app.py

import streamlit as st
from moviepy.editor import *
import tempfile, os
from transformers import pipeline

st.title("Simplified Text-to-Video (No TTS temporarily)")

@st.cache_resource()
def load_model():
    return pipeline('text-to-video-generation', model='cerspense/zeroscope_v2_576w')

video_gen = load_model()

input_text = st.text_area("Enter short text (max 50 chars):", max_chars=50)

if st.button("Generate Video"):
    if input_text:
        with st.spinner("Generating video..."):
            video_output = video_gen(input_text, num_frames=10)
            video_tensor = video_output["video"]
            video_np = (video_tensor * 255).astype('uint8')

            video_filename = tempfile.mktemp(".mp4")
            clips = [ImageClip(frame).set_duration(0.3) for frame in video_np]
            final_clip = concatenate_videoclips(clips, method="compose")
            final_clip.write_videofile(video_filename, fps=5)

            st.video(video_filename)

            os.remove(video_filename)
    else:
        st.warning("Please enter text.")