R50 — Cinematic Adverts System

Create cinematic video ads with AI using a unique dual-frame approach

Category: Content Creation Runtime: n8n AIOS Rating: 7/10

---

What It Does

Generates cinematic video advertisements using a unique dual-frame approach. Instead of generating a single image and animating it, R50 generates both a start frame and an end frame, then uses Veo 3's FIRST_AND_LAST_FRAMES_2_VIDEO mode to create smooth cinematic transitions between them. Includes AI music generation via Suno V5.

Four independent schedule triggers handle prompts, images, videos, and music — each polling Airtable for records at the correct status.

Architecture

TRIGGER 1: PROMPTS (polls Airtable)
Product brief + reference  →  OpenRouter writes  →  Start + end frame  →  Status = "Prompts Ready"
                                prompts (SEALCaM)    prompts per scene

TRIGGER 2: IMAGES (polls Airtable)
Status = "Prompts Ready"  →  NanoBanana Pro  →  Start frame + end frame  →  Status = "Images Ready"
                              (WaveSpeed)        per scene

TRIGGER 3: VIDEOS (polls Airtable)
Status = "Images Ready"  →  Veo 3 (Kie.ai)  →  Cinematic transitions  →  Status = "Video Ready"
                             FIRST_AND_LAST_FRAMES mode

TRIGGER 4: MUSIC (polls Airtable)
Status = "Video Ready"  →  Suno V5 (Kie.ai)  →  Background music  →  Status = "Complete"

The Dual-Frame Approach

This is what makes R50 unique:

Approach	How It Works	Result
Single-frame (R51, R54)	Generate 1 image → animate it	Camera/subject movement only
Dual-frame (R50)	Generate start + end images → Veo 3 interpolates	Smooth cinematic transitions between two compositions

The dual-frame method produces more cinematic results — think camera dollies, reveal shots, and dramatic transitions — because the AI knows where the scene starts AND where it needs to end.

Tech Stack

Platform	Role	Cost
OpenRouter	LLM gateway for prompt writing	Per-token
WaveSpeed AI	NanoBanana Pro for start + end frame images	Per-generation
Kie.ai	Veo 3 (dual-frame video) + Suno V5 (music)	Per-generation
Airtable	Pipeline state machine	Free tier
n8n	Workflow automation — 4 schedule triggers	Self-hosted

Project Structure

R50/
├── R50 _ Cinematic Adverts System.json    (n8n workflow)
└── Core Inputs/
    ├── core image.jpg                      (product reference)
    ├── core elements.png                   (brand elements)
    └── elements board inputs/              (mood board images)
        ├── 1.jpeg
        ├── 2.png
        ├── 3.jpg
        └── 4.jpg

Environment Variables

Variable	Source	Purpose
OPENROUTER_API_KEY	openrouter.ai	Prompt writing
WAVESPEED_API_KEY	wavespeed.ai	NanoBanana Pro images
KIE_API_KEY	kie.ai dashboard	Veo 3 video + Suno V5 music
AIRTABLE_API_KEY	airtable.com/create/tokens	Pipeline state
AIRTABLE_BASE_ID	Airtable URL	Target base

How to Use

1. Import the workflow JSON into your n8n instance
2. Configure credential nodes with API keys
3. Set up Airtable base with scene tracking
4. Add product images and brief to Core Inputs
5. Create Airtable row and set status to start the pipeline

Common Pitfalls

Problem	Cause	Fix
Veo 3 transition looks wrong	Start/end frames too different	Keep compositions related — same subject, different angle/position
Music doesn't fit	Generic prompt	Be specific with genre, tempo, mood in Suno prompt
Images inconsistent	No style reference	Include mood board images in Core Inputs for consistent aesthetic

---

Built by Copyweb — architecting how marketing teams operate in an AI-first world.