Spaces:
Sleeping
Sleeping
File size: 11,006 Bytes
5140a89 e575ced 5140a89 6ecc602 5140a89 e575ced e4b3a75 e575ced 6ecc602 e575ced ee4b9c0 e575ced ee4b9c0 e575ced ee4b9c0 e575ced ee4b9c0 e575ced ee4b9c0 e575ced 5140a89 ee4b9c0 6ecc602 ee4b9c0 6ecc602 ee4b9c0 6ecc602 ee4b9c0 6ecc602 ee4b9c0 e4b3a75 ee4b9c0 6ecc602 ee4b9c0 8e46c45 ee4b9c0 5140a89 ee4b9c0 6ecc602 ee4b9c0 8e46c45 6ecc602 ee4b9c0 6ecc602 ee4b9c0 6ecc602 ee4b9c0 6ecc602 ee4b9c0 8e46c45 6ecc602 ee4b9c0 5140a89 ee4b9c0 5140a89 ee4b9c0 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 |
import streamlit as st
from pathlib import Path
import torch
from transformers import pipeline
from PIL import Image, ImageDraw, ImageFont
import tempfile
import os
from moviepy.editor import *
import numpy as np
from gtts import gTTS
import textwrap
from concurrent.futures import ThreadPoolExecutor
import io
import unicodedata
import re
class FastVideoGenerator:
def __init__(self):
self.device = "cuda" if torch.cuda.is_available() else "cpu"
# Initialize text generation with efficient model
self.text_generator = pipeline(
'text-generation',
model='distilgpt2',
device=0 if self.device == "cuda" else -1
)
# Create temp directory
self.temp_dir = Path(tempfile.mkdtemp())
# Theme colors with opacity for better text visibility
self.themes = {
'Professional': {
'bg': (245, 245, 245),
'text': (33, 33, 33),
'accent': (0, 102, 204),
'overlay': (255, 255, 255, 180)
},
'Creative': {
'bg': (255, 240, 245),
'text': (51, 51, 51),
'accent': (255, 64, 129),
'overlay': (255, 255, 255, 180)
},
'Educational': {
'bg': (240, 249, 255),
'text': (25, 25, 25),
'accent': (0, 151, 167),
'overlay': (255, 255, 255, 180)
}
}
# Pre-load font
try:
self.font = ImageFont.truetype("arial.ttf", 40)
except:
self.font = ImageFont.load_default()
# Add text cleaner
self.text_cleaner = re.compile(r'[^\x00-\x7F]+')
@staticmethod
def clean_text(text):
"""Clean text to handle encoding issues"""
# Normalize unicode characters
text = unicodedata.normalize('NFKD', text)
# Replace special characters with standard ASCII
text = text.encode('ascii', 'ignore').decode('ascii')
# Replace common special characters
replacements = {
'โ': '-', # en dash
'โ': '-', # em dash
''': "'", # curly quote
''': "'", # curly quote
'"': '"', # curly quote
'"': '"', # curly quote
'โฆ': '...' # ellipsis
}
for old, new in replacements.items():
text = text.replace(old, new)
return text
@staticmethod
@st.cache_data
def generate_script_cached(prompt, style, length, temperature=0.7):
"""Cached script generation with proper text cleaning"""
style_prompts = {
'Professional': "Write a clear, professional video script about:",
'Creative': "Write an engaging, creative video script about:",
'Educational': "Write an informative educational video script about:"
}
prompt = FastVideoGenerator.clean_text(prompt)
with pipeline('text-generation', model='distilgpt2') as generator:
output = generator(
f"{style_prompts[style]} {prompt}. Make it {length} seconds long.",
max_length=min(length * 3, 1000),
num_return_sequences=1,
temperature=temperature
)
script = output[0]['generated_text']
script = script.replace(style_prompts[style], '').strip()
return FastVideoGenerator.clean_text(script)
def create_frame_fast(self, text, theme, frame_number, total_frames, size=(1280, 720)):
"""Create frame with cleaned text"""
# Clean text before rendering
text = self.clean_text(text)
# Create frame
frame = np.full((size[1], size[0], 3), theme['bg'], dtype=np.uint8)
img = Image.fromarray(frame)
draw = ImageDraw.Draw(img)
# Wrap text for better presentation
wrapped_text = textwrap.fill(text, width=50)
# Calculate text position
text_bbox = draw.textbbox((0, 0), wrapped_text, font=self.font)
text_x = (size[0] - (text_bbox[2] - text_bbox[0])) // 2
text_y = (size[1] - (text_bbox[3] - text_bbox[1])) // 2
# Draw text with background for better readability
text_bg = Image.new('RGBA', size, (0, 0, 0, 0))
text_draw = ImageDraw.Draw(text_bg)
text_draw.text((text_x, text_y), wrapped_text, fill=theme['text'], font=self.font)
# Add progress bar
progress = frame_number / total_frames
bar_width = int(1000 * progress)
draw.rectangle([140, 650, 1140, 660], fill=(200,200,200))
