Openai/6957d2e8-b738-8008-987a-09a5d9f29bc8

This conversation was exported from ChatGPT

gpt-5-2-thinking

User: ffmpeg cut from video the timestamsp 01:02 - 02:13 + 03:17 - 05:26 + 09:21 - 09:28 + 09:44 - 10:04 into a single video

ffmpeg cut from video the timestamsp 01:02 - 02:13 + 03:17 - 05:26 + 09:21 - 09:28 + 09:44 - 10:04 into a single video

Assistant: Here are two good ways to do it.

Here are two good ways to do it.

This will cut exactly at those timestamps and stitch into one file.

ffmpeg -i input.mp4 -filter_complex \
"[0:v]trim=start=62:end=133,setpts=PTS-STARTPTS[v0]; \
 [0:a]atrim=start=62:end=133,asetpts=PTS-STARTPTS[a0]; \
 [0:v]trim=start=197:end=326,setpts=PTS-STARTPTS[v1]; \
 [0:a]atrim=start=197:end=326,asetpts=PTS-STARTPTS[a1]; \
 [0:v]trim=start=561:end=568,setpts=PTS-STARTPTS[v2]; \
 [0:a]atrim=start=561:end=568,asetpts=PTS-STARTPTS[a2]; \
 [0:v]trim=start=584:end=604,setpts=PTS-STARTPTS[v3]; \
 [0:a]atrim=start=584:end=604,asetpts=PTS-STARTPTS[a3]; \
 [v0][a0][v1][a1][v2][a2][v3][a3]concat=n=4:v=1:a=1[v][a]" \
-map "[v]" -map "[a]" \
-c:v libx264 -crf 20 -preset medium \
-c:a aac -b:a 160k \
-movflags +faststart \
output.mp4

Timestamps converted to seconds:

• 01:02–02:13 → 62–133
• 03:17–05:26 → 197–326
• 09:21–09:28 → 561–568
• 09:44–10:04 → 584–604

This tries to avoid re-encoding. Often works, but accuracy depends on keyframes.

ffmpeg -ss 01:02 -to 02:13 -i input.mp4 -c copy -avoid_negative_ts 1 part1.mp4
ffmpeg -ss 03:17 -to 05:26 -i input.mp4 -c copy -avoid_negative_ts 1 part2.mp4
ffmpeg -ss 09:21 -to 09:28 -i input.mp4 -c copy -avoid_negative_ts 1 part3.mp4
ffmpeg -ss 09:44 -to 10:04 -i input.mp4 -c copy -avoid_negative_ts 1 part4.mp4

printf "file 'part1.mp4'\nfile 'part2.mp4'\nfile 'part3.mp4'\nfile 'part4.mp4'\n" > list.txt
ffmpeg -f concat -safe 0 -i list.txt -c copy output.mp4

If you tell me whether you want exact cuts (Option A) or fast/no re-encode (Option B), I can tailor the command for your codec/container constraints.

User: actually i want them separate in order to check them

actually i want them separate in order to check them also, i want to be able to cut up to miliseconds, if needed

Assistant: Got it — make separate clips, and allow millisecond precision.

Got it — make separate clips, and allow millisecond precision.

Use -ss/-to with input seeking (accurate) and re-encode. Put -ss after -i.

=====

ffmpeg -i input.mp4 -ss 00:01:02.000 -to 00:02:13.000 \ =====
-c:v libx264 -crf 20 -preset medium -c:a aac -b:a 160k \
-movflags +faststart clip_01_0102-0213.mp4

=====

ffmpeg -i input.mp4 -ss 00:03:17.000 -to 00:05:26.000 \ =====
-c:v libx264 -crf 20 -preset medium -c:a aac -b:a 160k \
-movflags +faststart clip_02_0317-0526.mp4

=====

ffmpeg -i input.mp4 -ss 00:09:21.000 -to 00:09:28.000 \ =====
-c:v libx264 -crf 20 -preset medium -c:a aac -b:a 160k \
-movflags +faststart clip_03_0921-0928.mp4

=====

ffmpeg -i input.mp4 -ss 00:09:44.000 -to 00:10:04.000 \ =====
-c:v libx264 -crf 20 -preset medium -c:a aac -b:a 160k \
-movflags +faststart clip_04_0944-1004.mp4

Millisecond edits: just change .000 to whatever you need, e.g. 00:03:17.237.

