Bilibili Live Stream UI Prompts

September 15, 2026

Bilibili has the most distinctive live stream layout of any major platform, and image models get it wrong in the same way every time. They draw a side chat column, which is Twitch, not Bilibili. Real Bilibili streams push danmaku straight across the middle of the video, stack gift buttons up the bottom-right corner, and show captain tiers next to every paying viewer's name. Miss those and the screenshot reads as generic, no matter how good the host render is.

Below are 25+ Bilibili live stream UI prompts organised by the five stream formats the site is known for, followed by a cheat sheet that turns each Bilibili interface element into plain English prompt text. For the cross-platform view, start with the AI livestream UI prompt collection; for the Chinese short-video look instead, the Douyin live stream UI prompts page covers that separately.

Why Bilibili UI Screenshots Are Hard to Fake

The first problem is danmaku placement. On Bilibili, comments fly across the video itself, in several horizontal lanes, with slight vertical offsets so they never collide. Models default to a static column on the right because that is what most training data shows. Naming the lanes, the direction and the speed is what changes the result, much the same way the classic livestream interface prompt improves the moment it stops describing a person and starts describing a screen.

The second problem is the paying-viewer furniture. Bilibili is unusually loud about who is supporting the stream: captain tiers sit beside usernames, charging counts appear in the gift area, and Super Chat messages pin themselves to the top of the screen in coloured cards. All of that is visual texture that makes the interface instantly recognisable, and none of it appears unless you ask for it.

The third problem is text. Danmaku are text, usernames are text, and AI models still struggle with dense lines of characters. The trick is to ask for danmaku as small blurred lines rather than readable sentences, then keep the rest of the interface icon-led. If you are generating anime hosts to go with the interface, the character-side advice in the prompt writing and editing guide is worth reading first.

Anime and Game Stream Prompts

This is the format Bilibili grew up on, and the busiest: danmaku across the whole frame, gifts stacked on the right, badges beside names. The clean live stream UI layout prompt and the typographic livestream portrait prompt are useful quieter baselines to build on.

Voice and Music Room Prompts

Voice rooms have no video feed at all, which confuses most models until you describe the empty stage and the avatar seats. This generic livestream broadcast prompt and the dance livestream broadcast prompt show two ways to fill a room without a camera subject.

Study and Chill Stream Prompts

Study streams are the calmest category on the site, and the trick is to keep the interface sparse on purpose. Compare the featured livestream screen prompt and the modern livestream UI prompt for how little chrome a screen actually needs.

Tech and Programming Stream Prompts

Tech streams are the least decorated format, which makes the small interface details do all the work. The tech livestream interface prompt and the icon-driven interface prompt are good references for panels and readouts.

Festival and Special Event Prompts

Event frames are the one place where maximum decoration is correct, so let the danmaku lanes fill and the props crowd the edges. This historic broadcast interface prompt is a useful contrast for how much layout has changed over the years.

Bilibili Live Stream UI Elements Cheat Sheet

Name two or three of these in a prompt and the frame usually stops looking generic. Two or three is enough; all of them at once turns the image into mush.

How to Write Prompts for Bilibili Screenshots

Write the danmaku layer first, because it is the single element that makes the screenshot unmistakably Bilibili. Specify lanes, direction and opacity, then move on. Second, pick the payment furniture your stream format would actually show: a study stream with no gifts at all, a music stream with Super Chat cards, a gaming stream with captain badges and a charging leaderboard. Choosing the wrong props is why some AI Bilibili screenshots feel like cosplay.

Third, decide portrait or desktop and say it out loud. Mobile screenshots compress everything into a 9:16 frame with the gift column hugging the right edge, while desktop player frames want 16:9 with a control bar along the bottom. Fourth, describe the liveliness of the screen: danmaku mid-flight, a gift animation crossing the frame, a Super Chat card mid-fade. A quiet Bilibili screen does not look like Bilibili.

Finally, keep the number of readable characters near zero. Ask for danmaku as small, slightly blurred text lines and let the badges and icons carry meaning. The interface prompt library and the livestream prompt library both have reference frames you can compare your output against before you commit to a final render.

FAQ

What makes an AI screenshot look like Bilibili instead of Twitch?

Danmaku across the middle of the video rather than in a right-hand column, plus a bottom-right gift column and a popularity gauge instead of a plain viewer count. Add a captain badge to two usernames and the impression is complete. Those three changes do more than any amount of host styling.

How do I prompt for danmaku without getting garbled text?

Ask for small semi-transparent comment lines in horizontal lanes and accept that the characters will be mostly unreadable. At screenshot scale, that reads as authentic. If your model insists on rendering crisp text, add "blurred motion on the text lines" and keep the rest of the interface icon-led.

Should a Bilibili screenshot be portrait or landscape?

Both exist, and they need different prompts. Portrait 9:16 for the mobile app, with a gift column along the right edge and a compact top bar. Landscape 16:9 for the desktop player, with a bottom control bar and wider danmaku lanes. Mixing the two is what produces oddly empty-looking frames.

Can I generate readable Super Chat cards?

Short usernames and a two or three digit amount render fine. Long messages do not, so keep Super Chat text to a few characters or leave it as a soft coloured block. The colour and stacking matter far more to recognition than the wording does.

Browse AI image prompts by model