Camera Intelligence has added Generative Video Editing to its Caira app, powered by Google's Gemini Omni and Runway's Aleph 2.0. It reworks footage already shot on the camera, no text-to-video, and launches in private beta today.
Following the slightly bumpy introduction of Camera Intelligence’s Caira, its AI-native Micro Four Thirds mirrorless camera, the company has just taken the wrappers off its Generative Video Editing feature. Having delivered the majority of Kickstarter units, Camera Intelligence says it surveyed 111 backers on what to build next; 77% said generative video editing would be useful.
Powered by Google's Gemini Omni 1.1 Flash and Runway's Aleph 2.0, the company says it is the first to embed video generation models with real-world scene understanding directly into a mirrorless camera system.
The feature works through the Caira app on footage already captured with Caira, applying visual and camera-motion effects via natural-language templates. Gemini Omni is a decent choice for this as it is capable of building a persistent understanding of a scene's geometry, lighting, materials, and motion. It’s not always 100%, but the aim is that a scene can be relit or extended while people, objects, and camera movement stay consistent across a clip.
It launches with four templates: Relight, Reframe, Camera Motion, and Set Extension, covering color grading and simulated camera moves such as drone-style sweeps. Templates are built by Camera Intelligence with community feedback, and can be further adjusted using plain-English instructions. Output is available in 720p or 1080p, taking one to three minutes to generate depending on resolution.
The company walked back its open-ended generative photo-editing prompt window in March 2026 after backlash over "vague or silly" test prompts, replacing it with curated buttons. It will be interesting to see how the video tools fare in comparison with their new template approach.
One interesting, and possibly popular, caveat is that there is no text-to-video function. Every edit starts from footage the user has shot with Caira, and a scene cannot be generated from nothing. Generated clips are saved as separate, clearly named files, watermarked with Google's SynthID, and original files are never modified. Caira's on-camera image pipeline remains non-generative, and the feature is optional.
On safeguards, the company says customer footage and images are not used for model training by Camera Intelligence or Google's API, and that it prohibits any alteration of a person's skin tone, ethnicity, or fundamental facial features.
Generative Video Editing will be part of Caira Studio, a premium software tier launching Autumn 2026 alongside Generative Image Editing and an unannounced third feature; pricing has not been set. It will be available in the US and UK only. A private beta opens today for existing Caira customers and five working filmmakers, via an application process at cameraintelligence.com/generative-video-editing or team@cameraintelligence.com.
Meanwhile, the fifth batch of cameras is available to preorder now for $999 and is expected to ship this month.