Video model · from 48 credits
Make a photo talk with InfiniTalk
InfiniTalk makes a photo speak: a portrait, an illustration or a generated character comes alive on the connected voice, lips in sync, gaze and expressions lively, up to twenty-eight seconds in one go. The model needs a face image, a voice audio and a short description of the scene; the video lasts exactly as long as the voice, measured by the server, and that duration is what gets billed, in 480p or 720p. It completes the Voice-over node and cloned voices: a personal voice and a photo are enough for a video message, a brand avatar or a story character taking the floor.
Capabilities
- Voice as input: the photo talks for the whole duration of the audio
- Resolutions: 480p, 720p
- Durations: 2–28 s
What InfiniTalk does
A single photo, a face seen from the front or three-quarters, and a voice: InfiniTalk renders a video where the person in the photo says the text, lips synchronized, with the blinks, head movements and expressions that go with speech. It is the talking avatar gesture, without a shoot: a generated portrait becomes a presenter, a drawn mascot makes its announcement, an old portrait tells its story. The optional prompt describes the attitude and the framing, not the text: the text is in the voice.
The duration is the voice's
InfiniTalk takes no duration: the video lasts as long as the wired audio, from two to twenty-eight seconds. The node measures the voice server-side and computes the number of frames to produce, one notch below so that the model accepts it. A voice longer than the cap is refused before generation, with no charge: split the text into two takes and chain them in the Montage node. The voice comes from a Voice-over node, catalog or cloned voice, or from an uploaded audio file.
How it is billed
The rate is per second produced, with two tiers, 480p for drafts and social networks, 720p for a render to project. The price shows on the button as soon as the voice is wired, computed on its measured duration, and a technical failure is refunded. A sharp portrait, lit from the front, mouth closed and visible, on a plain background, gives the most accurate lips. Glasses and beards work fine, a hand in front of the mouth does not.
How to use it on Imaginode
- 1Upload the photo into a Media node and write the lines to say in a Voice-over node, with the voice of your choice.
- 2Wire the image into the image port and the voice into the voice port of a Video node set to InfiniTalk: the video will last as long as the voice, the price shows before you click.
- 3Generate: the person speaks to camera, expressions and head movements included.
Frequently asked questions
Do I need a video to use InfiniTalk?
No: a single image is enough. That is what sets it apart from Sync Lipsync 3, which re-times the lips of an existing video. InfiniTalk invents the face motion from the photo and the voice.
How much text can the photo say?
Up to twenty-eight seconds of voice, about seventy words in English at a normal pace. Beyond that, split into several takes and edit them back to back in the Montage node, for free.
How much does InfiniTalk cost?
InfiniTalk is billed per second of video produced: 120 credits for 5s at 480p, about $1.44 (1 credit = $0.012). The duration is measured server-side on the wired file, never declared: the exact price shows on the button before you click, and a failure is refunded.
Can I try InfiniTalk for free?
Yes: signing up grants trial credits, no credit card required. Enough to test InfiniTalk in real conditions before deciding on a plan or a top-up.
Do I need to install anything to use InfiniTalk?
No: InfiniTalk runs server-side, in the browser, mobile included. Imaginode's node canvas wires its renders to the other models (references, camera, styles) with nothing to install.
Try InfiniTalk on a real canvas
133 models, one account, credits that work everywhere. The price always shows before you generate.
Get started
Kling 3.0 Motion Control
Grok Imagine (prolonger)
Sync Lipsync 3
Bria fond vidéo