Video

Video Generation

Video starts with the outcome you need, keeps every step and source attached, and saves finished media in Assets. You can also open the manual tools when you want to control one generation directly.

How to access

Open Create, then choose Video. Requires an active synfluencer.

Outcome recipes

• UGC Ad — Builds a source-backed short ad with scenes, captions, and a call to action. Choose an authorized reference person, an original fictional presenter, a product-only demo, or your own uploaded performance; likeness permission appears only for modes that use a real person.
• Founder Product Video — Turns approved product facts into four independently rendered founder/product acts plus a five-second end card. The approved Fish Audio master is divided into matching act tracks, so you can replace one weak act without paying to rerender the other three.
• Viral Hook — Creates measurable opening variants without promising that any hook will go viral. An optional existing video body is attached after each generated hook, with its original audio beginning after the hook.
• Explainer — Plans and assembles a clear explanation around real product screens and approved claims. When a presenter is selected, that model appears picture-in-picture over the walkthrough instead of replacing it.
• Content Director — Produces a shoot package and can finish validated footage later.
• VFX / 4K VFX — Applies a controlled treatment to authorized source footage, preserves approved audio, and labels upscaled 4K honestly.

Persona playbooks and Brand systems are managed together in the profile's Persona & brand workspace because their approved versions guide every drafting and media workflow. Voice & knowledge remains their evidence library, Assets remains their output library, and Podcast is not currently available.

Video tools

These direct tools stay visible below the outcome recipes for one-off clips. Talking-video choices are labeled by the model you will use, not by the company routing the request.

• Talking video — Generates a presenter video from your script and approved reference image. Choose HG Avatar IV, LC Avatar, or Generate both. The active authorized clone is used when selected; otherwise Synfluencer creates a generic Fish Audio track. Every paid video action stays disabled until you preview and explicitly approve that exact track.
• Generate motion — Creates a new video from a scene prompt. Pick a model from the dropdown — Veo 3.1 Fast (default), Veo 3.1, Kling v3.0 Pro/Standard, Seedance 2.0/Fast, Grok Imagine Video, or Hailuo 2.3. A reference photo is optional: select one for image-to-video, or leave it off for text-to-video.
• Adapt a video — Transfers motion from an authorized source video to an authorized reference image with Kling 2.6 Pro Motion Control. Model-specific duration, format, and reference limits still apply.

Step-by-step

1. Choose an outcome recipe for a planned deliverable, or choose Talking video, Generate motion, or Adapt a video under Video tools for one-off work.
2. Provide the approved brief, source, reference media, and any likeness, voice, claim, or source-rights confirmations the recipe requires.
3. For talking video, choose HG Avatar IV, LC Avatar, or Generate both. These are model choices; internal provider routing is not shown in the customer workflow. Both the manual tool and guided recipes create one Fish Audio track first: an active authorized clone when selected, otherwise a generic Fish voice. Listen to the exact audio, edit and regenerate the narration if needed, then approve it. Both model actions remain disabled until that approval.
4. Review the estimated range and maximum spend. Synfluencer does not submit paid steps until you confirm the maximum.
5. Follow each step's live status. You can cancel a run, retry a failed step without discarding successful work, and return after a worker restart.
6. Review the master, proxy, thumbnail, captions, scripts, costs, and model attribution. Send approved work to Review. Confirmed narration is saved under Assets → Audio with playback, script, voice attribution, and its final customer charge; finished video also remains in Assets.

Cloned voice

Each synfluencer can have its own cloned voice. In the synfluencer profile's Cloned Voice section, record a short sample (read the on-screen teleprompter — three parts, each in a different mood so the clone learns your full range; works on desktop and mobile) or upload an existing audio file. Once a clone exists, the Generate preview box lets you hear it — use Enhance with cues there to test how it performs emotion tags. Then turn on Use my cloned voice in Voice Selection to make it the active voice. In the Talking video tool, the Fish Audio panel lets you optionally enhance the script with emotion cues, generate the narration, preview it, and explicitly approve it before the HG Avatar IV or LC Avatar video action becomes available. If no active clone is selected, the same checkpoint uses a generic Fish voice. Approved audio is saved to Assets → Audio. Fish voice IDs are not portable between services; Synfluencer supplies the generated audio only where the selected video model accepts external audio.

Emotion cues

In the Fish Audio panel, the Enhance with cues button rewrites your script with inline tags such as [pause], [laugh], [whisper], and [excited]. The cued script stays editable, so you can add, remove, or move tags by hand before generating the audio.

How well the cues come through depends on your voice sample. A clone can only express what it heard when you recorded, so a flat or monotone sample limits the range the cues have to work with. Record with real energy and inflection for the best result. [pause] is the most reliable cue; richer emotion tags vary take to take.

Audio upgrades

After generating a video, you can upgrade its audio track. Select from available audio templates, preview the result, and save. Audio upgrades are also available from Assets.