test gradio

app.py

@@ -80,6 +80,39 @@ gif_pipe.load_lora_weights(lora_adapter_id, adapter_name="motion_lora")
 
 
 
+@spaces.GPU
+def generate_frames(prompt, reference_image, controlnet_conditioning_scale,num_frames):
+    massage_history = [{"role": "system", "content": """
+You are a scene designer tasked with creating sparse frames of a video. You will be given a prompt describing the desired video, and your goal is to design only the key frames (sparse frames) that represent major changes in the scene. Do not include repetitive or similar scenes—only capture distinct moments.
+
+Expected Format:
+Return the response as a JSON object with the key "frames". The value should be a list of dictionaries, where each dictionary has:
+
+"frame_index": an integer indicating the frame's position in the sequence.
+"description": a brief description of the scene in this frame.
+Example:
+If given a prompt like "A sunset over a beach with waves crashing and a ship sailing by," your response should look like this:
+
+```json
+{
+  "frames": [
+    {"frame_index": 0, "description": "Sunset over an empty beach, sky turning orange and pink"},
+    {"frame_index": 30, "description": "Waves gently crashing on the shore"},
+    {"frame_index": 60, "description": "A ship appears on the horizon, silhouetted by the sunset"},
+    {"frame_index": 90, "description": "Ship sailing closer, with waves becoming more dynamic"},
+    {"frame_index": 120, "description": "Sun dipping below the horizon, casting a golden glow over the water"}
+  ]
+}
+```
+This way, each frame represents a distinct scene, and there’s no redundancy between them."""},
+                       {"role": "user", "content": f"give me the frames to generate a video with prompt : `{prompt}`"},]
+    frames = ask_gpt(massage_history,return_str=False)['frames']
+    conditioning_frames = []
+    controlnet_frame_indices =[]
+    for frame in frames:
+        conditioning_frames.append(generate_image(frame['description'], reference_image, controlnet_conditioning_scale))
+        controlnet_frame_indices.append(frame['frame_index'])
+    return conditioning_frames
 
 @spaces.GPU
 def generate_gif(prompt, reference_image, controlnet_conditioning_scale,num_frames):
@@ -126,6 +159,23 @@ This way, each frame represents a distinct scene, and there’s no redundancy be
 
     return "animation.gif"
 
+# # Set up Gradio interface
+# interface = gr.Interface(
+#     fn=generate_gif,
+#     inputs=[
+#         gr.Textbox(label="Prompt"),
+#         # gr.Image( type= "filepath",label="Reference Image (Style)"),
+#         gr.File(type="filepath",file_count="multiple",label="Reference Image (Style)"),
+#         gr.Slider(label="Control Net Conditioning Scale", minimum=0, maximum=1.0, step=0.1, value=1.0),
+#         gr.Slider(label="Number of frames", minimum=0, maximum=1.0, step=0.1, value=1.0),
+#
+#     ],
+#     outputs="image",
+#     title="Image Generation with Stable Diffusion 3 medium and ControlNet",
+#     description="Generates an image based on a text prompt and a reference image using Stable Diffusion 3 medium with ControlNet."
+#
+# )
+
 # Set up Gradio interface
 interface = gr.Interface(
     fn=generate_gif,
@@ -137,7 +187,7 @@ interface = gr.Interface(
         gr.Slider(label="Number of frames", minimum=0, maximum=1.0, step=0.1, value=1.0),
 
     ],
-    outputs="image",
+    outputs="gallery",
     title="Image Generation with Stable Diffusion 3 medium and ControlNet",
     description="Generates an image based on a text prompt and a reference image using Stable Diffusion 3 medium with ControlNet."
 

requirements.txt

@@ -6,4 +6,5 @@ gradio
 sentencepiece
 spaces
 pillow
-peft
+peft
+openai