PixVerse Launches Multi-Keyframe Generation, Elevating AI Video Creation from 'Segments' to 'Narrative Expression'}
PixVerse's new multi-keyframe feature allows users to create coherent 30-second videos from up to 7 images, enhancing storytelling and control in AI-generated videos.
Today, PixVerse, used by over 60 million users worldwide, announced the addition of the 'Multi-Keyframe Generation' feature in its start-end frame module. Users can upload up to 7 images as keyframes to easily generate a continuous 30-second video, significantly improving creators' control over AI video storytelling.
▲ Users can add up to 7 keyframes via PixVerse web platform's start-end frame feature.
By analyzing the semantic relationships between keyframes, AI intelligently constructs smooth motion paths for actions and scenes, optimizing dynamic performance in static or slow scenes. This is especially useful for short dramas, product demos, and other deep scene scenarios. For example, turning a product display image into a rotating animation or transforming a storyboard sketch into a flowing short film.
Dr. Wang Changhu, founder and CEO of AiShi Technology, stated at the 2025 Beijing ZhiYuan Conference: “Allowing ordinary users and consumers to create previously impossible videos with AI — this is the 'GPT moment' for video generation. Good models lead to good products.”
The launch of multi-keyframe generation marks a shift in AI video creation from simple segments to complete, coherent narratives. By precisely defining start, transition, and end frames, AI can naturally connect character actions and scene changes (e.g., from walking to running), and mimic director’s shot language, such as switching between close-ups and wide shots, significantly enhancing storytelling. This breakthrough enables AI to better simulate event timelines, greatly improving efficiency for high-narrative scenarios like trailers, comics, and advertising.
Users can combine multiple photos of the same character (e.g., childhood to adulthood) or related scenes (e.g., family members in a group photo) to generate logically coherent narrative videos. Currently, many users have created videos of life memories, celebrity growth stories, family tales, and pet transformations.
Previously, AiShi’s self-developed video generation large model PixVerseV2 supported start-end frame functions, aligning similar images and generating smooth visual effects. As Travis Davids, a digital artist, described, placing similar compositions together is like magic. Overseas users have even created Simpsons-style short clips using this feature, garnering over a million views on social media.
▲ Travis Davids’ start-end frame GIF
Recently, AiShi was recognized at the AI for Good Global Summit as an impactful innovation. Co-founder Xie Xuzhang delivered a keynote titled “AI Video for Good,” emphasizing the goal to help billions of people who have never created or shared videos, making video creation accessible to all.
PixVerse’s global user base has surpassed 60 million. The domestic version launched on June 6, 2025, now supports nine languages including Chinese, English, Japanese, Korean, and French. AiShi aims to provide a better AI video creation experience, leading the development of large-scale AI video models and applications, meeting new creative and consumption needs in the AI era, and continuously lowering barriers to entry so everyone can be a director of their life.