AI News AI资讯 6h ago Updated 1h ago 更新于 1小时前 51

Black Forest Labs makes FLUX 3 Video generally available and claims it beats Seedance 2.0 黑森林实验室推出 FLUX 3 Video 通用版本,声称超越 Seedance 2.0

Black Forest Labs has made FLUX 3 Video generally available via API and select partners, marking a significant release in the generative video space. The model produces HD and Full HD clips up to 20 seconds with native audio including dialogue, sound effects, and ambient noise, supporting text-to-video, image-to-video, keyframes, video continuation, and multi-scene/camera-angle generation. FLUX 3 claims top Elo rankings: 1,135 for text-to-video and 1,051 for image-to-video, surpassing Gemini Omn Black Forest Labs 已通过 API 和精选合作伙伴将 FLUX 3 Video 正式发布,标志着生成式视频领域的一项重要发布。 该模型可生成高达 20 秒的高清和全高清片段,包含原生音频(包括对话、音效和环境噪音),支持文生视频、图生视频、关键帧、视频续写以及多场景/多机位生成。 FLUX 3 声称在 Elo 排名中位居榜首:文生视频为 1,135 分,图生视频为 1,051 分,超越了 Gemini Omni Flash、Minimax H3 和 Seedance 2.0。 它支持超过 14 种语言的对口型对话、场景内文字排版渲染、复杂提示词遵循,以及用于纪录片风格内容的世界

75
Hot 热度
70
Quality 质量
72
Impact 影响力

Analysis 深度分析

TL;DR

  • Black Forest Labs has made FLUX 3 Video generally available via API and select partners, marking a significant release in the generative video space.
  • The model produces HD and Full HD clips up to 20 seconds with native audio including dialogue, sound effects, and ambient noise, supporting text-to-video, image-to-video, keyframes, video continuation, and multi-scene/camera-angle generation.
  • FLUX 3 claims top Elo rankings: 1,135 for text-to-video and 1,051 for image-to-video, surpassing Gemini Omni Flash, Minimax H3, and Seedance 2.0.
  • It supports lip-synced dialogue in over 14 languages, typography rendering within scenes, complex prompt adherence, and world knowledge integration for documentary-style content.
  • Pricing is per-second based: Draft HD at $0.06 (T2V/I2V) / $0.12 (V2V); Full HD at $0.17 / $0.41; Full HD at $0.29 / $0.53, with audio included in all tiers.

Why It Matters

FLUX 3 Video's general availability signals intensifying competition in the generative video market, with Black Forest Labs positioning itself as a direct challenger to established players like Google and ByteDance. Its native audio generation, multilingual lip-syncing, and documentary-grade world knowledge integration address key gaps that have historically limited AI video tools for professional and narrative use cases. For AI practitioners, this release provides a competitively ranked, feature-rich alternative worth evaluating for production pipelines.

Technical Details

  • Video Generation Capabilities: Generates HD and Full HD clips up to 20 seconds, with support for text-to-video, image-to-video, keyframe conditioning, video continuation, and multiple scenes with varying camera angles within a single output.
  • Native Audio Integration: Includes built-in audio generation covering dialogue, sound effects, and ambient noise, eliminating the need for separate audio pipelines.
  • Multilingual Lip-Sync: Supports lip-synced dialogue in over 14 languages, enabling multilingual content creation without manual synchronization.
  • Advanced Prompting & Knowledge: Capable of rendering typography directly within scenes, following complex multi-step prompts, and leveraging world knowledge for contextually accurate outputs such as documentary-style videos.
  • Benchmark Performance: Achieved Elo scores of 1,135 (text-to-video) and 1,051 (image-to-video), ranking ahead of Gemini Omni Flash, Minimax H3, and Seedance 2.0 in BFL's internal evaluations.
  • Pricing Structure: Tiered per-second pricing with Draft (HD only) and Full Quality (HD/Full HD) modes, with video-to-video commands costing significantly more than text/image-to-video paths.

Industry Insight

  • The competitive Elo rankings suggest the generative video landscape is rapidly evolving, with Black Forest Labs closing the gap against well-resourced competitors like Google and ByteDance—practitioners should monitor whether these rankings hold in independent evaluations.
  • Native audio and multilingual lip-syncing reduce the friction of post-production workflows, making FLUX 3 particularly attractive for creators producing localized or dialogue-heavy content; teams should evaluate its output quality against existing video-plus-audio pipelines.
  • The per-second pricing model, especially the premium for video-to-video and Full HD modes, may limit cost-effective high-volume use cases—organizations should benchmark draft vs. full-quality outputs to optimize spend while maintaining acceptable fidelity.

摘要

Black Forest Labs 已通过 API 和精选合作伙伴将 FLUX 3 Video 正式发布,标志着生成式视频领域的一项重要发布。
该模型可生成高达 20 秒的高清和全高清片段,包含原生音频(包括对话、音效和环境噪音),支持文生视频、图生视频、关键帧、视频续写以及多场景/多机位生成。
FLUX 3 声称在 Elo 排名中位居榜首:文生视频为 1,135 分,图生视频为 1,051 分,超越了 Gemini Omni Flash、Minimax H3 和 Seedance 2.0。
它支持超过 14 种语言的对口型对话、场景内文字排版渲染、复杂提示词遵循,以及用于纪录片风格内容的世界知识整合。
定价按秒计算:草稿高清为 $0.06(文生视频/图生视频)/ $0.12(视频生成视频);全高清为 $0.17 / $0.41;全高清为 $0.29 / $0.53,所有层级均包含音频。

深度分析

简而言之

  • Black Forest Labs 已通过 API 和精选合作伙伴将 FLUX 3 Video 正式发布,标志着生成式视频领域的一项重要发布。
  • 该模型可生成高达 20 秒的高清和全高清片段,包含原生音频(包括对话、音效和环境噪音),支持文生视频、图生视频、关键帧、视频续写以及多场景/多机位生成。
  • FLUX 3 声称在 Elo 排名中位居榜首:文生视频为 1,135 分,图生视频为 1,051 分,超越了 Gemini Omni Flash、Minimax H3 和 Seedance 2.0。
  • 它支持超过 14 种语言的对口型对话、场景内文字排版渲染、复杂提示词遵循,以及用于纪录片风格内容的世界知识整合。
  • 定价按秒计算:草稿高清为 $0.06(文生视频/图生视频)/ $0.12(视频生成视频);全高清为 $0.17 / $0.41;全高清为 $0.29 / $0.53,所有层级均包含音频。

为何重要

FLUX 3 Video 的正式发布表明生成式视频市场竞争日益激烈,Black Forest Labs 将自己定位为直接挑战 Google 和字节跳动等成熟玩家的竞争者。其原生音频生成、多语言口型同步以及纪录片级世界知识整合,解决了长期以来限制 AI 视频工具在专业叙事用例中应用的关键短板。对于 AI 从业者而言,此次发布提供了一个具有竞争力排名且功能丰富的替代方案,值得在生产流程中评估。

技术细节

  • 视频生成能力:生成高达 20 秒的高清和全高清片段,支持

Disclaimer: The above content is generated by AI and is for reference only. 免责声明:以上内容由 AI 生成,仅供参考。

Video Generation 视频生成 Product Launch 产品发布 Multimodal 多模态