If you must keep original video quality, the only way to be frame/ms-accurate without re-encoding is basically impossible — stream copy can only cut cleanly on keyframes and audio packet boundaries.

However, you can do “almost-original” by re-encoding only around cut points, but that’s more complex and usually not worth it unless quality is critical.

With -ss after -i (like above), -to is treated as an end timestamp in the input timeline (works well for human timestamps).

If you ever prefer “duration”, use -t instead:

# from 01:02 for 71 seconds
ffmpeg -i input.mp4 -ss 00:01:02.000 -t 00:01:11.000 ...

====

in="input.mp4" ====

ffmpeg -i "$in" -ss 00:01:02.000 -to 00:02:13.000 -c:v libx264 -crf 20 -preset medium -c:a aac -b:a 160k -movflags +faststart clip_01.mp4
ffmpeg -i "$in" -ss 00:03:17.000 -to 00:05:26.000 -c:v libx264 -crf 20 -preset medium -c:a aac -b:a 160k -movflags +faststart clip_02.mp4
ffmpeg -i "$in" -ss 00:09:21.000 -to 00:09:28.000 -c:v libx264 -crf 20 -preset medium -c:a aac -b:a 160k -movflags +faststart clip_03.mp4
ffmpeg -i "$in" -ss 00:09:44.000 -to 00:10:04.000 -c:v libx264 -crf 20 -preset medium -c:a aac -b:a 160k -movflags +faststart clip_04.mp4

If you tell me your input codecs (paste ffprobe -hide_banner -i input.mp4), I can adjust settings to either keep the same FPS/resolution, preserve 5.1 audio, or speed it up with hardware encoding (NVENC/QSV/VAAPI).

User: why are "-c:v libx264 -crf 20 -preset medium -c:a aac -b:a 160k -movflags +faststart" needed?

why are "-c:v libx264 -crf 20 -preset medium -c:a aac -b:a 160k -movflags +faststart" needed?

Assistant: They’re not required to cut — they’re just the “safe defaults” that make the output widely playable and keep the cut ac…

They’re not required to cut — they’re just the “safe defaults” that make the output widely playable and keep the cut accurate.

Here’s what each one does, and what happens if you remove it:

Chooses the video encoder (H.264).

• If you don’t specify -c:v, ffmpeg will pick a default encoder for the output container (often still H.264, but not guaranteed).
• You could also use -c:v copy to avoid re-encoding, but then your cut won’t be millisecond-accurate (it’ll snap to keyframes / packet boundaries).

Quality target for x264 (“constant quality” mode). Lower = higher quality/bigger file; higher = lower quality/smaller file.

• Not required; if omitted, x264 uses its default CRF (commonly 23), which is a bit lower quality.
• Typical range: 18–23.

Encoding speed vs compression efficiency.

• Not required; x264 defaults to medium anyway.
• Slower presets = smaller file for same quality (but take longer).

Re-encodes audio to AAC at 160 kbps.

• Not required; ffmpeg might pick AAC anyway, but again not guaranteed.
• If your source audio is already AAC and you don’t need frame-accurate audio boundaries, you might do -c:a copy. But for “exact cuts”, re-encoding audio avoids weird edge cases (missing samples, bad timestamps, concat issues later).

Moves the MP4 “moov atom” to the beginning so the file can start playing before fully downloaded (web-friendly).

• Not required for local playback.
• It doesn’t change quality, only file layout.

You said: separate clips + millisecond precision if needed.

To cut with ms precision reliably, you generally need to decode and re-encode (at least video), because -c copy can only cut cleanly on keyframes and packet boundaries.

So the shortest “ms-accurate” version is basically:

ffmpeg -i input.mp4 -ss 00:01:02.000 -to 00:02:13.000 -c:v libx264 -c:a aac clip_01.mp4

Everything else (-crf, -preset, -b:a, +faststart) is tuning / compatibility / convenience.

