Veo3Gen Meta-Prompts for Consistent Brand Videos: A Copy-Paste "Prompt Builder" Template

TL;DR

A meta‑prompt is a reusable instruction set that outputs your final, Veo‑ready shot prompt. For consistent brand videos, use a 2‑layer system:

Brand DNA = non‑negotiable defaults (look, camera habits, lighting, audio policy, negatives).
Shot Builder = converts one idea into one shot prompt using a fixed schema: Camera → Subject → Action → Environment → Lighting → Style/Texture → Audio (optional), plus constraints.

This matches Veo prompt guidance: be explicit about camera framing/motion, style, lighting, character details, location, and action (https://deepmind.google/models/veo/prompt-guide/) and treat prompts like a one‑sentence shot list with constraints and negatives (https://www.visla.us/blog/guides/how-to-prompt-veo-3-and-veo-3-1/).

Key takeaways

Separate brand defaults from shot specifics so your look doesn’t drift between clips.
Force clarity with: one shot = one subject + one primary action.
Use a fixed output structure so prompts are scannable and reusable (https://invideo.io/blog/google-veo-prompt-guide/).
Control audio intentionally. Veo can generate audio (including dialogue) when prompted (https://deepmind.google/models/veo/prompt-guide/; https://replicate.com/blog/using-and-prompting-veo-3), so make “silent unless requested” a hard rule.
Keep prompts short by using defaults + deltas: Brand DNA holds the defaults; the shot prompt only states what changes.

Why your brand videos drift (and why meta‑prompts fix it)

Most inconsistency isn’t “model randomness.” It’s prompt variance:

You describe the product one way today and a different way tomorrow.
You forget to specify lighting/camera, so the model picks something new.
You accidentally ask for multiple beats (“unbox, apply, smile, walk outside”), and the generation becomes a montage.

Veo guidance rewards specific, repeatable shot language—framing and camera motion, style references, lighting descriptions, detailed characters, thorough locations, and clear actions (https://deepmind.google/models/veo/prompt-guide/). Meta‑prompts simply make that repeatable.

The 2‑layer system (use this for every campaign)

Layer 1: Brand DNA (compiled once)

Brand DNA is a compact block you reuse across every clip. It contains:

Style words (5–9 max)
Cinematography defaults (typical framing + motion)
Lighting defaults
Texture/finish defaults (e.g., crisp digital vs subtle grain)
Audio policy (what is allowed by default)
Guardrails + negatives (what must never appear)

Why it works: it prevents you from renegotiating your “house style” every time.

Layer 2: Shot Builder (run per idea)

Shot Builder takes:

your Brand DNA
one idea
constraints (aspect ratio, resolution, optional duration)
optional reference images

…and outputs a single shot prompt plus variants. The “shot list” approach is directly recommended in Veo prompting guides (https://www.visla.us/blog/guides/how-to-prompt-veo-3-and-veo-3-1/).

Copy‑paste template: Brand DNA Meta‑Prompt (fill once)

Paste this into your prompting doc, fill placeholders, and run it once to produce a compact BRAND_DNA block.

You are my Brand DNA compiler. Output a compact BRAND_DNA block to be pasted into a Veo3Gen Shot Builder.

Context
- Brand / Creator: {BRAND_NAME}
- Product or Offer: {PRODUCT}
- Audience: {AUDIENCE}
- Primary goal: {GOAL}

Brand style words (pick 5–9 max)
- {STYLE_WORDS}

Visual identity defaults
- Color palette cues: {COLORS}
- Wardrobe cues: {WARDROBE}
- Props / set dressing: {PROPS}
- Locations that feel on-brand: {LOCATIONS}

Cinematography defaults
- Framing habits: {FRAMING_DEFAULTS}
- Camera motion defaults: {CAMERA_MOTION_DEFAULTS}

Lighting defaults
- {LIGHTING_DEFAULTS}

Texture / finish defaults
- {TEXTURE_DEFAULTS}

Audio policy (IMPORTANT)
- Default: NO dialogue, NO voice-over, NO music unless explicitly requested.
- If audio is requested, keep it minimal and on-brand.

Non-negotiable guardrails
- ONE SHOT must contain: ONE subject + ONE primary action.
- Avoid cluttered scenes and background distractions.
- Do not add random on-screen text, logos, captions, or watermarks.

