Try MiniMax FastH3 from Fastvideo →
Sign InStart Building

Task History

Click to see output list

No tasks yet

Go to Models
Video GenerationActive

pruna / p-video-2

p-video-2

bypruna

Generate short videos from text, a start image, or an audio track. Export 720p or 1080p MP4 with 24 or 48 fps and optional last-frame control.

Text to VideoImage to Video
Model ID
p-video-2
Provider
pruna
Updated
1789048819
wiro playground—pruna/p-video-2
Reset to defaults
0 / 1
Maximum 1 image allowed

Optional: First frame for image-to-video.

Required: Describe the video to generate, or the motion to apply to the input image if provided

Required: Length of the generated video. An input audio longer than this is trimmed to this length

Required: Output resolution

Required for text to video. Ignored when an input image is provided

Sample outputs
Updated 1789048819

Overview

P-Video 2 by Pruna generates short MP4 videos from a written description. It can also animate a single still image into a moving shot, or condition motion and lip sync from an uploaded audio track. You pick duration, resolution, aspect ratio, and frame rate, then the model renders one continuous clip. This is useful when you need fast iterations for ads, avatars, and music-driven visuals.

What you can build

  • Talking avatars from one still image, with audio-driven lip sync
  • Social ad variations from one prompt, across multiple aspect ratios
  • Product shots that animate a static catalog image into a short loop
  • Music clip visuals that follow a supplied vocal or instrumental track
  • Quick story beats for short-form posts where you need one clean shot

Inputs

  • A required written description of the scene and motion you want. If you provide a start image, describe the motion to apply to that image.
  • An optional start-frame image to drive image-to-video. Provide a JPG, JPEG, PNG, or WebP file.
  • An optional last-frame reference image to steer how the clip ends. Provide a JPG, JPEG, PNG, or WebP file.
  • An optional audio track to condition the video. Provide FLAC, MP3, or WAV. When you add audio, the clip length follows the audio length.
  • A chosen clip length for text-only runs. On this listing, you can pick 5, 10, 15, or 20 seconds.
  • A chosen output resolution: 720p or 1080p.
  • A chosen aspect ratio for text-to-video: 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, or 1:1. If you provide a start image, the aspect ratio setting is ignored.
  • A chosen frame rate: 24 fps or 48 fps.
  • An optional draft preview switch that generates a faster, lower-quality clip for iteration.
  • An optional prompt enhancement switch that rewrites or expands your text before generation.
  • An optional setting to keep or drop audio in the final MP4.
  • An optional integer seed to repeat the same result more reliably across runs.

Outputs

  • One downloadable MP4 video file.
  • The clip matches your selected resolution, frame rate, and duration.
  • If you provide audio and keep audio enabled, the MP4 includes an audio track.

Recommended settings

  • Fast idea exploration: draft preview on, 720p, 24 fps, 5 seconds.
  • Final social export: draft preview off, 1080p, 48 fps, 10–20 seconds.
  • Vertical content: 1080p with 48 fps often helps preserve detail in 9:16 framing.
  • Repeatable variations: set a fixed seed, then change only 1 prompt detail per run.

Limitations

  • It’s not designed for extreme camera moves or complex multi-scene storytelling in one clip.
  • It doesn’t output 4K.
  • Audio results can degrade with more than 2 distinct speakers.
  • Speaker attribution can drift across lines, especially in longer dialogue.
  • Sound effects quality is limited compared to dedicated audio tools.
  • Low-quality inputs can reduce stability. Blurry images, heavy compression, and low-light photos can cause artifacts or identity drift.

Safety & compliance

  • The model includes a safety filter for prompts and input images, and some deployments expose a switch to disable it.
  • If you disable safety checks, you must apply your own content moderation and follow your platform rules.
  • Provider-run deployments commonly state that inputs and outputs aren’t used for model training, and aren’t retained after processing.

Example prompts

Great starting points for p-video-2.

The silk dress and her hair lift and ripple in a steady breeze as she slowly turns her head toward the camera and lets a faint smile form, camera pushes in gently and drifts a few degrees to the left, warm golden light flickering across the fabric, smooth continuous motion, no cuts, no morphingVideo Generation
The camera slowly orbits around the bottle while a thin curl of mist drifts across the slate and one water droplet slides down the glass, reflections and gold highlights shifting as the angle changes, everything else perfectly still, smooth continuous motion, loop-friendly, no morphingVideo Generation
Steam rises and curls upward from the broth as chopsticks lift a tangle of noodles into frame, broth dripping back into the bowl, camera drifts slowly downward and closer to the surface, warm light glinting on the egg yolk, smooth continuous motion, no cutsVideo Generation
The climber reaches up with one hand and settles into a higher hold, a small puff of chalk dust drifting off the rock as the rope sways slowly beneath, camera pulls back and tilts down to reveal the drop below, harsh sunlight steady on the granite, smooth continuous motion, no cuts, no morphingVideo Generation

API quick start

Run p-video-2 with a single API call.

POST https://api.wiro.ai/v1/Run/pruna/p-video-2
{
  "prompt": "The silk dress and her hair lift and ripp…",
  "duration": 5,
  "resolution": "720p",
  "inputImage": "https://your-cdn.com/input.png"
}
curl
curl -X POST "https://api.wiro.ai/v1/Run/pruna/p-video-2" \
  -H "Content-Type: application/json" \
  -H "x-api-key: YOUR_WIRO_API_KEY" \
  --data-binary @- <<'JSON'
{
  "prompt": "The silk dress and her hair lift and ripp…",
  "duration": 5,
  "resolution": "720p",
  "inputImage": "https://your-cdn.com/input.png"
}
JSON

Pricing: From $0.075/request

View full API docs