This is still ms-accurate for video (because it re-encodes video), and keeps audio untouched:

ffmpeg -i input.mp4 -ss 00:01:02.000 -to 00:02:13.000 -c:v libx264 -crf 20 -preset medium -c:a copy clip_01.mp4

Caveat: if you later concatenate clips, audio-copy can sometimes make concat pickier depending on timestamps/codec parameters.

====

ffmpeg -ss 00:01:02.000 -to 00:02:13.000 -i input.mp4 -c copy clip_01.mp4 ====

Fast, no quality loss — but cuts may be off by up to a few seconds if keyframes are sparse.

If you tell me whether your priority is (1) perfect accuracy or (2) no re-encode / perfect quality, I’ll give you the best exact command for that.

User: i want to both preserve quality and be exact

i want to both preserve quality and be exact

Assistant: You can’t have both in the strict, literal sense unless your cut points land exactly on keyframes and audio packet boun…

You can’t have both in the strict, literal sense unless your cut points land exactly on keyframes (and audio packet boundaries). Reason: with -c copy ffmpeg is not allowed to invent new frames, so it can only start a clip on an existing keyframe; “millisecond exact” usually requires decoding frames and re-encoding the first GOP.

So you have two practical ways to get what you mean by “preserve quality”:

This is the only “no quality change” method:

ffmpeg -ss 00:01:02.000 -to 00:02:13.000 -i input.mp4 -c copy clip_01.mp4

But it will only be exact to the ms if there’s a keyframe exactly at 00:01:02.000 (rare).

You will re-encode, but you can make it lossless (or visually lossless).

Per clip:

ffmpeg -i input.mp4 -ss 00:01:02.000 -to 00:02:13.000 \
-c:v libx264 -qp 0 -preset veryslow \
-c:a copy \
clip_01_lossless.mkv

Notes:

• -qp 0 = lossless for x264.
• I used MKV because it’s more forgiving; MP4 can be pickier with some codec combos.
• -c:a copy keeps audio unchanged.

=====

ffmpeg -i input.mp4 -ss 00:01:02.000 -to 00:02:13.000 \ =====
-c:v libx264 -crf 16 -preset slow \
-c:a copy \
clip_01.mkv

(You can use CRF 14–18 depending on how paranoid you are.)

If you want the final clips to be -c copy and exact, you can first create a version that has keyframes exactly at your cut points:

=====

ffmpeg -i input.mp4 \ =====
-c:v libx264 -qp 0 -preset veryslow \
-force_key_frames "00:01:02.000,00:02:13.000,00:03:17.000,00:05:26.000,00:09:21.000,00:09:28.000,00:09:44.000,00:10:04.000" \
-c:a copy \
keyed.mkv

=====

ffmpeg -ss 00:01:02.000 -to 00:02:13.000 -i keyed.mkv -c copy clip_01.mkv =====
ffmpeg -ss 00:03:17.000 -to 00:05:26.000 -i keyed.mkv -c copy clip_02.mkv
ffmpeg -ss 00:09:21.000 -to 00:09:28.000 -i keyed.mkv -c copy clip_03.mkv
ffmpeg -ss 00:09:44.000 -to 00:10:04.000 -i keyed.mkv -c copy clip_04.mkv

This still “changes” the video once (step 1), but with -qp 0 it’s lossless, so quality is preserved.

If you tell me your source codec/pix_fmt (paste ffprobe -hide_banner -i input.mp4), I can adjust for 10-bit / HDR, H.265, or ensure audio stays compatible with MP4 if you really need MP4 outputs.

User: here is the original video.

here is the original video. change your answer to fit its format/properties (in case any change does any improvement)

Assistant: Your source is:

Your source is:

• Container: MKV
• Video: H.264 High, 1920×1080, 30 fps, yuv420p, BT.709 / TV range
• Audio: Opus, 48 kHz, stereo

So the “best fit” output (same kind of file) is: MKV + H.264 + Opus, keeping 1080p/30fps/pix_fmt and BT.709 metadata.

Because you want exact cuts and “preserve quality”, the most faithful option is lossless H.264 re-encode (x264 -qp 0). That guarantees exact timestamps (ms) and no generation loss (but files can be bigger).

