Ffmpeg command

FFmpeg Command Builder

Translate a described media operation into a correct, efficient FFmpeg command.

Read the matching reference — every time

Before writing any command, identify the matching reference file(s) and read them with the Read tool. The recipes and gotchas in this file are summaries for quick recall. The references hold the parameter ranges, alternatives, edge cases, and tribal knowledge that separate a working command from a plausible-looking broken one.

This is non-negotiable — even when a recipe below seems to fit. Recipes show shape; references show why and what breaks. Skipping the read step is how broken commands get shipped.

If the operation involves...	Read
Any non-trivial command (read every time)	references/pitfalls.md
Codec choice, CRF, preset, tune, faststart, yuv420p, hvc1, 1-/2-pass	references/encoding.md
Filter graphs, -vf/-af/-filter_complex, stream selectors, split/concat/overlay, expressions	references/filters.md
Trimming, seeking, accuracy vs speed, -c copy interactions	references/seeking-and-trimming.md
Audio (extract, replace, mix, fades, format, normalize, channels)	references/audio.md
Scale, pad, crop, fps, speed, overlay, fades, vstack/hstack, jump cuts	references/video-effects.md
Drawtext or subtitles (burn or mux), ASS, font handling	references/text-and-subtitles.md
Image→video, slideshow, Ken Burns/zoompan, GIFs, thumbnails, storyboards	references/asset-generation.md
GPU encoding (NVENC, QSV, VAAPI)	references/gpu-acceleration.md
HLS packaging, VOD/live segments, ABR multi-bitrate	references/streaming.md

If the operation spans categories (e.g., "burn subtitles with custom styling and re-encode for web"), read every relevant reference. Always include pitfalls.md for non-trivial commands.

Workflow

1. Identify the operation type. Transcode? Remux? Trim? Filter? Generate from images? Mix audio?
2. Read the matching reference(s). Use the table above. Always include pitfalls.md for non-trivial commands. Use the Read tool — do not generate the command from memory.
3. Probe inputs only when codec, duration, or stream layout matters. Run ffprobe -show_streams -i <input> first if the request depends on input properties (e.g., "match the source frame rate", "keep the original audio codec", "trim the last 5 seconds"). Skip for plain remux/transcode.
4. Decide stream copy vs re-encode.
   • -c copy (remux only): changing container, muxing tracks, joining keyframe-aligned segments, moving metadata (faststart).
   • Re-encode: applying any filter, changing codec, frame-accurate trimming, compressing.
   • Details and edge cases: references/seeking-and-trimming.md, references/encoding.md.
5. Choose codec parameters for the target use case. Defaults below; verify against references/encoding.md for CRF ranges, preset trade-offs, and rate-control modes.
   • Web/VOD/archival: libx264 -crf 18 -preset veryslow -tune fastdecode -movflags +faststart
   • Apple H265: add -vtag hvc1
   • WebM: libvpx-vp9 -crf 31 -b:v 0 -c:a libopus
   • Live streaming: -preset ultrafast -tune zerolatency
6. Place flags in the correct position. Input flags (-ss, -loop, -t, -r, -skip_frame) come before -i; output flags (-c:v, -vf, -crf, -movflags, output path) come after.
7. Apply playback-compatibility defaults unless the user has a reason not to: format=yuv420p (or -pix_fmt yuv420p), setsar=1:1 after scale+pad, +faststart for web MP4. Full list in references/pitfalls.md.

Axis declaration

Many operations have a meaningful trade-off axis where the right default depends on user intent: GIF quality vs size, encoder preset speed vs compression, single-pass vs two-pass, CBR vs VBR, fast seek vs frame-accurate. Recipes for operations on a trade-off axis MUST:

1. Name the axis in the section heading or first sentence (e.g., "GIF — quality-first" vs "GIF — size-first").
2. Show both branches inline. Do not bury the alternative under "advanced".
3. List the user signals that flip the default. Words like "quality matters more than size", "for archival", "for fast turnaround", "bandwidth-constrained", "size budget under N MB", "live", "VOD" each push the choice.

Before writing a command for a trade-off-axis operation, re-read the user's request and pick the branch their wording supports. Do not lock in a single default just because the recipe shows one.

If you find yourself writing one set of defaults but the user's request mentions the opposite axis, you've taken the wrong branch — switch.

Quick recipes

Canonical shapes for common operations — for quick recall after you've already read the reference. The reference is authoritative; the recipe is a stub. Adapt input/output paths and parameter values; do not reorder flags. The linked reference covers variants, parameter ranges, and gotchas the recipe omits.

Remux / transcode

Remux MP4 to MKV (no re-encode):

ffmpeg -i input.mp4 -c copy output.mkv

Generic web-optimized re-encode (good default for archiving, streaming, edge devices):

ffmpeg -i input.mp4 -vf "scale=w=1080:h=1920:force_original_aspect_ratio=decrease,pad=1080:1920:(ow-iw)/2:(oh-ih)/2:color=black,setsar=1:1" -crf 18 -preset veryslow  -threads 0 -tune fastdecode  -movflags +faststart  output_scaled_optimized.mp4

→ references/encoding.md

Archive transcode (legacy AVI/WMV/MPEG-2 — likely interlaced)

ffmpeg -i input.avi -vf "yadif=mode=1:parity=auto,format=yuv420p" -c:v libx264 -crf 20 -preset slow -c:a aac -b:a 128k -movflags +faststart output.mp4

→ references/encoding.md (Archive transcode section)

