P-Video-Edit
P-Video-Edit is Pruna AI's instruction-based video editing model for changing an existing clip while preserving its motion, scene structure, and audio. Give it a source video and a text instruction to make attribute, object, or environment edits; up to four reference images can guide identity, style, or other supported changes. A draft mode supports faster iteration before rendering the final edit at full quality.
Editing video
How to edit a finished clip with Pruna P-Video-Edit: one source video, one written instruction, and the motion, framing and audio the model holds while it changes what you named.
Draft mode and reproducible reruns
How to use settings.draft in Pruna P-Video-Edit to preview an edit at a lower rate before committing, and how seed makes the run you approved repeatable.
Prompt upsampling
How settings.promptUpsampling rewrites your instruction before Pruna P-Video-Edit runs it, what it invents when a prompt is thin, and why a precise instruction barely notices it.
Editing with reference images
How to steer a Pruna P-Video-Edit edit with up to four reference images, when a still pins down a product or garment better than words, and how references differ from P-Video-Replace.
Removing objects from video
How to remove an object or a person from a clip with Pruna P-Video-Edit, naming the target and the surface that continues behind it so the fill holds for every frame.