FLUX 3 Video is here. Serious, fun, creative, real, cinematic, whatever you need it to be. Native audio, Text to Video, Image to Video with multiple frames, video continuation, dialogue in multiple languages. Comes with Draft mode so you can explore ideas fast at a fraction of Show more
視頻模型
FLUX 3 文生 / 圖生 / 首尾幀 / 延長視頻生成器
喺 Dreamega AI 用 Black Forest Labs 嘅多模態前沿模型 FLUX 3 生成視頻。寫提示詞做文生視頻，令參考圖郁起嚟做圖生視頻，用首幀同尾幀駁埋做過渡，或者用視頻延長續寫已有嘅 MP4 片段。每種模式都可以生成 5 至 20 秒、720p 或 1080p 視頻，自帶同步原生音頻同多語言對白。
Flux 模型系列
FLUX 3 視頻作品集
瀏覽用 FLUX 3 由提示詞生成嘅片段。每個示例都係單次生成，自帶同步音頻。
AI Video
遲到五十年嘅外賣員
雷雨夜，一個踩單車送外賣嘅速遞員嚟到公寓門口，撞見五十年後嘅自己，由文字生成。
Prompt
“A young bicycle courier arrives at an apartment door during a thunderstorm, but the digital display shows a date fifty years in the future. The scene opens with a fast tracking shot following him through wet neon streets, then becomes completely still as the door opens. An elderly man with the same face stands inside and looks at him in shock. The camera slowly rotates around both characters while lightning illuminates the corridor. The food bag begins glowing, and nearby objects float into the air. Emotional science-fiction mystery, realistic acting, controlled camera movement, cinematic neon lighting.”
Live PipelineTake 01 / 02
FLUX 3 圖生視頻作品集
睇下 FLUX 3 點樣令靜態圖片郁起嚟。第一段片段由一張參考圖生成；第二段就喺首幀同尾幀之間，跟提示詞完成過渡。
Source Feeds01 Inputs
Pipeline
00%
Program · On AirAI · Generated
Output
Transcript · 01
廢棄月台上嘅機械人
The old robot sits alone on an abandoned train platform, holding a faded ticket. Its head slowly rises as distant lights appear along the tracks. The station signs begin illuminating one by one, and a futuristic train silently enters the platform. The camera begins with a close-up of the ticket, tracks across the robot's weathered body, then cranes upward as the train arrives. Emotional science-fiction, quiet storytelling, atmospheric fog, warm and cold light contrast.
Live PipelineTake 01 / 01
FLUX 3 視頻延長作品集
將已有片段接落去變成新鏡頭。FLUX 3 由源視頻結尾繼續，保持角色、光線同動作連貫。
Source Feeds01 Inputs
Pipeline
00%
Program · On AirAI · Generated
Output
Transcript · 01
博物館骨架活返
Continue directly as the dinosaur skeleton begins moving. It carefully steps down from the exhibition platform, sending small bones and dust across the floor. The paleontologist runs through the gallery while other fossil displays begin moving behind glass cases. The camera tracks beside her, then swings around as she reaches the museum entrance and finds a living prehistoric jungle outside instead of the city. Maintain the same character, clothing, museum layout, lighting, and movement continuity.
FLUX 3 YouTube 視頻
睇社區示範、評測同教學，了解 FLUX 3 嘅實際表現
- FLUX 3 Is Here — Everything You NEED to Know - ElevenLabs
- Introducing FLUX 3 - Black Forest Labs
- Black Forest Labs FLUX 3: First Look - Bryan Software
- 100+ FLUX 3 AI Videos Scarily Close to Real Life - Airt
- Flux 3 Review: Everything You Need to Know (2026) - Thrive Media
- FLUX 3 - Incredible AI Video Model - Demo Reel - Jerrod Lew
FLUX 3 YouTube 視頻
睇社區示範、評測同教學，了解 FLUX 3 嘅實際表現
FLUX 3 在 X 上的熱門評價
睇下大家喺 X (Twitter) 上對 FLUX 3 嘅評價
Introducing FLUX 3. One multi-modal model for Image, Video, Audio and Action-Prediction. Creations are truer to life in every kind of style. FLUX 3 Video is now available in early access (link below). Jointly trained in one unified architecture, our model can be extended to Show more
FLUX 3 Video is #2 in the world. To celebrate, FLUX 3 Video is free to use in our playground until Sunday 16th, 11:59pm PT (link in the thread). This is just the beginning of what’s coming. Up next: 4K, video editing, and multiple images & videos as reference input.
Last week, @bfl_ai launched their first video model, FLUX 3 Video. They have been testing an update on @arena, and the new version is ranked #2 in the Text-to-Video Arena! With 1496 pts, the updated FLUX 3 Video is just 16 pts behind the #1 spot, Gemini Omni Flash (1512 pts).
