p-video

The p-video skill generates videos using Pruna's optimized P-Video and WAN models through the inference.sh CLI (belt). Pruna optimizes AI models for speed without sacrificing quality, so this skill is aimed at fast, economical text-to-video and image-to-video generation from the command line. The basic workflow is belt login followed by belt app run pruna/p-video with a JSON --input containing a prompt and optional settings.

Three models are documented: P-Video (pruna/p-video) for text-to-video and image-to-video with audio support; WAN-T2V (pruna/wan-t2v) for text-to-video at 480p/720p; and WAN-I2V (pruna/wan-i2v) for animating images at 480p/720p. Examples show text-to-video with duration and resolution, image-to-video by passing an image URL, audio input that syncs with the video, 1080p high-quality generation, and a draft mode (draft: true) for faster, cheaper output. Resolution and pricing differ by model: P-Video supports 720p and 1080p priced per second and varying by resolution and draft; WAN-T2V is $0.05 (480p) or $0.10 (720p) per video; and WAN-I2V is $0.05 (480p) or $0.11 (720p) per video. Users can browse Pruna apps with belt app list --namespace pruna.

The skill targets creators and developers who want quick, low-cost video generation and image animation. It requires the inference.sh CLI (belt), installable via npx skills add belt-sh/cli, and limits its allowed tool to Bash(belt *). Related skills cover the full inference.sh platform, all video-generation models, an image-to-video guide, Pruna image generation, and text-to-speech for narration.