FFmpeg

Input Seeking (Major Difference)

1. -ss BEFORE -i: fast seek, may be inaccurate—starts from nearest keyframe
2. -ss AFTER -i: frame-accurate but slow—decodes from start
3. Combine both: -ss 00:30:00 -i input.mp4 -ss 00:00:05—fast seek then accurate trim
4. For cutting, add -avoid_negative_ts make_zero to fix timestamp issues

Stream Selection

1. Default: first video + first audio—may not be what you want
2. Explicit selection: -map 0:v:0 -map 0:a:1—first video, second audio
3. All streams of type: -map 0:a—all audio streams
4. Copy specific: -map 0 -c copy—all streams, no re-encoding
5. Exclude: -map 0 -map -0:s—all except subtitles

Encoding Quality

1. CRF (Constant Rate Factor): lower = better quality, larger file—18-23 typical for H.264
2. -preset: ultrafast to veryslow—slower = smaller file at same quality
3. Two-pass for target bitrate: first pass analyzes, second pass encodes
4. -crf and -b:v mutually exclusive—use one or the other

Container vs Codec

1. Container (MP4, MKV, WebM): wrapper format holding streams
2. Codec (H.264, VP9, AAC): compression algorithm for stream
3. Not all codecs fit all containers—H.264 in MP4/MKV, not WebM; VP9 in WebM/MKV, not MP4
4. Copy codec to new container: -c copy—fast, no quality loss

Filter Syntax

1. Simple: -vf "scale=1280:720"—single filter chain
2. Complex: -filter_complex "[0:v]scale=1280:720[scaled]"—named outputs for routing
3. Chain filters: -vf "scale=1280:720,fps=30"—comma-separated
4. Filter order matters—scale before crop gives different result than crop before scale

Common Filters

1. Scale: scale=1280:720 or scale=-1:720 for auto-width maintaining aspect
2. Crop: crop=640:480:100:50—width:height:x:y from top-left
3. FPS: fps=30—change framerate
4. Trim: trim=start=10:end=20,setpts=PTS-STARTPTS—setpts resets timestamps
5. Overlay: overlay=10:10—position from top-left

Audio Processing

1. Sample rate: -ar 48000—standard for video
2. Channels: -ac 2—stereo
3. Audio codec: -c:a aac -b:a 192k—AAC at 192kbps
4. Normalize: -filter:a loudnorm—EBU R128 loudness normalization
5. Extract audio: -vn -c:a copy output.m4a—no video, copy audio

Concatenation

1. Same codec/params: concat demuxer—-f concat -safe 0 -i list.txt -c copy
2. Different formats: concat filter—-filter_complex "[0:v][0:a][1:v][1:a]concat=n=2:v=1:a=1"
3. list.txt format: file 'video1.mp4' per line—escape special characters
4. Different resolutions: scale/pad to match before concat filter

Subtitles

1. Burn-in (hardcode): -vf "subtitles=subs.srt"—cannot be turned off
2. Mux as stream: -c:s mov_text (MP4) or -c:s srt (MKV)—user toggleable
3. From input: -map 0:s—include subtitle streams
4. Extract: -map 0:s:0 subs.srt—first subtitle to file

Hardware Acceleration

1. Decode: -hwaccel cuda or -hwaccel videotoolbox (macOS)
2. Encode: -c:v h264_nvenc (NVIDIA), -c:v h264_videotoolbox (macOS)
3. Not always faster—setup overhead; benefits show on long videos
4. Quality may differ—software encoding often produces better quality

Common Mistakes

1. Forgetting -c copy when not re-encoding—defaults to re-encode, slow and lossy
2. -ss after -i for long videos—takes forever seeking
3. Audio desync after cutting—use -async 1 or -af aresample=async=1
4. Filter on stream copy—filters require re-encoding; -c copy + -vf = error
5. Output extension doesn't set codec—output.mp4 without -c:v uses default, may not be H.264