Reusable NEGATIVE / EXCLUSIONS snippet
- {NEGATIVES}

Output format
Return only:
BRAND_DNA:
- Brand summary (1 line)
- Defaults (bullets)
- Guardrails (bullets)
- NEGATIVE_SNIPPET (one line)

No commentary.

The fields above map directly to Veo guidance: be specific about camera, lighting, style, character details, location, and action (https://deepmind.google/models/veo/prompt-guide/).

Copy‑paste template: Veo3Gen Shot Builder Meta‑Prompt (predictable structure)

This is the meta‑prompt that writes the prompt you actually generate with.

You are my Veo3Gen Shot Builder. Convert ONE idea into ONE shot prompt using the BRAND_DNA below.

BRAND_DNA:
{PASTE_BRAND_DNA_HERE}

User inputs
- Idea: {IDEA}
- Aspect ratio: {ASPECT_RATIO} (16:9 or 9:16)
- Resolution target: {RESOLUTION} (720p, 1080p, or 4K)
- Duration: {DURATION} (optional; if not provided, omit)
- Reference images: {REFERENCE_IMAGES} (if any; otherwise "none")
- Audio requested?: {AUDIO_REQUESTED} (yes/no)
- If yes: {AUDIO_TYPE} (dialogue / VO / SFX / music / ambience) + {AUDIO_NOTES}

Hard rules
1) ONE SHOT = ONE SUBJECT + ONE PRIMARY ACTION.
2) Output must be concise and scannable.
3) Do not introduce extra characters/animals unless explicitly specified.
4) Do not add any on-screen text, logos, captions, or watermarks.
5) Only include an Audio line if Audio requested = yes.
6) Include a Negative line using NEGATIVE_SNIPPET.

Output format (exact structure)
Camera: ... (framing + motion; lens only if helpful)
Subject: ...
Action: ... (single verb phrase)
Environment: ... (location + 2–3 anchors)
Lighting: ...
Style/Texture: ... (use brand style words)
Audio: ... (ONLY if requested)
Constraints: ... (aspect ratio, resolution, duration if provided, reference images)
Negative: ...

Then provide:
- Variant A (more cinematic): only change Camera/Lighting/Style.
- Variant B (more creator-native): only change Camera/Style.

No headings. No explanation.

That field order is aligned with common Veo prompt formulas that layer camera, subject, action/physics, environment, lighting, style/texture, and audio (https://invideo.io/blog/google-veo-prompt-guide/).

Worked example (with a before/after you can copy today)

Below is a concrete “brand consistency” problem: the same product idea generates wildly different looks because the prompt is vague.

Before (typical vague prompt)

Make a short video ad for a vitamin C serum. Clean and premium. Show a morning routine. Make it look cinematic.

What goes wrong: multiple beats (“morning routine”), no camera plan, no lighting specifics, and “cinematic” is undefined.

After (meta‑prompt output: one shot, structured)

Assume your BRAND_DNA prefers minimal sets, warm even lighting, gentle push‑ins, and silent by default.

Shot prompt (paste into Veo3Gen)

Camera: Tight product close-up, steady, slow gentle push-in.
Subject: One hand holding a vitamin C serum bottle, label facing camera.
Action: The hand dispenses one glossy drop onto a fingertip (single dispense).
Environment: Minimal bright bathroom vanity; uncluttered counter; neutral towel and mirror softly out of focus.
Lighting: Warm even lighting across the frame; soft highlights on the glass.
Style/Texture: Clean, premium, minimal; crisp digital finish; subtle controlled contrast.
Constraints: 9:16, 1080p; reference images: none.
Negative: extra limbs/fingers, warped label text, text artifacts, flickering, sudden camera whip, chaotic motion.

Variant A (more cinematic):
Camera: Macro close-up, slow dolly-in, shallow depth of field.
Subject: One hand holding the serum bottle, label readable.
Action: Single dispense of a luminous drop catching the key light.
Environment: Minimal vanity; faint steam haze near mirror.
Lighting: Soft key light + gentle edge highlight on bottle.
Style/Texture: Premium beauty ad; subtle film grain.
Constraints: 9:16, 1080p; reference images: none.
Negative: extra limbs/fingers, warped label text, text artifacts, flickering, sudden camera whip, chaotic motion.

