Picsart’s new Reshoot node for Flow changes cast, products or locations while preserving the original footage’s timing, camera movement and edit structure.
Picsart has added a new Reshoot node to Flow, its AI video editing workflow, giving creators a way to rework existing footage without rebuilding the shot from scratch. According to Picsart’s announcement, the tool takes an original clip, combines it with reference images and a written instruction, and then generates a new version that preserves the timing, camera movement and cut points of the source material while changing the people, products or locations in frame.
The feature sits within a broader Flow suite that Picsart markets as an AI video production environment for short-form content. On its video animation page, Picsart says Flow is designed for polished clips for TikTok, Reels and Shorts, with support for motion templates, animated text, subtitles and multiple output formats. That places Reshoot inside a system aimed at rapid versioning, rather than one-off generative clips.
Reshoot is built around two modes. Recast is intended for broader replacements, such as swapping cast, wardrobe and set design, while Swap is narrower and changes only the elements pointed to by the supplied references. Picsart says the distinction matters because the original motion, camera work and edit structure are retained in both cases, making the feature useful for controlled variations of the same scene.
The company also frames the tool as a way to stretch a single production into multiple campaign assets. In Picsart’s description, one product shoot can be re-rendered for different colourways, models or locations, with each output appearing as a separate node beside the original on the canvas. The aim is to let marketers and content teams build a family of variants from one take, rather than organising repeated shoots for each version.
Technical limits are part of the workflow. Picsart says clips must be between four and 30 seconds long, users can supply between one and 30 reference images, and the output can be rendered at 480p, 720p or 1080p. The company also says users can link parts of their prompt to a specific reference image with an “@”. That makes the instruction set more explicit, particularly when several references are used in one run.
The release arrives as other AI video platforms push deeper into post-production rather than text-only generation. Digital Camera World reported that Luma AI’s Ray3 Modify can alter real footage while keeping performance and timing intact, including wardrobe, lighting and backgrounds. TechRadar has also reported on Camera Intelligence’s forthcoming generative video editing for its Caira camera, which is limited to footage shot on that device. The direction of travel is clear: AI video tools are increasingly focused on changing existing footage with precision, not merely inventing scenes from prompts.
For Picsart, that emphasis on edited reality is likely the point. By keeping the underlying take fixed, Reshoot offers a more predictable route to localised product films, cast changes and scene substitutions than fully synthetic video. It is a practical, production-oriented approach that may appeal most to teams that already have usable footage and want to multiply it quickly across formats and audiences.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.
