Danmu are viewer comments tied to a precise second in a video. When the scene reaches that second, the comments cross the picture—even if they were written weeks earlier. The result is less like reading a comment section after a show and more like hearing a room react while the show is still playing.
The useful question is not simply “what does danmu mean?” It is what happens when an audience becomes part of the image.
A joke arrives on the frame that makes it funny
An ordinary comment has to describe its subject: “At 03:18, look at the person on the left.” A danmu can simply say “left side” because it appears while that person is visible. Timing carries part of the sentence.
That makes several familiar danmu moves possible:
- a warning arrives one beat before a jump scare;
- the same phrase floods the screen when a recurring character enters;
- viewers type a lyric in time with a song;
- someone adds a name, translation or reference exactly where it is needed;
- a wave of laughter becomes visible before the scene has finished.
The comments do not need to form a thread. Repetition, placement and the video itself supply the connection.
The audience is together, but not necessarily live
Bilibili describes danmu as a way for viewers to cross the limits of time and create the feeling of watching together. That feeling is real at the interface level, but it can be misunderstood.
A comment flying across the screen is not proof that its writer is watching now. The writer may have arrived yesterday; you may arrive next month. Both of you meet at the same playback second. Researchers describe this as a pseudo-synchronous experience: shared timing without a shared clock.
This is why danmu can make an old video feel socially alive. It is also why the screen should not be read as a live opinion poll.
Repetition is part of the language
When hundreds of viewers repeat one short phrase, the phrase stops working like an individual statement. It becomes applause, a chant, a warning sign or an in-joke.
The meaning can depend on where it appears. The same words may celebrate a long-awaited entrance in one video and mock an overused trope in another. Translation software can carry the literal sentence while losing the shared reference.
For a first viewing, watch for three things before trying to translate every line:
- When does the wave begin? The triggering image or sound often explains more than the words.
- Is the phrase copied? A repeated line may be a ritual response rather than a debate.
- Does it point or perform? Some comments identify details; others are there to join the moment.
The layer can improve a video—or bury it
One well-timed note can catch a visual clue. A dense curtain can cover a face, subtitle or important action. Danmu’s strength and failure mode come from the same property: the audience occupies the picture.
That matters for accessibility. Moving text competes with captions and can make a stable image difficult to follow. Bilibili and related players provide visibility, density and filtering controls, but the exact buttons change. Turning danmu off is not “watching it wrong.” It is often the best way to see the work first.
Try two passes on a video with an active comment layer. Watch once with danmu off, then replay a short section with it on. The first pass shows what the creator made. The second shows what the participating audience made from it.
Danmu are not subtitles or live chat
Subtitles belong to the work or its distributor. They carry dialogue, translation or accessibility information. Danmu belong to viewers and can be wrong, ironic, late or deliberately distracting.
Live chat follows a broadcast clock. Danmu follow playback time. When a recorded video starts again, its danmu can reappear at the same frames. Ordinary comments sit outside the picture and usually require their own ordering and reply structure.
That distinction is the reason danmu deserves attention as a cultural form, not just a software feature. It changes where audience speech lives: from below the work to inside its timing.
What to watch next
Open Bilibili’s 2026 Spring Festival Gala danmu invitation for a clear, platform-made example of communal viewing. Then choose a familiar music performance or comedy clip and compare thirty seconds with the layer on and off. A familiar work makes it easier to notice what the audience adds.
Sources and scope
- Bilibili: About us — the platform’s own account of time-synced comments and the shared-viewing effect.
- Bilibili annual filing — the company’s formal definition of bullet chat as a contextual interaction feature.
- Community-Building on Bilibili — independent research on synchronization and community interaction.
- Coherence through sedimentation — research separating the time of the video, the time of posting and the time of later viewing.
This page explains a viewing form through documented mechanics and one concrete platform example. It does not treat one video’s comments as representative of Chinese public opinion.
What changed, 13 September 2026: the former glossary-style page was rebuilt around the experience of watching, with a playable current example and practical ways to read the audience layer.