ShengShu Technology has introduced the preview version of its flagship video generation model, Vidu Q4 Preview. The main advantages of this version include more convincing character acting and a significantly reduced initial usage cost.
According to a report by IT Home on October 10th, and a review on Tencent News from October 8th, the model allows uploading up to 15 images as references, which provides a more complete visual representation of characters, scenes, and props. Furthermore, up to three audio clips can be used to maintain the consistency of the character's voice.
ShengShu claims that the model has improved camera movement dynamics, frame changes, and transitions between complex scenes. Output content is available in resolutions of 540p, 720p, 1080p, 2K, and 4K with 10-bit color depth, and one generation can last up to 16 seconds. Currently, Q4 Preview only supports image-to-video and reference-to-video generation.
Another important aspect is the pricing. For its proprietary SaaS product Vidu, the preview version offers a two-month promotional price starting from 0.09 yuan per second, making a 10-second clip cheaper than 1 yuan. However, the standard price after the promotion ended was not disclosed in any of the reports. The reviewer noted that in China, the main API prices for video models usually range from 0.5 to 3 yuan per second, and since creators often generate many variations before obtaining a suitable clip, the cost per second is critical for determining the accessibility of experiments.
During testing, which included fantasy advertising, a short anime film, and a cinematic style scene, the model successfully handled a slow camera pull-out from a close-up of a dessert to a wide garden landscape, avoiding the abrupt zoom-like cuts characteristic of previous models and maintaining continuity between frames. The reviewer described a recognizable stylistic approach favoring energetic camera movements, anime, and cinematic action. It should be noted that these are the impressions of one reviewer, not benchmarking results.
ShengShu positions Q4 Preview for use in creating short dramatic videos, advertisements, social content, and film production. If the third quarter generation focused on ensuring consistent AI character behavior in production processes, the fourth quarter combines improved performance with a sufficiently low price to encourage repeated content creation attempts. The release date for the full Q4 version has not yet been announced.