prompt, style)
 
     #short_cap = get_short_caption(iid_img)
     
@@ -150,18 +150,23 @@ with gr.Blocks(css=css) as demo:
             with gr.Column():
                 face_in = gr.Image(type="filepath", label="Face to copy", value="monalisa.png")
             with gr.Column():
-                with gr.Row():
-                    camera_shot = gr.Dropdown(
-                        label = "Camera Shot", 
-                        info = "Use standard camera shots vocabulary, or drop your custom shot as conditional pose (1280*720 ratio is recommended)",
-                        choices = [
-                            "custom", "close-up", "medium close-up", "medium shot", "cowboy shot", "medium full shot", "full shot"
-                        ],
-                        value = "custom"
+                with gr.Group():
+                    with gr.Row():
+                        camera_shot = gr.Dropdown(
+                            label = "Camera Shot", 
+                            info = "Use standard camera shots vocabulary, or drop your custom shot as conditional pose (1280*720 ratio is recommended)",
+                            choices = [
+                                "custom", "close-up", "medium close-up", "medium shot", "cowboy shot", "medium full shot", "full shot"
+                            ],
+                            value = "custom"
+                        )
+                        style = gr.Dropdown(label="Style template", info="InstantID legacy templates", choices=STYLE_NAMES, value=DEFAULT_STYLE_NAME)
+    
+                    condition_shot = gr.Image(type="filepath", label="Custom conditional shot (Important) [1280*720 recommended]")
+                    controlnet_selection = gr.CheckboxGroup(
+                        ["pose", "canny", "depth"], label="Controlnet", value=["pose"],
+                        info="Use pose for skeleton inference, canny for edge detection, and depth for depth map estimation. You can try all three to control the generation process"
                     )
-                    style = gr.Dropdown(label="Style template", info="InstantID legacy templates", choices=STYLE_NAMES, value=DEFAULT_STYLE_NAME)
-
-                condition_shot = gr.Image(type="filepath", label="Custom conditional shot (Important) [1280*720 recommended]")
                 prompt = gr.Textbox(label="Short Prompt (keeping it short is better)")
                 chosen_model = gr.Radio(label="Choose a model", choices=["i2vgen-xl", "stable-video"], value="i2vgen-xl", interactive=False, visible=False)
             
@@ -188,6 +193,7 @@ with gr.Blocks(css=css) as demo:
             face_in,
             camera_shot,
             condition_shot,
+            controlnet_selection,
             prompt,
             style,
             chosen_model