Trim by time (frame-accurate)

ffmpeg -i input.mp4 -ss 00:00:10 -to 00:00:25 output_trimmed.mp4

→ references/seeking-and-trimming.md

Audio

Extract AAC without re-encoding:

ffmpeg -i  input.mp4 -map 0:a:0 -acodec copy output.aac

Replace audio in video (trim video to audio length):

ffmpeg -i input.mp4 -i input.mp3 -map 0:v -map 1:a  -shortest -c:v copy -c:a aac output_replace_audio.mp4

Mix new audio over original at lower volume:

ffmpeg -i input.mp4 -i input.mp3 -filter_complex "[1:a]volume=0.2[a1];[0:a][a1]amix=inputs=2:duration=shortest" -shortest -map 0:v -c:v copy -c:a aac output_mix_audio.mp4

→ references/audio.md

Resize / pad

Scale to 1080×1920 keeping aspect ratio with black padding:

ffmpeg -i input.mp4 -vf "scale=w=1080:h=1920:force_original_aspect_ratio=decrease,pad=1080:1920:(ow-iw)/2:(oh-ih)/2:color=black,setsar=1:1" output_resized_pad.mp4

→ references/video-effects.md

Overlay logo (timed)

ffmpeg -i input.mp4 -i logo.png -filter_complex "overlay=x=(main_w-overlay_w)/8:y=(main_h-overlay_h)/8:enable='gte(t,1)*lte(t,7)'" -c:v libx264 -c:a  output_logo.mp4

→ references/video-effects.md

Burn subtitles (custom font)

ffmpeg -i input.mp4 -ss 00:00 -to 00:40 -vf "subtitles=sample_subtitles.srt:fontsdir=.:force_style='FontName=Poppins,FontSize=24,PrimaryColour=&HFFFFFF,OutlineColour=&H4066B66B,Outline=1,BorderStyle=3'" -c:v libx264  -c:a copy output_subtitles.mp4

→ references/text-and-subtitles.md

Speed change (1.5×, no audio pitch shift)

ffmpeg -i input.mp4 -filter_complex "[0:v]setpts=PTS/1.5[v];[0:a]atempo=1.5[a]" -map "[v]" -map "[a]" output_sped_up.mp4

→ references/video-effects.md

Thumbnail at second 7

ffmpeg -i input.mp4 -ss 00:00:07 -frames:v 1 output_thumbnail.png

GIF (quality-first canonical — pick the branch matching user intent)

ffmpeg -i input.mp4 -vf "fps=20,scale=640:-1:flags=lanczos,split[s0][s1];[s0]palettegen=stats_mode=full[p];[s1][p]paletteuse=dither=sierra2_4a" -loop 0 -an output.gif

This is the quality-first default. For size-first (smaller file, lower fps/resolution, stats_mode=diff + bayer dither) or the stylized fast-cut variant, see the GIF section in the reference — it has both branches and a chooser table.

→ references/asset-generation.md

GPU transcode (NVIDIA)

ffmpeg -i input.avi -c:v h264_nvenc output_gpu_264.mp4

→ references/gpu-acceleration.md

HLS VOD (single rendition — rendi.dev house style)

ffmpeg -i input.mp4 -c:v h264 -c:a aac -b:v 2000k -b:a 128k -hls_time 5 -hls_list_size 0 -f hls playlist.m3u8

This is the minimalist form. For hardened-VOD playback (older players, ABR-ready, explicit GOP alignment) or multi-bitrate ABR, see the reference. The reference also has a parallel DASH recipe.

→ references/streaming.md

Top gotchas (summary — pitfalls.md is the source of truth)

The most common LLM failure modes when generating FFmpeg commands. Read references/pitfalls.md for the full 20-item list with explanations. The summary below is for quick recall; trust the reference over the summary if they ever diverge.

1. format=yuv420p (or -pix_fmt yuv420p) when re-encoding video for general playback — without it QuickTime and many devices refuse the file or render wrong colors.
2. -movflags +faststart for web MP4/MOV/M4A — moves the moov atom to the front so playback starts before download finishes.
3. -vtag hvc1 for H265 + Apple devices — without it AirDrop/QuickTime reject the file.
4. setsar=1:1 after scale + pad — otherwise FFmpeg may set a non-square SAR to compensate, distorting output on some players.
5. -c copy is incompatible with filters and frame-accurate trimming. Filters require re-encoding; copy can only cut at keyframes (often produces black frames at the start).
6. Input flags before -i, output flags after -i. -ss/-loop/-t before apply to the input; after, they apply to the output.
7. After trim/atrim, reset timestamps with setpts=PTS-STARTPTS / asetpts=PTS-STARTPTS before concat — otherwise concat produces wrong durations or gaps.
8. force_original_aspect_ratio=decrease shrinks-to-fit (then pad fills); =increase grows-to-cover (then crop). Pick deliberately.
9. Subtitles FontName= is the font's internal name (open the font file to see it), not the filename. Set fontsdir to the directory holding the font.
10. -vsync 0 is deprecated — use -fps_mode passthrough (or cfr/vfr).
11. zoompan jitters without first upscaling: scale=8000:-1,zoompan=....
12. Don't re-encode when remuxing. If only the container changes, use -c copy.

Glossary

Common flags and selectors:

• -vf / -filter:v — video filter chain
• -af / -filter:a — audio filter chain
• -filter_complex — multi-input/multi-output filter graph
• [0:v] — video stream of first input; [1:a] — audio stream of second input
• 0:v:0 — first input, first video stream (0-based)
• [name] — named filter output, used inside -filter_complex
• -map — select streams for output
• -y — overwrite output without prompting (add to every command)
• -c copy — stream copy (remux without re-encoding)
• -c:v / -c:a — video / audio codec (libx264, libx265, libvpx-vp9, aac, libmp3lame, libopus, pcm_s32le)
• -an — disable audio in output
• -ss — seek (placement matters: input vs output)
• -shortest — finish output when shortest input ends