# Kling Video 4.0 API: Enable 4K HDR and Verify the File Before Delivery

The November 19 update adds native 4K and HDR to Kling's video endpoints. Validate the settings, inspect the delivered file and confirm pricing before scaling.

By Paras Madan | 2026-11-21


Kling's [November 19 API update](https://kling.ai/document-api/updates/api) adds native 4K and HDR to the Kling Video 4.0 text-to-video, image-to-video and Omni video routes. This is a change to the existing video API, not the separate Kling Image 4.0 release. A prompt that says “4K HDR” is not enough to request this output.

If you generate clips for an editor, make the delivery format an explicit setting. Test a short clip through the same download, editing and export pipeline your team will actually use.

## Set resolution and dynamic range separately

The [Omni video API reference](https://kling.ai/document-api/api/video/4-0) puts both fields under `settings`. For 4K HDR, send the following settings fragment in an otherwise valid request:

```json
{
  "settings": {
    "resolution": "4k",
    "dynamic_range": "hdr"
  }
}
```

HDR is supported at `1080p` and `4k`. Do not offer it for `720p`. If dynamic range is omitted, the Omni reference lists SDR as the default. The [developer overview](https://kling.ai/dev/api) describes the new output as 10-bit HDR.

Keep the request and the returned media metadata next to each other. If the editor sees an SDR asset, you want to know whether the original generation requested HDR, the delivered source was tagged differently, or a later transcode converted it.

## Separate the creative review from the delivery test

Start with a short clip you can afford to repeat. Keep the prompt and other inputs fixed, then compare 1080p SDR, 1080p HDR and 4K HDR. Add 4K SDR if you need to isolate the resolution change. This compares modes without making every test a high-resolution HDR job.

Use an example with dark detail and bright highlights. Check the visual result as a creative asset, but do not use appearance alone to infer its file properties. An application can show a tone-mapped HDR clip on an SDR display, making it hard to tell what the underlying file contains.

For scheduled jobs, validate the combinations before submission. A preset labelled “social draft” can stay SDR while a specific delivery preset requests HDR. Do not silently upgrade everything in a queue because the endpoint has a new option.

## Inspect the downloaded output

For an original asset you have already downloaded, a local metadata check can look like this:

```bash
ffprobe -v error -select_streams v:0 \
  -show_entries stream=codec_name,width,height,pix_fmt,color_primaries,color_transfer,color_space \
  -of json clip.mp4
```

Check the actual width and height against the requested aspect ratio and delivery requirements rather than assuming every 4K clip has the same dimensions. Inspect pixel format and colour metadata separately. A format that supports ten bits is not, by itself, proof of the transfer function your delivery pipeline expects.

Then import the file into the editor and inspect an exported sample using the same procedure. If the final delivery is SDR, make the conversion a conscious step and review the result rather than letting an arbitrary downstream tool make the choice.

## Confirm charges before increasing duration

The November 19 notes say HDR adds a charge per second of **output** video. The [billing guide](https://kling.ai/document-api/productBilling/billingMethod) also explains that base video charges can depend on input video duration as well as output duration. The surcharge basis and base billing duration are not necessarily the same.

The public billing table consulted for this update did not list a usable 4K base rate for Kling 4.0. Confirm the current rates for your account and compare billed usage for a small job before budgeting a batch. Do not reuse the 1080p price or an older Kling 3.0 4K rate as a stand-in.

Once the pipeline is proven, keep the short SDR preset for iteration and promote only the chosen shots to the final format. The success check is a file your editor can deliver, with the expected colour behavior and a known charge, not just an API task marked complete.

I'm Paras. I work at the intersection of AI, engineering and distribution.

I help AI teams explain launches and API changes through technical writing developers can use.

[Book a 30-minute call](https://cal.com/paras-madan/30min)