###

ffmpeg -i "NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv" -filter_complex \
"[0:v]trim=start=62:end=133,setpts=PTS-STARTPTS[v];[0:a]atrim=start=62:end=133,asetpts=PTS-STARTPTS[a]" \
-map "[v]" -map "[a]" \
-c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p -profile:v high -level 4.0 \
-color_range tv -colorspace bt709 -color_trc bt709 -color_primaries bt709 \
-c:a libopus -b:a 160k -ar 48000 -ac 2 \
clip_01_0102-0213.mkv

=====

ffmpeg -i "NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv" -filter_complex \ =====
"[0:v]trim=start=197:end=326,setpts=PTS-STARTPTS[v];[0:a]atrim=start=197:end=326,asetpts=PTS-STARTPTS[a]" \
-map "[v]" -map "[a]" \
-c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p -profile:v high -level 4.0 \
-color_range tv -colorspace bt709 -color_trc bt709 -color_primaries bt709 \
-c:a libopus -b:a 160k -ar 48000 -ac 2 \
clip_02_0317-0526.mkv

=====

ffmpeg -i "NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv" -filter_complex \ =====
"[0:v]trim=start=561:end=568,setpts=PTS-STARTPTS[v];[0:a]atrim=start=561:end=568,asetpts=PTS-STARTPTS[a]" \
-map "[v]" -map "[a]" \
-c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p -profile:v high -level 4.0 \
-color_range tv -colorspace bt709 -color_trc bt709 -color_primaries bt709 \
-c:a libopus -b:a 160k -ar 48000 -ac 2 \
clip_03_0921-0928.mkv

=====

ffmpeg -i "NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv" -filter_complex \ =====
"[0:v]trim=start=584:end=604,setpts=PTS-STARTPTS[v];[0:a]atrim=start=584:end=604,asetpts=PTS-STARTPTS[a]" \
-map "[v]" -map "[a]" \
-c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p -profile:v high -level 4.0 \
-color_range tv -colorspace bt709 -color_trc bt709 -color_primaries bt709 \
-c:a libopus -b:a 160k -ar 48000 -ac 2 \
clip_04_0944-1004.mkv

Just use decimals in the start= / end= values, e.g.:

• trim=start=197.237:end=326.041
• atrim=start=197.237:end=326.041

Same commands, just replace -qp 0 -preset veryslow with:

• Very high quality: -crf 16 -preset slow
• Still excellent: -crf 18 -preset slow

Example swap:

-c:v libx264 -crf 16 -preset slow

With MKV + Opus (and your file starting at -0.007s), trim/atrim + setpts/asetpts reliably:

• cuts exactly (including ms),
• resets each clip to start at t=0 cleanly,
• avoids weird negative start timestamps.

If you want, tell me one clip where you need ms precision (start/end), and I’ll plug those exact values into the command for you.

User: note that i still might need to cut up to miliseconds

note that i still might need to cut up to miliseconds

Assistant: Yep — the approach I gave still works fine with milliseconds.

Yep — the approach I gave still works fine with milliseconds. You just put decimals on the cut points.

You can do it either as seconds with decimals:

• start=62.137:end=133.904

or as HH:MM:SS.mmm (ffmpeg accepts this format too, but inside trim it’s simplest to use seconds).

Say you decide clip 2 should be 03:17.237 → 05:26.041.

03:17.237 = 197.237 seconds 05:26.041 = 326.041 seconds

ffmpeg -i "NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv" -filter_complex \
"[0:v]trim=start=197.237:end=326.041,setpts=PTS-STARTPTS[v]; \
 [0:a]atrim=start=197.237:end=326.041,asetpts=PTS-STARTPTS[a]" \
-map "[v]" -map "[a]" \
-c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p -profile:v high -level 4.0 \
-color_range tv -colorspace bt709 -color_trc bt709 -color_primaries bt709 \
-c:a libopus -b:a 160k -ar 48000 -ac 2 \
clip_02_0317p237-0526p041.mkv

Your video is 30 fps, so frames are spaced about 33.333 ms apart.

That means:

• You can type millisecond timestamps,
• but the actual video cut can only land on a frame boundary (≈33ms steps).
• Audio can be trimmed much more finely (sample-level).

If you want, paste the exact ms cut points you end up with (like 01:02.153–02:13.087) and I’ll rewrite all 4 commands with those exact values.

