yt-dlp Command Generator for Bilibili

Build a yt-dlp command for Bilibili, starting with the browser impersonation that avoids most 403 errors.

Video, playlist or channel URL

Optional. Leave it empty and the command ends with URL for you to fill in.

Your command

yt-dlp -f 'bv*[height<=1080]+ba/b[height<=1080]' --merge-output-format mp4 --no-playlist --impersonate chrome --cookies-from-browser chrome https://www.bilibili.com/video/BV1xx411c7mD

Basics

What do you want

Video Audio only

Audio-only extracts the soundtrack and drops the video stream.

Video and quality

Maximum quality -f

Caps the height. yt-dlp still picks the best stream at or below it.

Container --merge-output-format needs ffmpeg

The file type separate video and audio streams are merged into.

Advanced options (3)

Custom format selector -f

Overrides the quality dropdown. For people who already know the format syntax.

Remux into --remux-video needs ffmpeg

Repackages the streams without re-encoding. Fast and lossless.

Re-encode into --recode-video needs ffmpeg

Re-encodes the video. Slow and lossy — only when remuxing will not do.

Audio

Audio format --audio-format needs ffmpeg

The file the audio is converted to.

Audio quality --audio-quality needs ffmpeg

VBR gives the best size-to-quality ratio; fixed bitrates are more predictable.

Subtitles

Embedding puts them inside the video; saving writes a separate .srt next to it.

Subtitle languages --sub-langs

Comma separated. Use all for every available language.

Auto-generated subtitles --write-auto-subs Include machine-generated captions when no human ones exist.

Playlists and channels

Playlist handling

What to do when the URL belongs to both a video and a playlist.

Which items -I

Ranges and lists both work, for example 1-10,15 or -5::2 for every second item from the end.

Advanced options (1)

Archive file --download-archive

Records what has been downloaded so re-running skips it. The way to keep a channel in sync.

Sections and chapters

Advanced options (2)

Download only part of it --download-sections needs ffmpeg

A time range like *00:01:30-00:03:00, or a chapter name. The leading asterisk means it is a timestamp.

Split into chapters --split-chapters needs ffmpeg Writes one file per chapter instead of a single long one.

Metadata and artwork

Embed thumbnail --embed-thumbnail needs ffmpeg Uses the video thumbnail as cover art.

Embed metadata --embed-metadata needs ffmpeg Writes title, uploader and date into the file itself.

Embed chapters --embed-chapters needs ffmpeg Adds chapter markers your player can jump between.

Advanced options (3)

Save thumbnail file --write-thumbnail Writes the cover image next to the video.

Save description --write-description Writes the description to a .description file.

Save metadata JSON --write-info-json Writes everything yt-dlp knows about the video to a .info.json file.

SponsorBlock

Remove segments --sponsorblock-remove needs ffmpeg

Cuts community-marked segments out of the file using SponsorBlock data.

Advanced options (1)

Mark segments as chapters --sponsorblock-mark

Keeps the segments but marks them, so you can skip them yourself.

Files and folders

File name template -o

Field names go in %(...)s. Slashes create folders.

Safe (ASCII) file names --restrict-filenames Strips accents, spaces and anything a stubborn filesystem might reject.

Overwrite existing files --force-overwrites Replaces files that are already there.

Resume partial downloads -c Picks up where an interrupted download stopped.

Advanced options (4)

Download folder -P

Where files are written. Combine with the template above for subfolders.

Windows-safe names --windows-filenames Removes characters Windows refuses even when running elsewhere.

Trim names to --trim-filenames

Maximum file name length in characters.

Never overwrite -w Skips anything already downloaded.

Filters

Advanced options (5)

Uploaded after --dateafter

YYYYMMDD, or a relative date like now-1month.

Uploaded before --datebefore

YYYYMMDD, or a relative date.

Minimum size --min-filesize

Skips anything smaller. Accepts 5M, 500K and similar.

Maximum size --max-filesize

Skips anything larger.

Stop after --max-downloads

Number of files to download before stopping.

