OmniHuman and Kling in VibeMV: What Changed [2026]

Earlier versions of this guide described OmniHuman-1.5 for Pro lip-sync and Kling V3 Pro for Pro scene generation. You may still see those names in an older tutorial or saved link, but they are no longer the public choices you should look for in VibeMV.

The stable product controls are Base model and Pro model, plus the Scene, Lipsync, and Dance Modes. The implementation behind a product-level model can change, so check the current selector in the app rather than assuming an old provider name is still active.

Historical names mapped to today's controls

This table maps the creator's task, not a live provider guarantee. It does not mean that OmniHuman-1.5 or Kling V3 Pro still serves production requests.

Choose the Mode before the model tier

Use Scene when the Shot should carry story or atmosphere without readable singing. It can run under vocal audio, but if the audience needs to see a performer sing, use Lipsync. Lipsync follows the audio; it does not read or transcribe lyrics. Fast delivery, extreme profile angles, tiny faces, or a covered mouth make the task harder.

Use Dance when choreography is the main event. It is a separate generation path at 12 credits per generated second, rather than a Base/Pro Scene substitute.

Choose Base or Pro with evidence from your Shot

Base currently costs 2 credits per generated second; Pro costs 8. Start with Base when you are still testing the Start frame, prompt, camera direction, or edit rhythm. Consider Pro for a selected hero Shot or visible performance only when the extra cost is justified by the result.

For a fair comparison, keep the same Shot, song section, Start frame, Mode, and prompt, then generate one Base version and one Pro version. Review face stability, intended motion, framing, sync where relevant, and continuity with the neighboring Shots. A single comparison tells you which result works for that Shot; it is not a universal benchmark for every style.

Example mixed workflow

• Instrumental intro: Base model + Scene while you establish the visual language.
• Singer-facing chorus: compare Base and Pro with Lipsync on the same short vocal Shot; keep the version that reads best.
• Atmospheric cutaway: Base model + Scene for a lower-cost transition.
• Eight-count hook break: Dance for one clear performer and one continuous choreographed phrase.

There is no required mix. Model and Mode are Shot-level decisions, so a music video can reserve Pro credits for the moments that survive an actual A/B review.

Credits and review

Scene and Lipsync use the selected Base or Pro rate. Generated Start frames, regeneration, Dance, and the optional 2K version add credits. The Base vs Pro model comparison includes a same-Shot A/B method and a mixed-tier budget; pricing lists current plan terms.

Start with Create a music video from your song, then review each Shot for face stability, motion, framing, sync, and continuity before export.

Frequently asked questions

Can I still select OmniHuman-1.5 or Kling V3 Pro in VibeMV?

Those are historical provider names, not current public selector labels. In the product, choose Base model or Pro model for a Shot, then choose Scene or Lipsync for the task. The provider behind a product-level model can change.

What replaced OmniHuman-1.5 in the workflow?

For the same visible-singing job, choose Pro model with Lipsync. This is a workflow mapping, not a claim that the historical OmniHuman implementation still runs in production.

What replaced Kling V3 Pro in the workflow?

For the same image-to-video scene job, choose Pro model with Scene. Use Lipsync instead when a visible performer should sing.

How much do the current choices cost?

Scene and Lipsync currently cost 2 credits per generated second with Base model or 8 with Pro model. Dance is a separate Mode at 12 credits per generated second.