Variant B (more creator-native):
Camera: Handheld phone-feel, stable, medium close-up angled slightly down.
Subject: Creator’s hand holding the serum at arm’s length.
Action: Single dispense onto fingertip.
Environment: Realistic tidy bathroom; morning vibe.
Lighting: Bright soft window light; no harsh overheads.
Style/Texture: Clean, relatable, creator-native; crisp digital.
Constraints: 9:16, 1080p; reference images: none.
Negative: extra limbs/fingers, warped label text, text artifacts, flickering, sudden camera whip, chaotic motion.

Notice what changed:

One subject, one primary action.
Camera + lighting are stated plainly (Veo guidance explicitly calls out camera framing/motion and lighting descriptions) (https://deepmind.google/models/veo/prompt-guide/).
Audio is absent on purpose (because Veo can generate audio when prompted, you prevent surprises by defaulting to silent) (https://replicate.com/blog/using-and-prompting-veo-3).

Using Veo3Gen modes without changing your prompt system

Your meta‑prompt system stays the same regardless of mode. In Veo3Gen you can pick:

Veo 3.1 Fast (quick, great default)
Veo 3.1 Quality (max fidelity)
Veo 3.1 Lite (cheapest, preview)

Veo3Gen supports text‑to‑video and image‑to‑video, plus first‑and‑last‑frame control on Veo 3.1. Supported outputs include 720p, 1080p, and 4K (4K on Fast/Quality) with 16:9 and 9:16 aspect ratios. Generations include native, synchronized audio (dialogue/SFX/music) in a single pass when you ask for it. Pricing is pay‑as‑you‑go credits plus optional monthly plans, and purchased credits do not expire; new users get free credits. There’s also a developer API for programmatic generation.

If your goal is a consistent brand pipeline, start by generating 3–5 “anchor shots” with the templates above, then reuse the same Brand DNA across your campaign.

CTA (mid‑article): If you want to iterate quickly without rebuilding prompts every time, run this two‑layer system in Veo3Gen and switch between Fast/Quality/Lite while keeping the same structured output.

Checklist

Use this before you spend credits.

One shot: exactly one subject and one primary action.
Camera: framing + motion are explicit (and only one main move).
Environment: named location + 2–3 concrete anchors (what’s physically there).
Lighting: simple, filmable language (e.g., warm even lighting; spotlight in one area) (https://deepmind.google/models/veo/prompt-guide/).
Style/Texture: uses your fixed brand style words; no new “vibe soup.”
Audio: either explicitly requested (type + notes) or explicitly absent (Veo can generate dialogue/audio when prompted) (https://deepmind.google/models/veo/prompt-guide/; https://replicate.com/blog/using-and-prompting-veo-3).
Constraints: aspect ratio + resolution; duration only if you truly need it; reference images listed if used.
Negative line: present and reusable.

FAQ

How do I make prompts consistent across 30+ brand videos?

Use a Brand DNA block for unchanging defaults, then a Shot Builder that outputs the same schema every time. This mirrors the “shot list + constraints” approach recommended for Veo prompting (https://www.visla.us/blog/guides/how-to-prompt-veo-3-and-veo-3-1/).

How do I stop random dialogue or music?

Make “no audio unless requested” a hard rule and only include an Audio: line when you want it. Veo can generate dialogue and prompts can provide specific lines (https://deepmind.google/models/veo/prompt-guide/), and Veo audio can include dialogue/voice‑over/SFX/music (https://replicate.com/blog/using-and-prompting-veo-3).

What’s the simplest structure that still works well?

Use a fixed, scannable schema: Camera → Subject → Action → Environment → Lighting → Style/Texture → Audio (optional). This aligns with repeatable prompt formulas (https://invideo.io/blog/google-veo-prompt-guide/).

How do I keep a product or character consistent between clips?

Use reference images and keep “reference images: …” as an explicit input field. Prompting guidance notes that attaching reference images helps keep faces, products, and environments consistent (https://www.visla.us/blog/guides/how-to-prompt-veo-3-and-veo-3-1/).

Generate your first video now: Get started
Compare plans and pay-as-you-go pricing: See pricing