AnimationDiff / app.py
Ai
Update app.py
2b29e94
raw
history blame
2.2 kB
import gradio as gr
import torch
from diffusers import AnimateDiffPipeline, MotionAdapter, DDIMScheduler
from diffusers.utils import export_to_gif
import random
def generate_gif(image, animation_type):
# Load the motion adapter
adapter = MotionAdapter.from_pretrained("guoyww/animatediff-motion-adapter-v1-5-2", torch_dtype=torch.float16)
# Load SD 1.5 based finetuned model
model_id = "SG161222/Realistic_Vision_V6.0_B1_noVAE"
pipe = AnimateDiffPipeline.from_pretrained(model_id, motion_adapter=adapter, torch_dtype=torch.float16)
# Scheduler setup
scheduler = DDIMScheduler(
clip_sample=False,
beta_start=0.00085,
beta_end=0.012,
beta_schedule="linear",
timestep_spacing="trailing",
steps_offset=1
)
pipe.scheduler = scheduler
# Enable memory savings
pipe.enable_vae_slicing()
pipe.enable_model_cpu_offload()
# Load ip_adapter
pipe.load_ip_adapter("h94/IP-Adapter", subfolder="models", weight_name="ip-adapter_sd15.bin")
# Load the selected motion adapter
pipe.load_lora_weights(f"guoyww/animatediff-motion-lora-{animation_type}", adapter_name=animation_type)
# Generate a random seed
seed = random.randint(0, 2**32 - 1)
prompt = "best quality, high quality,photorealisric"
# Set adapter weights for the selected adapter
adapter_weight = [0.75]
pipe.set_adapters([animation_type], adapter_weights=adapter_weight)
# Generate GIF
output = pipe(
prompt=prompt,
num_frames=16,
guidance_scale=7.5,
num_inference_steps=30,
ip_adapter_image=image,
generator=torch.Generator("cpu").manual_seed(seed),
)
frames = output.frames[0]
gif_path = "output_animation.gif"
export_to_gif(frames, gif_path)
return gif_path
# Gradio interface
iface = gr.Interface(
fn=generate_gif,
inputs=[gr.Image(type="pil"), gr.Radio(["zoom-out", "tilt-up", "pan-left"])],
outputs=gr.Image(type="pil", label="Generated GIF"),
title="AnimateDiff + IP Adapter Demo",
description="Upload an image and select an motion module type to generate a GIF!"
)
iface.launch(debug=True,share=True)