draw.rectangle([140, 650, 140+bar_width, 660], fill=theme['accent'])
return np.array(img)
def generate_audio_chunks(self, script, chunk_size=1000):
"""Generate audio with cleaned text"""
# Clean text before TTS
script = self.clean_text(script)
chunks = textwrap.wrap(script, chunk_size)
audio_paths = []
for i, chunk in enumerate(chunks):
chunk_path = self.temp_dir / f"audio_chunk_{i}.mp3"
try:
tts = gTTS(text=chunk, lang='en', slow=False)
tts.save(str(chunk_path))
audio_paths.append(chunk_path)
except Exception as e:
# If TTS fails, try with further cleaning
cleaned_chunk = re.sub(r'[^a-zA-Z0-9\s.,!?-]', '', chunk)
tts = gTTS(text=cleaned_chunk, lang='en', slow=False)
tts.save(str(chunk_path))
audio_paths.append(chunk_path)
return audio_paths
def create_optimized_video(self, script, theme, duration=30):
"""Create video with optimized processing"""
fps = 24
total_frames = duration * fps
# Create frames efficiently
def make_frame(t):
frame_number = int(t * fps)
return self.create_frame_fast(
script,
self.themes[theme],
frame_number,
total_frames
)
# Generate video with reduced memory usage
clip = VideoClip(make_frame, duration=duration)
# Generate audio in background while processing video
with ThreadPoolExecutor() as executor:
future_audio = executor.submit(self.generate_audio_chunks, script)
# Process video
output_path = self.temp_dir / "output_video.mp4"
temp_video = self.temp_dir / "temp_video.mp4"
# Write video without audio first
clip.write_videofile(
str(temp_video),
fps=fps,
codec='libx264',
audio=False,
preset='ultrafast'
)
# Get audio paths and combine audio
audio_paths = future_audio.result()
audio_clips = [AudioFileClip(str(path)) for path in audio_paths]
final_audio = concatenate_audioclips(audio_clips)
# Combine video and audio
video = VideoFileClip(str(temp_video))
final_clip = video.set_audio(final_audio)
final_clip.write_videofile(str(output_path), fps=fps, codec='libx264')
# Cleanup
video.close()
final_clip.close()
for clip in audio_clips:
clip.close()
return output_path
def main():
st.set_page_config(
page_title="โก Fast Video Generator",
layout="wide",
initial_sidebar_state="expanded"
)
# Custom CSS
st.markdown("""
<style>
.stButton>button {
width: 100%;
height: 3em;
background-color: #FF4B4B;
color: white;
}
.stProgress > div > div > div > div {
background-color: #FF4B4B;
}
</style>
""", unsafe_allow_html=True)
if 'video_generator' not in st.session_state:
st.session_state.video_generator = FastVideoGenerator()
with st.sidebar:
st.title("๐ฎ Video Settings")
theme = st.selectbox(
"Theme Style",
["Professional", "Creative", "Educational"],
help="Choose the visual style of your video"
)
duration = st.slider(
"Duration (seconds)",
min_value=30,
max_value=300,
value=60,
step=30,
help="Videos up to 5 minutes supported"
)
quality = st.select_slider(
"Generation Speed",
options=["High Quality", "Balanced", "Fast"],
value="Balanced",
help="Faster generation may reduce video quality"
)
st.title("โก Fast Video Generator")
st.markdown("Create longer videos with optimized performance!")
text_input = st.text_area(
"Video Topic",
height=100,
placeholder="Enter your topic here..."
)
if st.button("๐ฌ Generate Video", use_container_width=True):
if text_input:
try:
progress_bar = st.progress(0)
status = st.empty()
# Script generation
status.text("โ๏ธ Creating script...")
script = FastVideoGenerator.generate_script_cached(
text_input, theme, duration
)
progress_bar.progress(30)
# Video creation
status.text("๐จ Generating video...")
video_path = st.session_state.video_generator.create_optimized_video(
script, theme, duration
)
progress_bar.progress(100)
status.text("โจ Video ready!")
# Display results
tab1, tab2 = st.tabs(["๐ฝ๏ธ Video", "๐ Script"])
with tab1:
st.video(str(video_path))
with open(str(video_path), 'rb') as f:
st.download_button(
"โฌ๏ธ Download Video",
f,
file_name="generated_video.mp4",
mime="video/mp4"
)
with tab2:
st.markdown("### Generated Script")
st.write(script)
except Exception as e:
st.error(f"๐ฅ Error: {str(e)}")
st.error("Please try again with different settings")
else:
st.warning("โ ๏ธ Please enter a topic first!")
if __name__ == "__main__":
main()
|