Kling 4.0 上线原生 4K 与 10-bit HDR 输出
Kling 4.0 Adds Native 4K and 10-Bit HDR Output
Kling 4.0 已支持原生 4K 视频输出和 10-bit HDR,HDR 在 4K 与 1080p 下均可使用,并附带 21:9 超宽画幅。Kling 称 10-bit HDR 在逆光、夜景和霓虹场景中能保留更多高光与阴影细节、减少色带,为后期调色留出空间。官方公告未说明所用 HDR 标准、文件编码与封装格式,也未公布 4K 或 HDR 渲染的积分价格与时长限制。
Kling 4.0 把原生 4K 与 10-bit HDR 从预告变为可用,但 HDR 标准、编码格式和渲染成本仍未公布,读者可据此判断能否接入现有后期流程。
VP Land
Nov 20, 2026 · 3 min read

Kling 4.0 has added native 4K video output and 10-bit HDR, which works at both 4K and 1080p, according to Kling AI's post on X. AI shots can now come out of the model at delivery resolution with more color depth for the grade, without an upscale pass in between.
- 4K is now live in Kling 4.0. Kling AI had listed native 4K output as coming soon on its Kling 4.0 feature page, and the post says it's now supported.
- 10-bit HDR isn't tied to 4K. HDR output also works at 1080p, so the extra color depth is available at the smaller resolution too.
- 21:9 framing comes with it. The Kling 4.0 feature page lists 21:9 ultrawide framing alongside output up to 4K, which Kling says gives wider framing for cinematic shots and large-scale scenes.
- The upgrade lands on a model built for long takes. Kling 4.0 already focuses on long single-pass generations, keyframes and reference-driven control.
Kling says 10-bit HDR keeps highlights and shadows in backlit, night and neon scenes
The Kling 4.0 user guide describes the 10-bit HDR output as delivering "richer color gradation and smoother transitions between light and shadow." It names backlighting, night scenes and neon-lit environments as cases where more highlight and shadow detail is preserved, with reduced color banding.
Kling also says the output provides "greater flexibility for post-production color grading." That's the claim colorists will want to test on their own shots before planning a grade around generated footage.
Native 4K joins Kling 4.0's longer, reference-heavy generations
Kling's launch release billed Kling 4.0 as a next-generation video model. Per the user guide and feature page, the resolution upgrade sits alongside these capabilities:
- Up to 30 seconds per generation. Video Extension continues forward or backward across passes to reach up to 2 minutes in total.
- Up to 10 keyframes. They guide character poses, scene changes and shot transitions along the timeline.
- Up to 15 reference assets per task. That covers up to 10 images, 5 videos (up to 30 seconds total), 7 elements and 3 video elements.
- Stereo audio and lip sync. Two-channel stereo with more accurate lip sync keeps voices and dialogue in time with performances.
- Targeted video editing. You can change expressions, camera positions, backgrounds and style while the rest of the scene stays stable.
Kling 3.0 already offered native 4K alongside multi-shot video and built-in audio. Kling 4.0 had listed native 4K and 10-bit HDR as coming soon, and this update turns both on.
Resolution has been climbing on Kling's image side as well, where Kling Image 4.0 launched with native 8K output.
Kling's announcement doesn't name the HDR standard, file format or 4K render cost
The announcement covers resolution and bit depth but leaves out the details a post team needs to slot the footage into a pipeline:
- HDR standard and format. Kling's announcement and feature page don't say which HDR transfer function the output uses or what codec and container the files come in.
- Cost and limits. The announcement doesn't list credit pricing or duration limits for 4K or HDR renders.
- Third-party access. fal's description of Kling 4.0 doesn't mention 4K or HDR, though the model runs on fal's Kling 4.0 page in the playground and fal Agent, with API access coming soon.
KEEP READING
NVIDIA's MotionBricks Adds a Pretrained Human Model for Real-Time Character Animation From Target Poses
NVIDIA released MotionBricks, the real-time motion generation model from its SIGGRAPH paper, in its own official GitHub repository with a new pretrained model for SOMA human characters, after a first release that cove…
Nov 19, 2026 · 4 min read
Runway Labs Introduces Diffusion-UI, a Research App That Generates Interactive Interfaces in Real Time and Can Use Structured Data
Runway Labs introduced Diffusion-UI, a research application in which a real-time model generates every pixel of an interactive interface, and shared an early look at generated interfaces that can use structured data,…
Nov 19, 2026 · 2 min read
Apple Releases FastFrames Video Models for Interactive Generation, With Frames in as Little as 36 ms on One B200 GPU
Apple released FastFrames on Hugging Face, two autoregressive video diffusion models that generate 480p video at interactive latency, with the 1.3B FastFrames-S producing a frame every 36 ms on one B200 GPU and the 14…
Nov 19, 2026 · 3 min read
来源:VP Land · vp-land.com