Update app.py

app.py

@@ -65,7 +65,7 @@ model = load_model()
 
 # Text-to-video generation function
 @spaces.GPU(duration=120)
-def generate_video(image, prompt, duration, guidance_scale, video_guidance_scale, progress=gr.Progress(track_tqdm=True)):
+def generate_video(prompt, image=None, duration=5, guidance_scale=9, video_guidance_scale=5, progress=gr.Progress(track_tqdm=True)):
  multiplier = 0.8 if is_canonical else 2.4
  temp = int(duration * 0.8) # Convert seconds to temp value (assuming 24 FPS)
  torch_dtype = torch.bfloat16 if MODEL_DTYPE == "bf16" else torch.float32
@@ -129,7 +129,8 @@ def generate_video(image, prompt, duration, guidance_scale, video_guidance_scale
 
 # Gradio interface
 with gr.Blocks() as demo:
- gr.Markdown("# Pyramid Flow text-to-video demo")
+ gr.Markdown("# Pyramid Flow")
+ gr.Markdown("Pyramid Flow is a training-efficient Autoregressive Video Generation model based on Flow Matching. It is trained only on open-source datasets within 20.7k A100 GPU hours")
  gr.Markdown("[[Paper](https://arxiv.org/pdf/2410.05954)], [[Model](https://huggingface.co/rain1011/pyramid-flow-sd3)], [[Code](https://github.com/jy0205/Pyramid-Flow)]")
  
  #with gr.Tab("Text-to-Video"):
@@ -155,9 +156,19 @@ with gr.Blocks() as demo:
  <p>to use privately and generate videos up to 10s</p>
  </div>
  """)
+ gr.Examples(
+ examples=[
+ "A movie trailer featuring the adventures of the 30 year old space man wearing a red wool knitted motorcycle helmet, blue sky, salt desert, cinematic style, shot on 35mm film, vivid colors",
+ "Beautiful, snowy Tokyo city is bustling. The camera moves through the bustling city street, following several people enjoying the beautiful snowy weather and shopping at nearby stalls. Gorgeous sakura petals are flying through the wind along with snowflakes"
+ ]
+ fn=generate_video,
+ inputs=prompt,
+ ouutputs=t2v_output,
+ cache_examples="lazy"
+ )
  t2v_generate_btn.click(
  generate_video,
- inputs=[i2v_image, t2v_prompt, t2v_duration, t2v_guidance_scale, t2v_video_guidance_scale],
+ inputs=[t2v_prompt, i2v_image, t2v_duration, t2v_guidance_scale, t2v_video_guidance_scale],
  outputs=t2v_output
  )