P-Video 2
Pruna AI · AI videoP-Video 2 is Pruna AI's quality-focused video model and the successor to P-Video. It generates video from text, a starting image, or an audio track, with native-speech lip-sync, sharp close-ups, and strong identity consistency.
It returns video at up to 1080p and 48 fps with optional native audio, supports explicit 1–20 second durations, multi-aspect ratios, and a draft mode for faster previews.
On VidMachine, scene generation uses image-to-video at 720p via Replicate at 6 credits per second. For AI Influencer talking-head segments, use P-Video Avatar instead.
Key features and benefits
Image-to-video with optional audio
Animate a starting frame from a motion prompt. Attach soundtrack audio to condition generation; when audio is provided, output length follows the audio instead of the duration parameter.
Native speech and lip-sync
Write dialogue directly in the prompt with audio saving enabled for native speech and lip-sync, or supply your own audio track when generation needs to follow a specific recording.
Longer clips
Durations run from 1 to 20 seconds, giving more room for multi-beat action and dialogue than the original P-Video.
Quality-focused renders
P-Video 2 targets premium output with strong identity consistency. Draft mode is available for iteration, but VidMachine always runs full-quality (non-draft) renders at 720p.
Technical specifications
ProviderPruna AI (P-Video 2)
Primary inputsText prompt; start image for image-to-video; optional last-frame image; optional audio (flac, mp3, wav)
Resolutions720p or 1080p (VidMachine: 720p only)
Frame rate24 or 48 fps (VidMachine: 24 fps)
Aspect ratio16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 1:1 when no input image; with an image, framing follows the source
Duration1–20 seconds when no audio is set; with audio, length follows the track
Use cases and applications
Animate scene start frames into longer motion clips for AI Video projects—multi-beat action, camera moves, and product or character motion from a strong still.
Produce short-form social creatives that need native speech or dialogue without a separate lip-sync step.
Why this model
Choose P-Video 2 when you want Pruna AI's highest-quality Replicate-backed model for general scene motion, with longer durations and optional audio conditioning.
For lip-synced AI Influencer talking heads, use P-Video Avatar. Pair P-Video 2 with other models in your priority list when you need a fallback.
How VidMachine uses it
VidMachine generates scene clips via image-to-video at 720p (24 fps, full quality—not draft mode) using each scene's starting frame and motion prompt. Optional last-frame conditioning is passed when available. Duration is clamped to 1–20 seconds for standard AI Video scenes.
Generation runs on Replicate (prunaai/p-video-2). Billing is 6 credits per second of output video at 720p.
What you should know
What resolution does VidMachine use?
Always 720p. The Replicate API also supports 1080p, but VidMachine locks resolution to 720p for consistent cost and pipeline behavior.
How does P-Video 2 differ from P-Video?
P-Video 2 is Pruna AI's quality-focused successor. It supports longer 1–20 second durations, stronger identity consistency, and native speech, at a higher rate of 6 credits per second versus 4 for P-Video.
Can I use P-Video 2 for AI Influencer projects?
No. Influencer talking-head segments use P-Video Avatar (p-video-avatar) on influencer_video_model_priority. P-Video 2 remains available for AI Video scenes.
Does VidMachine use draft mode?
No. Scene generation always runs with draft disabled for production-quality output.
How is P-Video 2 billed?
6 credits per second of generated scene video duration.