P Video 2

Prerequisites

Install and load these skills before generating (skip if already in context via @pruna):

Or install the full suite once: npx skills add PrunaAI/pruna-skills@pruna -y

Follow each skill's Before generating / craft sections — do not restate guide content here.

Agent habit

Routing: `p-video-2` is the quality clip generator (text / image / audio). Use `p-video` for simpler, quicker clips. Talking-head-only (no native scene audio) → `p-video-avatar`. Instruction edit of existing footage → `p-video-edit`.

In the first reply, name `p-video-2` in backticks, confirm PRUNA_API_KEY (or stop with signup links from pruna-api), then ask for required inputs. Open intake → generation-diversity clarification intake before the first POST. When drafting motion prompts, follow Prompt craft (dynamic + faithful) and video-prompting (p-video-2-prompting) — do not paste skill examples. Redirect when When NOT to use fits better.

Prompt craft (dynamic + faithful)

Every input.prompt must be fresh and specific, and must match the user's beat. Diversity never overrides the brief. Craft deltas vs p-video: video-prompting (p-video-2-prompting).

Fidelity check (before pay): if you remove the user's named subject/action/setting from the prompt, the job is wrong — rewrite. Free axes only fill what the brief left open.

When showing a drafted prompt, still name `p-video-2` (guides help craft; this tool owns the call).

Skill boundary

This skill = one p-video-2 prediction per invocation.

Out of scope (do not execute from this skill):

• Multi-scene assembly, concat, subagent orchestration, or parallel scene batches
• Motion transfer from a template video → p-video-animate
• Talking-head-only / lip-sync host without native scene audio → p-video-avatar
• Instruction edit of existing footage → p-video-edit

If the request exceeds one clip, stop and recommend: image-to-video (one narrated beat), visual-transition-reel (multi-scene visual), or narrated-multi-scene (multi-scene + VO).

When NOT to use

Use a different skill instead:

HTTP (curl)

Upload for image-to-video / frame anchors

curl -X POST "https://api.pruna.ai/v1/files" \
  -H "apikey: ${PRUNA_API_KEY}" \
  -F "content=@/path/to/first-frame.png"

Pass urls.get as input.image (first frame) and/or input.last_frame_image (last frame). Upload TTS/music the same way for input.audio.

Create (async text-to-video — recommended)

curl -X POST 'https://api.pruna.ai/v1/predictions' \
  -H 'Content-Type: application/json' \
  -H "apikey: ${PRUNA_API_KEY}" \
  -H 'Model: p-video-2' \
  -d '{
    "input": {
      "prompt": "A sports car drifting through a neon-lit city at night, cinematic aerial shot",
      "duration": 5,
      "resolution": "720p",
      "aspect_ratio": "16:9"
    }
  }'

Poll and download: follow pruna-api.

Complete the random seed ritual from generation-diversity before writing prompts — do not pass the ritual string as API seed.

Billing is per returned second (not the requested duration). Rate limit: 250 requests per minute.

First / last frame (visual transition)

curl -X POST 'https://api.pruna.ai/v1/predictions' \
  -H 'Content-Type: application/json' \
  -H "apikey: ${PRUNA_API_KEY}" \
  -H 'Model: p-video-2' \
  -d '{
    "input": {
      "prompt": "OPEN: hold wide on wet alley, neon flicker. MID: slow dolly in, rain ticks on pavement. CLOSE: settle on end pose.",
      "image": "https://api.pruna.ai/v1/files/START_ID",
      "last_frame_image": "https://api.pruna.ai/v1/files/END_ID",
      "duration": 5,
      "resolution": "720p",
      "fps": 24
    }
  }'

Scene anchor triple (image + last_frame_image + audio)

curl -X POST 'https://api.pruna.ai/v1/predictions' \
  -H 'Content-Type: application/json' \
  -H "apikey: ${PRUNA_API_KEY}" \
  -H 'Model: p-video-2' \
  -d '{
    "input": {
      "prompt": "Close-up of a singer performing the uploaded track. Natural lip-sync, expressive face, stage lighting, camera holds steady on the performer.",
      "image": "https://api.pruna.ai/v1/files/SCENE_START",
      "last_frame_image": "https://api.pruna.ai/v1/files/SCENE_END",
      "audio": "https://api.pruna.ai/v1/files/SCENE_NARRATION",
      "resolution": "720p",
      "fps": 24,
      "save_audio": true
    }
  }'

Omit duration when audio is set. Clip length = min(audio length, 20s) — keep TTS ≤ ~19s (ffprobe before render). Audio-conditioned jobs bill the audio length.

Create (sync — quick test only)

curl -X POST 'https://api.pruna.ai/v1/predictions' \
  -H 'Content-Type: application/json' \
  -H "apikey: ${PRUNA_API_KEY}" \
  -H 'Model: p-video-2' \
  -H 'Try-Sync: true' \
  -d '{
    "input": {
      "prompt": "A sports car drifting through a neon-lit city at night, cinematic aerial shot",
      "duration": 5,
      "resolution": "720p",
      "draft": true
    }
  }'

Before generating

1. Complete Prerequisites guide reading order (generation-diversity → video-prompting).
2. Ritual seed → draft a dynamic + faithful motion prompt (section above) → confirm mode (T2V / I2V / frame pair / audio), duration (omit to let the model choose from the prompt, or when audio-driven), resolution, fps, draft, and prompt with the user.
3. Pruna notes: when image is set, aspect_ratio is ignored. With audio, omit duration and keep save_audio: true (default) to keep narration. For on-camera speech, lead with T2V native speech (save_audio: true, no imported audio) — pause / line / rest lip-sync is the quality jump vs p-video. Imported audio keeps a cleaner face; viseme lock is mixed — do not treat a muxed wav as the lip-sync demo. Iterate in draft: true, then final with draft: false. If the request is multi-scene — stop (see Skill boundary). Native 4K is not supported.

Required input

• prompt (string)

Common optional fields

Related

Related skills: