Update app.py

app.py

@@ -1,35 +1,28 @@
 import torch
-import gradio as gr
-from diffusers import StableDiffusionPipeline
+from diffusers.utils import export_to_video
+from diffusers import AutoencoderKLWan, WanPipeline
+from diffusers.schedulers.scheduling_unipc_multistep import UniPCMultistepScheduler
 
-# Load model manually from Hugging Face model hub or your uploaded files
-model_path = "sarthak247/Wan2.1-T2V-1.3B-nf4"  # Replace with your model path
-pipe = StableDiffusionPipeline.from_pretrained(model_path, torch_dtype=torch.float16)
-pipe.to("cuda")  # If running on GPU
+model_id = "Wan-AI/Wan2.1-T2V-14B-Diffusers"
+vae = AutoencoderKLWan.from_pretrained(model_id, subfolder="vae", torch_dtype=torch.float32)
+pipe = WanPipeline.from_pretrained(model_id, vae=vae, torch_dtype=torch.bfloat16)
+flow_shift = 3.0  # 5.0 for 720P, 3.0 for 480P
+pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config, flow_shift=flow_shift)
+pipe.to("cuda")
 
-def generate_video(prompt):
-    """
-    Generates a video from the provided prompt using the pre-loaded model.
-    """
-    try:
-        # Generate video using the model pipeline
-        video = pipe(prompt).videos[0]  # Assuming output is a video tensor
-        
-        # Return the generated video
-        return video
+pipe.load_lora_weights("NIVEDAN/wan2.1-lora")
 
-    except Exception as e:
-        print(f"Error during video generation: {e}")
-        return "Error generating video"
+pipe.enable_model_cpu_offload() #for low-vram environments
 
-# Gradio UI for video generation
-iface = gr.Interface(
-    fn=generate_video,
-    inputs=gr.Textbox(label="Enter Text Prompt"),
-    outputs=gr.Video(label="Generated Video"),
-    title="Text-to-Video Generation with Wan2.1-T2V",
-    description="This app generates a video based on the text prompt using the Wan2.1-T2V model."
-)
+prompt = "nivedan"
+negative_prompt = "Bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards"
 
-# Launch the Gradio app
-iface.launch()
+output = pipe(
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    height=480,
+    width=832,
+    num_frames=81,
+    guidance_scale=5.0,
+).frames[0]
+export_to_video(output, "output.mp4", fps=16)