Upload 3 files

Dockerfile

@@ -4,7 +4,7 @@ WORKDIR /app
 
 # 安装FFmpeg和Noto Sans CJK SC字体
 RUN apt-get update && \
-    apt-get install -y ffmpeg fonts-noto-cjk fonts-noto-cjk-extra && \
+    apt-get install -y ffmpeg fonts-roboto && \
     apt-get clean && \
     rm -rf /var/lib/apt/lists/*
 

app.py

@@ -16,6 +16,8 @@ import ssl
 import json
 from fastapi.staticfiles import StaticFiles
 from pydub import AudioSegment
+import shlex
+from ffmpeg import probe as ffmpeg_probe  # 需要安装ffmpeg-python包
 
 # 配置日志
 logging.basicConfig(level=logging.INFO)
@@ -104,7 +106,7 @@ def get_audio_duration(audio_path):
 # 创建caption字幕文件（底部显示）
 def create_caption_subtitle_file(project_dir, captions, panel_start_times, panel_durations):
     try:
-        subtitle_file = os.path.join(project_dir, "captions.srt")
+        subtitle_file = os.path.join(project_dir, "captions.ass")
         
         with open(subtitle_file, "w", encoding="utf-8") as f:
             subtitle_index = 1
@@ -130,7 +132,7 @@ def create_caption_subtitle_file(project_dir, captions, panel_start_times, panel
 # 创建speech字幕文件（顶部显示）
 def create_speech_subtitle_file(project_dir, speeches, panel_start_times, panel_durations):
     try:
-        subtitle_file = os.path.join(project_dir, "speeches.srt")
+        subtitle_file = os.path.join(project_dir, "speeches.ass")
         
         with open(subtitle_file, "w", encoding="utf-8") as f:
             subtitle_index = 1
@@ -249,6 +251,16 @@ async def create_audio_file(project_dir, captions, speeches):
         logger.error(traceback.format_exc())
         return None, {}, [], []
 
+def get_video_dimensions(video_path):
+    """获取视频分辨率"""
+    try:
+        probe = ffmpeg_probe(video_path)
+        video_stream = next(s for s in probe['streams'] if s['codec_type'] == 'video')
+        return int(video_stream['width']), int(video_stream['height'])
+    except Exception as e:
+        logger.warning(f"Can't detect video dimensions: {e}, using default 1920x1080")
+        return (1920, 1080)
+
 # 创建视频
 def create_video(project_dir, image_paths, caption_subtitle_file, speech_subtitle_file, audio_file, output_video, audio_durations, panel_start_times, panel_durations):
     try:
@@ -276,30 +288,75 @@ def create_video(project_dir, image_paths, caption_subtitle_file, speech_subtitl
         ]
         subprocess.run(cmd1, check=True)
 
-        # 处理字幕路径转义（两个不同的字幕文件）
-        caption_path = caption_subtitle_file.replace(os.sep, '/').replace(':', '\\:')
-        speech_path = speech_subtitle_file.replace(os.sep, '/').replace(':', '\\:')
-        # 合并滤镜的正确写法（使用两个subtitles滤镜）
-        combined_filter = f"subtitles='{caption_path}':force_style='Alignment=2,WrapStyle=1,Fontsize=15,MarginV=30'," \
-                                    f"subtitles='{speech_path}':force_style='Alignment=8,WrapStyle=1,Fontsize=15,MarginV=30'"
+        # 动态获取视频参数
+        video_width, video_height = get_video_dimensions(temp_video)
+        base_fontsize = max(24, video_width // 50)
+        
+        # 智能样式计算
+        style_config = {
+            "caption": {
+                "fontsize": int(base_fontsize * 0.9),
+                "margin_v": video_height // 10,
+                "alignment": 2  # 底部居中
+            },
+            "speech": {
+                "fontsize": base_fontsize,
+                "margin_v": video_height // 12,
+                "alignment": 8  # 顶部居中
+            }
+        }
+        # 安全处理字幕路径
+        def process_sub_path(path):
+            return shlex.quote(
+                Path(path).resolve().as_posix()
+                    .replace(':', '\\:')
+            )
+        # 构建复合滤镜
+        combined_filter = (
+            f"subtitles={process_sub_path(caption_subtitle_file)}:"
+            f"force_style="
+            f"'Fontname=Roboto-Bold,"
+            f"Fontsize={style_config['caption']['fontsize']},"
+            f"Alignment={style_config['caption']['alignment']},"
+            f"MarginV={style_config['caption']['margin_v']},"
+            f"Outline=1,Shadow=1,BackColour=&H40000000,"
+            f"BorderStyle=3,WordWrap=1,WrapStyle=2;"
+            f"subtitles={process_sub_path(speech_subtitle_file)}:"
+            f"force_style="
+            f"'Fontname=Roboto-Bold,"
+            f"Fontsize={style_config['speech']['fontsize']},"
+            f"Alignment={style_config['speech']['alignment']},"
+            f"MarginV={style_config['speech']['margin_v']},"
+            f"Outline=1,Shadow=1,BackColour=&H40000000,"
+            f"BorderStyle=1,WordWrap=1,WrapStyle=2'"
+        )
+        # 优化视频处理命令
         cmd_combined = [
             "ffmpeg", "-y",
             "-i", temp_video,
             "-vf", combined_filter,
             "-c:a", "copy",
-            output_video  # 直接输出到最终文件，无需temp_video2
+            "-c:v", "libx264",
+            "-preset", "fast",
+            "-movflags", "+faststart",
+            "-x264-params", "log-level=error",
+            "-hide_banner",
+            "-loglevel", "error",
+            output_video
         ]
         
+        # 执行并计时
+        logger.info("Starting video processing with optimized subtitles")
+        start_time = time.time()
         subprocess.run(cmd_combined, check=True)
-        
-        # 删除临时文件（只需删除temp_video）
+        logger.info(f"Video processing completed in {time.time()-start_time:.2f}s")
+        # 清理临时文件
         os.remove(temp_video)
-        
+        Path(frames_list).unlink(missing_ok=True)
         return output_video
     except Exception as e:
-        logger.error(f"Error creating video: {e}")
-        import traceback
-        logger.error(traceback.format_exc())
+        logger.error(f"Video creation failed: {str(e)}")
+        logger.debug(f"Filter chain used: {combined_filter}")
         return None
 
 # 使用本地存储

requirements.txt

@@ -4,4 +4,5 @@ aiohttp>=3.8.4
 openai>=1.2.0
 python-multipart>=0.0.6
 pydantic>=1.10.7
-pydub
+pydub
+ffmpeg-python