new approach for animate diff model

app.py

@@ -1,30 +1,36 @@
 import gradio as gr
 import torch
-from diffusers import AnimateDiffPipeline
-from PIL import Image
-import tempfile
-import os
+from diffusers import AnimateDiffPipeline, MotionAdapter, DDIMScheduler
+from diffusers.utils import export_to_gif
 
-pipe = AnimateDiffPipeline.from_pretrained(
-    "ByteDance/AnimateDiff-Lightning",
-    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32
+# Load the motion adapter
+adapter = MotionAdapter.from_pretrained("guoyww/animatediff-motion-adapter-v1-5-2")
+# Load the base model
+model_id = "SG161222/Realistic_Vision_V5.1_noVAE"
+pipe = AnimateDiffPipeline.from_pretrained(model_id, motion_adapter=adapter)
+# Set up the scheduler
+scheduler = DDIMScheduler.from_pretrained(
+    model_id, subfolder="scheduler", clip_sample=False, timestep_spacing="linspace", steps_offset=1
 )
-pipe.to("cuda" if torch.cuda.is_available() else "cpu")
+pipe.scheduler = scheduler
 
-def generate_video(prompt):
-    frames = pipe(prompt).frames
+# Enable memory savings
+pipe.enable_vae_slicing()
+pipe.enable_model_cpu_offload()
 
-    with tempfile.NamedTemporaryFile(suffix=".mp4", delete=False) as temp_video:
-        pipe.export_to_video(frames, temp_video.name)
-        return temp_video.name
-
-demo = gr.Interface(
-    fn=generate_video,
-    inputs=gr.Textbox(label="Prompt"),
-    outputs=gr.Video(label="Generated Animation"),
-    title="AnimateDiff Demo",
-    description="Generate animations from text prompts using AnimateDiff."
-)
+def generate_animation(prompt):
+    output = pipe(
+        prompt=prompt,
+        negative_prompt="bad quality, worse quality",
+        num_frames=16,
+        guidance_scale=7.5,
+        num_inference_steps=25,
+        generator=torch.Generator("cpu").manual_seed(42),
+    )
+    frames = output.frames[0]
+    export_to_gif(frames, "animation.gif")
+    return "animation.gif"
 
+# Create Gradio interface
+demo = gr.Interface(fn=generate_animation, inputs="text", outputs="image")
 demo.launch()
-