How to extract broadcaster watermark
./logo_extractor input_file.ts logo_name.txt start_time end_time
- Some of logo_name.txt is available under logo/ directory.
- Specify analysis range of input_file.ts by start_time and end_time.
- All frames in analysis range should have broadcaster warermark.
./mecenc input_file_1.ts [input_file_2.ts ...]
- CM detection.
- Encode by x264.
- Convert the frame ratio from keep_fps to 24fps and deinterlace.
- Shrink the movie to 1280 x 720 since almost all animations are created in this resolution.
- Don't use broadcaster watermark detection for CM detection.
- Working / output / log directory is in ~/enc/.
./mecenc --encoder x265 input_file.ts
Enable broadcaster watermark detection
./mecenc --logo logo_name input_file.ts
- logo_name.png and logo_name.txt should be in a logo directory.
CM detection only (don't encode)
./mecenc --analyze input_file.ts
Encode with detected CM information (single files)
./mecenc --scenefile /path/to/scene.txt input_file.ts
Encode with detected CM information (multiple files)
./mecenc --scenelistfile /path/to/yyyy-mm-dd_hhmm_scenelist.txt
Encode with interlaced movie
./mecenc --interlaced input_file.ts
Encode with keeping the original fps (mainly for 30 fps animation)
./mecenc --keep_fps input_file.ts
./mecenc --no_scale input_file.ts
./mecenc --tempdir tempdir --destdir destdir --logdir logdir input_file.ts
Don't clean the working directory
./mecenc --no_clean input_file.ts
Enable aggressive CM analysis mainly for manual CM detection.
./mecenc --aggressive_analysis input_file.ts
./mecenc --no_lock input_file.ts
- Without this option, mecenc creates /tmp/encode_movie.lock to lock other process.
- g++
- python-opencv
- python-numpy
- libopencv-dev
- libstdc++6:i386 # for NeroAacEnc(32bit) on 64bit Linux.
- ImageMagick
- sox
- x264
- x265 (optional)
- neroAacEnc
- ffmpeg (libx264, libx265 and libpng support is required)
- tesseract-ocr
- tesseract-ocr-jpn
- TBD...