FoodReel — the spec

FoodReel is a social media ad video generator built for solo food business owners who need professional-quality content without a production team. A user uploads still photos of their food, setup, or location, writes a prompt describing their vision, and gets a photorealistic 15–30 second ad video ready to post on TikTok, Reels, or YouTube Shorts — no editing skills required.

---

Decisions locked

---

The core loop

1. User creates an account and starts a new project for a product or campaign.
2. User bulk-uploads still photos (product shots, location, setup); the system auto-organizes them by detected product group.
3. User opens the prompt builder, selects a food-video template (e.g. "sizzling close-up", "behind-the-counter story"), and writes or refines a text prompt.
4. FoodReel sends the photo set and prompt to the video generation engine and produces a photorealistic 15–30 second ad.
5. User reviews the rendered video, downloads it in their target format (TikTok 9:16, Reels 9:16, YouTube Shorts 9:16/16:9), and publishes.
6. Performance data from published videos feeds back into the analytics dashboard, informing the next prompt.

---

v1 scope

Bulk photo upload with auto-organization by product
Accepts multiple image files in one pass and groups them by product using visual similarity. Lives on the Project workspace screen.

Text prompt builder with food-video templates
A structured prompt interface with pre-built templates tuned for food content (close-up sizzle, ambient dining, prep sequence, etc.). The user fills in or overrides fields; the final prompt string is passed directly to the generation engine. Lives on the Prompt Builder screen.

Photorealistic video generation from still images
The generation engine animates the uploaded stills into a fluid, realistic 15–30 second video using the prompt as the creative brief. Renders within the project and is viewable on the Video Review screen.

Download in TikTok / Reels / YouTube Shorts formats
One-click export at the correct resolution and aspect ratio for each platform. No manual re-encoding. Lives on the Video Review screen alongside the rendered output.

Video performance analytics dashboard
Tracks views, engagement rate, and watch time for published videos. Surfaces which prompts and photo sets produced the best-performing ads. Lives on the Analytics screen.

---

Deliberately later

• Real-time preview before final render — render infrastructure cost and latency need to be validated at scale before adding a preview tier; v2 once usage patterns are established.
• Style presets (moody, bright, ASMR, storytelling) — templates in v1 prove which aesthetic directions users actually reach for; presets are built from that real data in v2.
• Auto-music and caption suggestions — licensing and sync logic add meaningful legal and technical complexity; ship after the core video loop is stable.

---

Data model sketch

• users — id, email, password_hash, plan_tier, created_at
• projects — id, user_id, name, created_at, updated_at
• photos — id, project_id, file_url, product_group_label, upload_order, created_at
• prompts — id, project_id, template_id, prompt_text, created_at
• templates — id, name, category, base_prompt_structure, is_active
• videos — id, project_id, prompt_id, render_status, video_url, duration_seconds, created_at
• exports — id, video_id, format (tiktok/reels/shorts), file_url, downloaded_at
• analytics_events — id, video_id, platform, views, watch_time_seconds, engagement_rate, recorded_at

---

Screens

• Dashboard — lists all projects with thumbnail preview and last-activity date.
• Project workspace — displays the uploaded photo library organized by product group; entry point for new uploads.
• Prompt builder — template selector plus free-text prompt editor; confirms the photo set and fires the generation request.
• Video review — shows the rendered video, duration, and download buttons per platform format.
• Analytics dashboard — aggregated and per-video performance metrics tied back to the originating prompt and photo set.
• Account / settings — plan management, connected platform tokens, notification preferences.

---

How to use this document

• Bring it into Mind Palace, where the guided platform and its AI coach pick up from exactly this document and build FoodReel with you step by step.
• Hand it to a developer as the complete v1 brief — scope, data model, and screens are defined; nothing is left open.
• Keep it as the single source of truth that resolves any scope disagreement during the build.