User: first one is 01:02.2 till 02:13.2

first one is 01:02.2 till 02:13.2

Assistant: Cool — for clip 1 you want 00:01:02.200 → 00:02:13.200 i.e.

Cool — for clip 1 you want 00:01:02.200 → 00:02:13.200 (i.e. 62.2s → 133.2s).

=====

ffmpeg -i "NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv" -filter_complex \ =====
"[0:v]trim=start=62.2:end=133.2,setpts=PTS-STARTPTS[v]; \
 [0:a]atrim=start=62.2:end=133.2,asetpts=PTS-STARTPTS[a]" \
-map "[v]" -map "[a]" \
-c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p -profile:v high -level 4.0 \
-color_range tv -colorspace bt709 -color_trc bt709 -color_primaries bt709 \
-c:a libopus -b:a 160k -ar 48000 -ac 2 \
clip_01_0102p2-0213p2.mkv

Quick note: your video is 30 fps, so the video cut will land on the nearest frame boundary (~33.3 ms steps). Audio can still be trimmed more finely.

User: $ ffmpeg -i "NEA MADALIN E DRAG QUEEN ACUM P9gZEDOz9eE.mkv" -filtercomplex \

$ ffmpeg -i "NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv" -filter_complex \ "[0:v]trim=start=62.2:end=133.2,setpts=PTS-STARTPTS[v]; \

[0:a]atrim=start=62.2:end=133.2,asetpts=PTS-STARTPTS[a]" \

-map "[v]" -map "[a]" \ -c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p -profile:v high -level 4.0 \ -color_range tv -colorspace bt709 -color_trc bt709 -color_primaries bt709 \ -c:a libopus -b:a 160k -ar 48000 -ac 2 \ clip_01_0102p2-0213p2.mkv ffmpeg version 4.4.2-0ubuntu0.22.04.1 Copyright (c) 2000-2021 the FFmpeg developers

 built with gcc 11 (Ubuntu 11.2.0-19ubuntu1)
 configuration: --prefix=/usr --extra-version=0ubuntu0.22.04.1 --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --arch=amd64 --enable-gpl --disable-stripping --enable-gnutls --enable-ladspa --enable-libaom --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libcodec2 --enable-libdav1d --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libjack --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librabbitmq --enable-librubberband --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvidstab --enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzimg --enable-libzmq --enable-libzvbi --enable-lv2 --enable-omx --enable-openal --enable-opencl --enable-opengl --enable-sdl2 --enable-pocketsphinx --enable-librsvg --enable-libmfx --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-chromaprint --enable-frei0r --enable-libx264 --enable-shared
 WARNING: library configuration mismatch
 avcodec     configuration: --prefix=/usr --extra-version=0ubuntu0.22.04.1 --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --arch=amd64 --enable-gpl --disable-stripping --enable-gnutls --enable-ladspa --enable-libaom --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libcodec2 --enable-libdav1d --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libjack --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librabbitmq --enable-librubberband --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvidstab --enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzimg --enable-libzmq --enable-libzvbi --enable-lv2 --enable-omx --enable-openal --enable-opencl --enable-opengl --enable-sdl2 --enable-pocketsphinx --enable-librsvg --enable-libmfx --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-chromaprint --enable-frei0r --enable-libx264 --enable-shared --enable-version3 --disable-doc --disable-programs --enable-libaribb24 --enable-libopencore_amrnb --enable-libopencore_amrwb --enable-libtesseract --enable-libvo_amrwbenc --enable-libsmbclient
 libavutil      56. 70.100 / 56. 70.100
 libavcodec     58.134.100 / 58.134.100
 libavformat    58. 76.100 / 58. 76.100
 libavdevice    58. 13.100 / 58. 13.100
 libavfilter     7.110.100 /  7.110.100
 libswscale      5.  9.100 /  5.  9.100
 libswresample   3.  9.100 /  3.  9.100
 libpostproc    55.  9.100 / 55.  9.100

Input #0, matroska,webm, from 'NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv':

 Metadata:
   COMPATIBLE_BRANDS: iso6avc1mp41
   MAJOR_BRAND     : dash
   MINOR_VERSION   : 0
   ENCODER         : Lavf58.76.100
 Duration: 00:25:49.28, start: -0.007000, bitrate: 2264 kb/s
 Stream #0:0: Video: h264 (High), yuv420p(tv, bt709, progressive), 1920x1080 [SAR 1:1 DAR 16:9], 30 fps, 30 tbr, 1k tbn, 60 tbc (default)
   Metadata:
     HANDLER_NAME    : ISO Media file produced by Google Inc.
     VENDOR_ID       : [0][0][0][0]
     DURATION        : 00:25:49.266000000
 Stream #0:1(eng): Audio: opus, 48000 Hz, stereo, fltp (default)
   Metadata:
     DURATION        : 00:25:49.281000000

Stream mapping:

 Stream #0:0 (h264) -> trim
 Stream #0:1 (opus) -> atrim
 setpts -> Stream #0:0 (libx264)
 asetpts -> Stream #0:1 (libopus)

Press [q] to stop, [?] for help x264 [error]: high profile doesn't support lossless [libx264 @ 0x5874e72a1640] Error setting profile high. [libx264 @ 0x5874e72a1640] Possible profiles: baseline main high high10 high422 high444 Error initializing output stream 0:0 -- Error while opening encoder for output stream #0:0 - maybe incorrect parameters such as bit_rate, rate, width or height [libopus @ 0x5874e739cb80] 1 frames left in the queue on closing Conversion failed!

Assistant: That error is because of this combo:

That error is because of this combo:

• -qp 0 = lossless
• -profile:v high = forces “High” profile

x264 refuses that: “high profile doesn't support lossless”. For lossless, x264 switches to a different profile (typically High 4:4:4 Predictive) automatically. When you force high, it can’t.

Use this:

ffmpeg -i "NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv" -filter_complex \
"[0:v]trim=start=62.2:end=133.2,setpts=PTS-STARTPTS[v]; \
 [0:a]atrim=start=62.2:end=133.2,asetpts=PTS-STARTPTS[a]" \
-map "[v]" -map "[a]" \
-c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p \
-c:a libopus -b:a 160k -ar 48000 -ac 2 \
clip_01_0102p2-0213p2.mkv

That will cut exactly (including milliseconds) and keep video lossless.

Even though your pixel format stays yuv420p, a lossless x264 encode often signals a profile like High 4:4:4 Predictive. Some hardware players (TVs/phones) may not like that, even if VLC/mpv do.

If you want maximum compatibility while still being visually indistinguishable from the source, do this instead:

ffmpeg -i "NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv" -filter_complex \
"[0:v]trim=start=62.2:end=133.2,setpts=PTS-STARTPTS[v]; \
 [0:a]atrim=start=62.2:end=133.2,asetpts=PTS-STARTPTS[a]" \
-map "[v]" -map "[a]" \
-c:v libx264 -crf 16 -preset slow -pix_fmt yuv420p -profile:v high -level 4.0 \
-c:a libopus -b:a 160k -ar 48000 -ac 2 \
clip_01_0102p2-0213p2.mkv

Just change 62.2 / 133.2 to e.g. 62.237 / 133.041.

If you send the ms-precise times for the other 3 clips, I’ll spit out all 4 final commands in the same style.

User: second is 03:17.10 to 05:26.20

second is 03:17.10 to 05:26.20

Assistant: Great — that’s:

Great — that’s:

• start 00:03:17.100 = 197.1 seconds
• end 00:05:26.200 = 326.2 seconds

Here’s the matching command (same settings as the fixed lossless version — no forced -profile:v high):

ffmpeg -i "NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv" -filter_complex \
"[0:v]trim=start=197.1:end=326.2,setpts=PTS-STARTPTS[v]; \
 [0:a]atrim=start=197.1:end=326.2,asetpts=PTS-STARTPTS[a]" \
-map "[v]" -map "[a]" \
-c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p \
-c:a libopus -b:a 160k -ar 48000 -ac 2 \
clip_02_0317p10-0526p20.mkv

If you want the filename to reflect milliseconds more clearly, you can rename it after (or I can format it differently).

Send clip 3 + 4 ms times when ready.

