ShengShu Launches Vidu Q4 Preview With 4K Output and Low Entry Price
ShengShu Technology has released Vidu Q4 Preview, the first look at its next flagship video generation model, and the company is selling it on two claims. Character performances should look more convincing than before, and the entry price sits well below what rival video models charge for comparable work. Both claims point at the same buyer, teams producing short drama, advertising, and social video at volume.
The model takes up to 15 reference images so that characters, scenes, and props carry a fuller visual reference. It also takes up to three reference audio clips, which hold a character voice steady across separate shots. ShengShu says it has reworked dynamic camera movement, shot changes, and transitions between complex shots, while output reaches 540p, 720p, 1080p, 2K, and 4K at a colour depth of 10 bits, with a single generation running as long as 16 seconds.
But the list of supported formats is wider than the list of supported tasks. Vidu Q4 Preview handles two jobs for now, turning a still image into video and turning supplied reference material into video, so a written prompt on its own is not yet an input path. Buyers who expected a text driven model will have to wait for the full release.
Pricing is the louder headline. The preview carries a two month launch offer on Vidu's own software service starting at RMB 0.09 per second, which puts a 10 second clip below RMB 1. Neither of the reports behind the launch gave the standard price that will apply once the offer closes.
The width of that gap is the point. Mainstream video model interfaces in China typically charge between RMB 0.5 and RMB 3 per second, according to the reviewer, and creators usually burn through several attempts before a clip is usable. Per second cost therefore settles who can afford to experiment at all.
And a hands on review published ahead of the announcement described the results in favourable terms. Across a fantasy advertisement, an anime short, and a film style scene, the model managed a slow pull from a close up of a dessert plate to a wide garden landscape without the abrupt zoom style cuts that dogged earlier generations, and it kept cuts between shots continuous. The reviewer called the house style recognisable, favouring energetic camera moves, anime, and cinematic action.
Those are one reviewer's impressions rather than benchmark results, and no independent measurement of quality has been published. ShengShu has pitched the model at AI short dramas, advertising, social content, and film production, markets where clip volume matters more than one perfect take.
The generation before it was built around a different problem, getting AI characters to act consistently inside real production workflows. With the new preview, the company pairs that consistency work with a price low enough to encourage repeated drafts, and the reviewer's notes suggest the camera work improved alongside the performances. The full release has no announced date.
So the pitch rests on cost as much as on craft. Pandaily reported the launch, citing IT Home for the release and Tencent News for the hands on review, and ShengShu's own materials describe the build as a preview rather than a finished product.
Vidu Q4 Preview accepts up to 15 reference images and three reference audio clips
Output reaches 4K at a colour depth of 10 bits, in single clips up to 16 seconds long
Launch pricing starts at RMB 0.09 per second for two months, against RMB 0.5 to RMB 3 across mainstream Chinese video model interfaces
Image to video and reference to video are the only supported generation modes so far
Source: Pandaily