Network

Speed limit -r

For example 2M or 500K. Useful when you need the rest of your bandwidth.

Advanced options (5)

Parallel fragments -N

Speeds up fragmented streams. Four is a sensible starting point.

Retries -R

How many times to retry a failed piece. infinite is allowed.

Proxy --proxy

HTTP, HTTPS or SOCKS proxy URL.

Bypass geo-restriction --geo-bypass Fakes the request origin. Works on some sites, not all.

Impersonate a browser --impersonate

Makes requests look like a real browser. The usual fix for sudden 403 errors.

Cookies from browser --cookies-from-browser

Reads your logged-in session directly. Needed for private, age-restricted or members-only content.

Advanced options (1)

Cookies file --cookies

A Netscape-format cookies.txt exported from your browser.

Live streams

Advanced options (2)

Record from the start --live-from-start Starts a live stream from the beginning instead of the current moment.

Wait for the stream --wait-for-video

Polls until a scheduled stream begins. MIN-MAX seconds between checks.

Workarounds

Advanced options (3)

User agent --user-agent deprecated

Deprecated in favour of --add-headers. Kept because plenty of older guides still use it.

Referer --referer deprecated

Deprecated in favour of --add-headers.

Extractor arguments --extractor-args

Site-specific switches, for example youtube:player_client=web.

Output and debugging

List formats only -F Prints the available formats and exits without downloading.

Advanced options (2)

Keep going on errors -i Skips items that fail instead of stopping. Essential for large playlists.

Simulate -s Does everything except write the file.

Reset

The options that matter here

Impersonate a browser --impersonate Makes requests look like a real browser. The usual fix for sudden 403 errors.
Cookies from browser --cookies-from-browser Reads your logged-in session directly. Needed for private, age-restricted or members-only content.
Maximum quality -f Caps the height. yt-dlp still picks the best stream at or below it.

403 errors and quality limits

Bilibili rejects requests that do not look like a browser, which is why a plain command often returns 403. Impersonation is switched on in this preset because it is the fix in most cases.

1080p and above are usually reserved for logged-in or premium accounts, so cookies matter here for quality and not just for access. Multi-part videos appear as playlist entries.

Bilibili is the platform where a stale yt-dlp breaks fastest — the site reworks its API more often than any Western platform, and extractor updates land in yt-dlp releases within days of each change. When a download that worked last month starts failing, update first: yt-dlp -U. Impersonation (on by default in this preset) covers the browser-looking-headers part of the 403 problem; the rest is usually a missing or expired session.

Multi-part uploads — the P1/P2/P3 structure common for series — arrive as playlist entries, so whole-playlist handling fetches every part, and the playlist index in the output template keeps the parts in order. Danmaku (the scrolling comments) are a subtitle track to yt-dlp: --write-subs saves them as an .ass file, which is the only format they exist in. If you do not want them, do nothing — they are not merged unless asked.

Interaction qualities — the "dolby" and premium ladders — exist only for accounts holding them, and no flag reaches them without cookies from such an account. Region-locked content is the other wall: some uploads are visible only from mainland China, and a proxy in-region is the only route when the block is enforced at the CDN rather than the page. Geo-bypass occasionally works for the page-level block and never for the CDN one.

Bilibili serves its higher qualities as DASH with separate audio, so ffmpeg merges just like YouTube — the merge step is in this preset already. Audio-only extraction with -x works and is the fastest way to take just the sound from a long multi-part upload. For live replays, the replay becomes an ordinary VOD after the stream ends, so download it like any video; the live-recording options on this site are for the stream while it is happening.

Frequently asked questions

› Why does Bilibili return 403?

Bilibili rejects requests that do not look like a browser. --impersonate chrome fixes this most of the time; add cookies for member-only content.

› How do I download 1080p or higher from Bilibili?

Higher resolutions are reserved for logged-in and sometimes premium accounts. Use --cookies-from-browser with such an account.

› How do I get every part of a multi-part video?

Set playlist handling to the whole playlist — the parts are exposed as playlist entries.