User: 09:21:00 till 09:28:30

09:21:00 till 09:28:30

Assistant: That’s:

That’s:

• start 00:09:21.000 = 561.0 seconds
• end 00:09:28.300 = 568.3 seconds

ffmpeg -i "NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv" -filter_complex \
"[0:v]trim=start=561.0:end=568.3,setpts=PTS-STARTPTS[v]; \
 [0:a]atrim=start=561.0:end=568.3,asetpts=PTS-STARTPTS[a]" \
-map "[v]" -map "[a]" \
-c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p \
-c:a libopus -b:a 160k -ar 48000 -ac 2 \
clip_03_0921p00-0928p30.mkv

Send the last interval (09:44–10:04 with ms) and I’ll format that one too.

User: 09:44:10 to 10:04:20

09:44:10 to 10:04:20

Assistant: That’s:

That’s:

• start 00:09:44.100 = 584.1 seconds
• end 00:10:04.200 = 604.2 seconds

ffmpeg -i "NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv" -filter_complex \
"[0:v]trim=start=584.1:end=604.2,setpts=PTS-STARTPTS[v]; \
 [0:a]atrim=start=584.1:end=604.2,asetpts=PTS-STARTPTS[a]" \
-map "[v]" -map "[a]" \
-c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p \
-c:a libopus -b:a 160k -ar 48000 -ac 2 \
clip_04_0944p10-1004p20.mkv

If you want, here are all four commands together (same settings) so you can paste once:

in="NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv"

ffmpeg -i "$in" -filter_complex "[0:v]trim=start=62.2:end=133.2,setpts=PTS-STARTPTS[v];[0:a]atrim=start=62.2:end=133.2,asetpts=PTS-STARTPTS[a]"  -map "[v]" -map "[a]" -c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p -c:a libopus -b:a 160k -ar 48000 -ac 2 clip_01_0102p20-0213p20.mkv
ffmpeg -i "$in" -filter_complex "[0:v]trim=start=197.1:end=326.2,setpts=PTS-STARTPTS[v];[0:a]atrim=start=197.1:end=326.2,asetpts=PTS-STARTPTS[a]" -map "[v]" -map "[a]" -c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p -c:a libopus -b:a 160k -ar 48000 -ac 2 clip_02_0317p10-0526p20.mkv
ffmpeg -i "$in" -filter_complex "[0:v]trim=start=561.0:end=568.3,setpts=PTS-STARTPTS[v];[0:a]atrim=start=561.0:end=568.3,asetpts=PTS-STARTPTS[a]" -map "[v]" -map "[a]" -c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p -c:a libopus -b:a 160k -ar 48000 -ac 2 clip_03_0921p00-0928p30.mkv
ffmpeg -i "$in" -filter_complex "[0:v]trim=start=584.1:end=604.2,setpts=PTS-STARTPTS[v];[0:a]atrim=start=584.1:end=604.2,asetpts=PTS-STARTPTS[a]" -map "[v]" -map "[a]" -c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p -c:a libopus -b:a 160k -ar 48000 -ac 2 clip_04_0944p10-1004p20.mkv

User: ok, now in a single ffmpeg command create the combined video of the cuts

ok, now in a single ffmpeg command create the combined video of the cuts

Assistant: Here’s a single ffmpeg command that cuts your 4 timestamp ranges with milliseconds and concats them into one output.

Here’s a single ffmpeg command that cuts your 4 timestamp ranges (with milliseconds) and concats them into one output.

Combined output (exact cuts, lossless video, MKV + H.264 + Opus)

ffmpeg -i "NEA MADALIN E DRAG QUEEN ACUM [P9gZEDOz9eE].mkv" -filter_complex \
"[0:v]trim=start=62.2:end=133.2,setpts=PTS-STARTPTS[v0]; \
 [0:a]atrim=start=62.2:end=133.2,asetpts=PTS-STARTPTS[a0]; \
 [0:v]trim=start=197.1:end=326.2,setpts=PTS-STARTPTS[v1]; \
 [0:a]atrim=start=197.1:end=326.2,asetpts=PTS-STARTPTS[a1]; \
 [0:v]trim=start=561.0:end=568.3,setpts=PTS-STARTPTS[v2]; \
 [0:a]atrim=start=561.0:end=568.3,asetpts=PTS-STARTPTS[a2]; \
 [0:v]trim=start=584.1:end=604.2,setpts=PTS-STARTPTS[v3]; \
 [0:a]atrim=start=584.1:end=604.2,asetpts=PTS-STARTPTS[a3]; \
 [v0][a0][v1][a1][v2][a2][v3][a3]concat=n=4:v=1:a=1[v][a]" \
-map "[v]" -map "[a]" \
-c:v libx264 -preset veryslow -qp 0 -pix_fmt yuv420p \
-c:a libopus -b:a 160k -ar 48000 -ac 2 \
combined_cuts.mkv

