veo git:20260810.4dbaa58A

Immutable. This exact content never changes and is served at /api/v1/blob/d52f96d6311f534b.

---
name: veo
description: |
  This skill should be used when the user specifically asks for Veo, or
  names a Veo-specific control: "generate video with Veo", "extend this
  video" / "video-to-video", "first and last frame transition", "ingredients
  to video" / "consistent characters across shots", or integrating with an
  existing Veo pipeline.

  Not the default for a plain "generate a video" / "make a video from this
  photo" request with no Veo-specific control named — prefer Gemini Omni
  Flash for that (Google's current default for video generation: better
  coherence, multi-input reasoning, character consistency, multi-turn
  editing). Reach for this skill only when the task specifically needs
  Veo's own controls.
context: fork
model: sonnet
---

# Veo Video Generation

Generate video with Google Veo 3.1 via the Gemini API. Supports
text-to-video, image-to-video, video extension, and multi-shot workflows
with cinematography control and synchronized audio.

**This is not the default video-generation route.** Google's current
guidance is to reach for Gemini Omni Flash first — it gives better video
coherence, reasons over text/image/audio/video together, keeps characters
consistent, and supports multi-turn conversational editing via the
Interactions API. Come here specifically for Veo's own controls: scene
extension, last-frame control, or integrating with an existing Veo
pipeline. If the request is a plain "generate a video of X," that is
Omni Flash's job, not this skill's.

```bash
export GEMINI_API_KEY="your-api-key"
SCRIPT="${CLAUDE_PLUGIN_ROOT}/skills/veo/scripts/generate_video.py"

uv run "$SCRIPT" "A neon hologram of a cat driving at top speed"
uv run "$SCRIPT" "Slow dolly shot, cinematic lighting" --image photo.jpg
uv run "$SCRIPT" "Zoom out to reveal the skyline" --model veo-3.1-fast-generate-preview --duration 4
```

The path goes through `CLAUDE_PLUGIN_ROOT` because this runs with the
user's project as the working directory, not this one. A bare relative
path does not resolve.

The script declares its own dependency inline (PEP 723), so `uv run`
resolves it; there is no install step. `--help` carries the current flags.

The script covers text-to-video and image-to-video. Video extension,
first/last-frame transitions, and ingredients-to-video (reference images)
take more inputs than a CLI comfortably carries — see
[references/api-examples.md](references/api-examples.md) for worked code
to adapt directly.

## Choosing a Veo variant

| Model | Use when |
|---|---|
| `veo-3.1-generate-preview` | Best fidelity, production/hero shots |
| `veo-3.1-fast-generate-preview` | Faster iteration, drafts |
| `veo-3.1-lite-generate-preview` | Cheapest, quick previews, high volume |

Price scales per second of output, and the same two levers move it for
every variant: resolution (720p/1080p/4k) and whether audio is generated
alongside the video (video+audio costs more than video-only). Check
[Google's Vertex AI generative-AI pricing page](https://cloud.google.com/vertex-ai/generative-ai/pricing)
for the current per-second figures before a spend decision — a number
copied here would go stale silently.

## Duration

Supported single-generation durations are **4, 6, or 8 seconds** —
anything else is rejected by the API. `--duration` defaults to 8.

## The Prompting Formula

For consistent, high-quality results, structure prompts using:

**[Cinematography] + [Subject] + [Action] + [Context] + [Style & Ambiance]**

### Example

```
Medium shot, a tired corporate worker, rubbing his temples in exhaustion,
in front of a bulky 1980s computer in a cluttered office late at night.
The scene is lit by the harsh fluorescent overhead lights and the green
glow of the monochrome monitor. Retro aesthetic, shot as if on 1980s
color film, slightly grainy.
```

For detailed cinematography language (camera movements, composition, lens
techniques), see [references/prompting-guide.md](references/prompting-guide.md).

## Audio Direction

Veo 3.1 generates complete soundtracks based on text instructions in the
same prompt — there is no separate audio parameter.

- **Dialogue** — quotation marks: `A woman says, "We have to leave now."`
- **Sound effects** — described explicitly: `SFX: thunder cracks in the distance`
- **Ambient noise** — background soundscape: `Ambient noise: the quiet hum of a starship bridge`

## Advanced Workflows

For complex projects requiring precise control, combine Veo with an image
model (e.g. Gemini's image generation) to prepare inputs, then feed them
to Veo:

- **First and Last Frame** — controlled transition between two specific
  viewpoints (`last_frame` in the generation config).
- **Ingredients to Video** — consistent characters/objects across shots via
  `reference_images` in the generation config.
- **Timestamp Prompting** — direct a multi-shot sequence with precise
  timing inside a single generation, via the prompt text alone.

See [references/prompting-guide.md](references/prompting-guide.md) for
workflow instructions and [references/api-examples.md](references/api-examples.md)
for the code.

## Watermarking

Google DeepMind documents that Veo output carries an invisible SynthID
watermark — invisible does not mean absent, and it is meant to be
machine-detectable rather than seen. Whether API output additionally
carries a *visible* watermark, the way some consumer surfaces (the Gemini
app, Flow) do, is **not verified here** — do not assume API output is
visibly watermark-free, and do not assume it is watermarked, without
checking a generated clip directly.

## Resources

- [Veo API Reference](https://ai.google.dev/gemini-api/docs/video)
- [Google GenAI Python SDK](https://github.com/googleapis/python-genai)