Alibaba's Wan 3.0 moves your frame with smooth, coherent motion and renders the voices, ambience, and effects with it.
Start creatingWan's cinematic look, with sound generated in the same shot.
Ambience and effects are generated with the picture, so a splash or a footstep lands on the frame it belongs to.
Write a line into the shot and Hypernatural passes it to Wan 3.0, which speaks it on camera in the same clip.
Hypernatural hands Wan 3.0 each shot's opening frame, and Wan sets it moving smoothly without drifting from its look.
Hypernatural lines Wan 3.0 shots up on a timeline and runs your narration, music, and captions across them.
She crosses the clifftop at an easy pace with the grass bending in the wind, and the moment the cloud lifts her flows straight out of her stride.
The Dazey bag tilts as avocados, limes, and chips sail past, and its wordmark, mascot, and flavor read as clearly in the last frame as in the first.
She moves the can from beside her face into both hands and glances down at it, and neither her fingers nor the Sol logo warp as her grip changes.
Shots where the sound belongs to the scene: someone speaking to camera, a product moment with its own fizz and clink, a quiet beat that needs the room around it. It keeps the cinematic look Wan is known for, and Hypernatural charges far less per second for it than for Veo 3.1 or Seedance 2.5, so it can carry a whole sequence rather than a single hero shot.
Yes. Speech, ambience, and effects come out of the same generation as the picture, timed to the motion rather than dubbed on afterward. Hypernatural then lays your narration, music, and captions over the top in the same timeline.
Yes. Give the shot its line and Hypernatural passes it to Wan 3.0 as speech, which Wan voices in the clip along with the picture. If you give an open-ended brief with someone talking to camera but no script, Hypernatural writes the line itself, and it directs a shot with no line as non-speaking.
Both sit in Hypernatural's picker, and the clearest difference is sound: Wan 3.0 generates audio with the picture, while Wan 2.2 renders silent clips. Wan 2.2 costs a little less per second, so it still suits shots that will sit under narration or music anyway. The picker shows what each one charges when you choose.
Hypernatural generates each Wan 3.0 shot at up to 8 seconds, sized to the shot in whole seconds rather than padded out to a preset clip length. A finished video is as many shots as the story needs, chained on Hypernatural's timeline.
Yes. Create the character once in Hypernatural's library and write their @handle in every shot they appear in — a bare name attaches nothing. Hypernatural builds each shot's opening frame from that portrait, and Wan 3.0 animates from it, so the same person comes back every time. Their outfit follows what the shot describes.
Hypernatural Desktop
Hypernatural Desktop has Wan 3.0 built in. Ask your agent to use it in any workflow, like these.
Turn a raw webcam or talking-head recording full of retakes into a tight rough cut of the best takes, then make it EXTRA in the client's own brand: kinetic titles, karaoke captions, synthesized sound effects and motion graphics that react to the speaker's face and hands, with type behind them that's never more than a little hidden. Use for founder or creator explainers, launch and feature videos, and any 'cut my takes and go wild with graphics' request.
Make a finished narrative video from a written script: adapt it into script.json, cast and perform the voices, design the sets, stage the cast in stills, compose every shot with its dialogue spoken in the cast voices, composite titles over the footage, then make the cuts and deliver. For scripted shorts, skits, training dramas, quiz shows and multi-character scenes made from nothing but text.
Make a ~35 s faceless marketing explainer for an app or product from its own code repo or website: a voiceover over kinetic type in the brand's colours and the product's real screenshots in a phone frame, with a music bed, built as one hn4a-compositor spec cued from the voice take's own word times. Problem, brand, three how-it-works steps, one feature, social proof, end card. Covers mining the repo for copy, facts and images, a script and aspect gate, the opening scene in a few looks on estimated word times shown live minutes after the script, beside the voice and music auditions, then the whole layout in the picked look cued to the picked voice, shown live before the render. Costs well under $1. For an ad with a presenter avatar and rebuilt UI, use workflow-product-ui-video.
Community
© 2026 Hypernatural Systems, Inc.