Action: An early version of FLUX 3 is now running on robots. @mimicrobotics was one of the first partners to gain early access to FLUX 3. Together we developed FLUX-mimic, a video-action model combining the FLUX 3 backbone with mimic's expertise in robot learning for dexterous Show more
Black Forest Labs just announced FLUX 3: a unified multimodal model for image, video, audio and action prediction. Founded in Freiburg, Germany, one of the few globally significant European companies in the AI sector. The current release is gated early access, not open source. Show more
FLUX 3 Video is here. Serious, fun, creative, real, cinematic, whatever you need it to be. Native audio, Text to Video, Image to Video with multiple frames, video continuation, dialogue in multiple languages. Comes with Draft mode so you can explore ideas fast at a fraction of
Black Forest Labs video model, which I posted about when it was first teased on their site back in August 2024, is about to arrive in the form of a multimodal Flux 3. A placeholder page was briefly up yesterday at bfl.ai/models/flux-3 before being quickly taken down. It Show more
Reel · Specifications
咩係 FLUX 3
Black Forest Labs 嘅多模態前沿模型，可以生成自帶原生音頻嘅視頻，已上線 Dreamega AI
- · 015-20s片段時長
- · 021080p最高解析度
- · 034模式：文生 / 圖生 / 首尾幀 / 延長
- · 047寬高比
FLUX 3 係 Black Forest Labs 第一個多模態前沿模型：佢唔再分開訓練圖像、視頻同音頻系統，而係喺一個統一架構入面一齊學晒三樣嘢，同一個主幹網絡仲用嚟做機械人動作預測。視頻版本可以生成最長 20 秒嘅片段，自帶同步原生音頻、有口型同步嘅多語言對白同電影級運鏡控制。喺 Dreamega AI 提供四種模式：文生視頻用提示詞直接生成，圖生視頻令參考圖郁起嚟，首尾幀生視頻駁埋兩張關鍵幀，視頻延長就續寫已有嘅 MP4 片段。每種模式都可以生成 5 至 20 秒，支持 720p 或 1080p。
Reel · Capabilities
FLUX 3 嘅核心能力
Black Forest Labs 多模態模型喺文生、圖生同視頻片段驅動生成上嘅表現
- Feature 01 / 08
統一嘅多模態模型
FLUX 3 喺單一統一架構入面同時學習圖像、視頻同音頻，而唔係將獨立系統夾埋一齊，所以動作、聲音同畫面都嚟自對場景嘅同一份理解。
- Feature 02 / 08
原生同步音頻
對白、音效同環境聲喺生成視頻嘅同一過程入面產生，同畫面發生嘅事緊密對應。音頻預設開啟，而且唔會增加任何積分消耗。
- Feature 03 / 08
多語言對白同口型同步
角色可以用英文、中文、西班牙文、法文、德文、日文、葡萄牙文、俄文、意大利文、印地文等多種語言講嘢，口型動作同對白相匹配。
- Feature 04 / 08
單段最長 20 秒
時長可以喺 5 至 20 秒之間任揀整數秒，成段片一次過生成完，足夠容納多段落場景或者一次完整運鏡。
- Feature 05 / 08
首尾幀控制
上載首幀同尾幀圖片，FLUX 3 會生成兩者之間嘅過渡，並按照你嘅提示詞呈現中間嘅動作、鏡頭運動同氛圍。
- Feature 06 / 08
視頻延長
將一段最長 15 秒嘅現有 MP4 片段續寫做新嘅 5 至 20 秒鏡頭，同時保持相同嘅角色、服裝、燈光同動作連貫性。
- Feature 07 / 08
720p 同 1080p 輸出
草稿同社交平台交付揀 720p，成片揀 1080p。每檔都係按秒計費，短測試成本低，淨係成片先付全價。
- Feature 08 / 08
七種寬高比
可以渲染 21:9、2:1、16:9、4:3、1:1、3:4 或者 9:16，一次過覆蓋超寬銀幕、寬螢幕、方形社交同直向短片，唔使事後重新構圖。
點樣用 FLUX 3 文生視頻
用一段文字描述生成 5 至 20 秒有聲音嘅片段
寫提示詞
描述場景、主體、運鏡、光線同氛圍。如果想要對白，將台詞放喺引號入面，仲要註明語言。
FLUX.2 係 Black Forest Labs 嘅圖像生成同編輯系列。FLUX 3 係全新嘅多模態前沿模型，喺同一個架構入面聯合學習圖像、視頻同音頻，首個發布版本係視頻模型：可以生成最長 20 秒嘅片段，自帶原生同步音頻同多語言對白。Dreamega 透過 WaveSpeed 提供呢個模型。
四種。文生視頻用提示詞直接生成片段；圖生視頻令你上載嘅參考圖郁起嚟，同時保持主體外觀；首尾幀生視頻生成首幀同尾幀之間嘅過渡；視頻延長將一段最長 15 秒嘅現有 MP4 片段續寫做新鏡頭。
輸出係 720p 或者 1080p。時長可以係 5 至 20 秒之間任何整數秒，成段片一次生成完成。提供七種寬高比：21:9、2:1、16:9、4:3、1:1、3:4 同 9:16。
會。音頻同畫面一齊產生而且預設開啟，涵蓋對白、音效同環境聲。對白支持多種語言，包括英文、中文、西班牙文、法文、德文、日文、葡萄牙文、俄文、意大利文同印地文，仲有口型同步。如果淨係想要無聲畫面，都可以關咗音頻。
視頻延長接收單個 MP4 檔案，最長 15 秒、最大 50 MB。新片段由輸入結束嘅位置繼續，所以請喺提示詞入面描述接落嚟嘅動作，並將寬高比設定成同來源視頻一致。
所有模式都係按生成視頻嘅秒數計費：720p 係每秒 26 積分，1080p 係每秒 44 積分。5 秒 720p 片段係 130 積分，10 秒 1080p 片段係 440 積分。音頻生成已經包含在內，唔會額外收費。