If you decide “lossless” is overkill and you want maximum playback compatibility (High profile) while still looking identical to most eyes, swap the video settings to:

-c:v libx264 -crf 16 -preset slow -pix_fmt yuv420p -profile:v high -level 4.0

(Everything else can stay the same.)

User: how much will be duration of the new video?

how much will be duration of the new video?

Assistant: Let’s add up each kept segment:

Let’s add up each kept segment:

1. 01:02.200 → 02:13.200 133.2 − 62.2 = 71.0 s
2. 03:17.100 → 05:26.200 326.2 − 197.1 = 129.1 s
3. 09:21.000 → 09:28.300 568.3 − 561.0 = 7.3 s
4. 09:44.100 → 10:04.200 604.2 − 584.1 = 20.1 s

Total = 71.0 + 129.1 + 7.3 + 20.1 = 227.5 seconds = 3 minutes 47.5 seconds (i.e., 00:03:47.500)

User: do cummulative sum and also convert into minutes so i can see where the cuts are of the total

do cummulative sum (and also convert into minutes) so i can see where the cuts are of the total

Assistant: Sure — here are the segment lengths and the cumulative timeline in the new combined video.

Sure — here are the segment lengths and the cumulative timeline in the new combined video.

Segment durations

• S1 (01:02.200 → 02:13.200): 71.000 s = 1.1833 min = 01:11.000
• S2 (03:17.100 → 05:26.200): 129.100 s = 2.1517 min = 02:09.100
• S3 (09:21.000 → 09:28.300): 7.300 s = 0.1217 min = 00:07.300
• S4 (09:44.100 → 10:04.200): 20.100 s = 0.3350 min = 00:20.100

Cumulative positions in the combined output

| Segment | In source | Duration | Starts at (new) | Ends at (new) | End (minutes) | | ------- | ------------------- | --------- | --------------- | ------------- | ------------- | | S1 | 01:02.200–02:13.200 | 71.000 s | 00:00.000 | 01:11.000 | 1.1833 min | | S2 | 03:17.100–05:26.200 | 129.100 s | 01:11.000 | 03:20.100 | 3.3350 min | | S3 | 09:21.000–09:28.300 | 7.300 s | 03:20.100 | 03:27.400 | 3.4567 min | | S4 | 09:44.100–10:04.200 | 20.100 s | 03:27.400 | 03:47.500 | 3.7917 min |

Total combined duration: 227.500 s = 3.7917 min = 03:47.500

User: make the table ascii only

make the table ascii only

Assistant:

+---------+-------------------------+-----------+----------------+----------------+-------------+
| Segment | In source                | Duration  | Starts at (new) | Ends at (new)   | End (min)   |
+---------+-------------------------+-----------+----------------+----------------+-------------+
| S1      | 01:02.200 - 02:13.200   | 01:11.000 | 00:00.000      | 01:11.000      | 1.1833 min  |
| S2      | 03:17.100 - 05:26.200   | 02:09.100 | 01:11.000      | 03:20.100      | 3.3350 min  |
| S3      | 09:21.000 - 09:28.300   | 00:07.300 | 03:20.100      | 03:27.400      | 3.4567 min  |
| S4      | 09:44.100 - 10:04.200   | 00:20.100 | 03:27.400      | 03:47.500      | 3.7917 min  |
+---------+-------------------------+-----------+----------------+----------------+-------------+
| TOTAL   |                         | 03:47.500 |                |                | 3.7917 min  |
+---------+-------------------------+-----------+----------------+----------------+-------------+