Subscribers enjoy more affordable credit package prices
Everything you need to know about structured JSON prompting for Veo 3.1 & Sora 2
Veo 3.1 offers surgical precision for brand-safe, predictable outputs. It's ideal for structured content like product videos, ads, and corporate messaging where consistency matters. Sora 2 provides higher creative variance with diverse visual styles—perfect for experimental content, artistic projects, or exploring multiple aesthetic directions. Both use the same JSON prompt structure, so you can switch models without rewriting. Pro tip: Use Sora for rapid A/B testing of creative concepts, then lock in final versions with Veo for consistent batch production.
Choose Veo 3.1 when you need: (1) Brand consistency across 10+ videos, (2) Predictable outputs for client approvals, (3) Tight control over camera, lighting, and audio specs. Choose Sora 2 when you want: (1) Creative exploration with varied visual styles, (2) Quick concept testing before finalizing direction, (3) Artistic or experimental content with higher visual diversity. Many users combine both: Sora for ideation phase (2-3 variations in 5 minutes), Veo for production phase (locked brand assets at scale). Both models share the same JSON structure, so your template library works across both.
Text prompts = creative lottery. JSON prompts = predictable results. With text prompts, you're hoping AI guesses your intent correctly. Most pros need 10+ generations to get one usable shot. With JSON prompts, you organize by structure: shot sequence, camera (lens, movement), lighting specs, audio layers. The AI follows exact instructions, not vague descriptions. Real impact: Agencies report going from 10 attempts per shot to 2 attempts — that's 80% faster with 5x cost savings. Think storyboards (structured) vs search queries (vague).
If you've used a form or spreadsheet, you already get the concept. JSON prompting is just filling in labeled fields: shot (what's happening), camera (close-up, wide, tracking), lighting (golden hour, studio), audio (dialogue, ambient sound). Start here: (1) Pick a pre-built template for product video, ad, or vlog (2) Change 2-3 fields like product name or color palette (3) Generate and see results in 60 seconds (4) Tweak one field at a time to learn what each does. Most users create custom prompts within 15-20 minutes. No coding required — our visual editor builds the JSON for you. Advanced users write JSON directly for maximum control.
Yes — that's the whole point of structured prompts. One JSON template creates unlimited variations. Same structure, different products: Keep shot sequence, camera moves, lighting. Change product name, color palette, brand voice. Result: 10 product videos with consistent style in 30 minutes. Same content, different formats: Keep subject, action, audio. Change aspect_ratio field (16:9 to 9:16 to 1:1). Result: YouTube, TikTok, Instagram versions from one prompt. Same brand, different campaigns: Save your brand's camera style, color grading, audio signature as a template. Duplicate and tweak message only. Result: Every video stays on-brand automatically. This is why agencies love JSON — one template serves 50+ clients with minimal edits.
Lock the variables, change only what you want different. For character consistency: Define once in JSON (physical features, speaking style, wardrobe), then copy-paste these fields across all prompts for the same character in every video. For brand consistency: Create a brand kit JSON template with fixed camera settings (50mm lens, eye-level), color palette, audio style (upbeat corporate, 95 BPM), and duration. Pro tip from Reddit users: Change one field per generation. If you change 5 things, you won't know what caused the improvement. Change camera angle only, test, then change lighting next, test again. Result: From 10 attempts to 2 attempts per usable shot.
Yes. Our API and integrations plug into your current stack. Popular workflows: CMS triggers (product added to Shopify auto-generates JSON, renders video, posts to Instagram), Spreadsheet batch (upload 100 product rows, each becomes a JSON prompt, queue renders overnight), n8n or Zapier (webhook receives order, generates thank-you video with customer name, emails automatically), API for dev teams (POST JSON payload, receive video URL, embed in your app). You can automate: template library sync, brand kit application, multi-format export (16:9, 9:16, 1:1 from one JSON), trend monitoring. Real use case from Reddit: Built a Telegram bot where you say 'Make me a Rolex ad' and the workflow generates JSON, Veo 3.1 renders, video delivered in 2 minutes.
Yes — this is called reverse engineering, and it's one of the most powerful features for creators. How it works: (1) Upload or link to any video from TikTok, Instagram Reel, YouTube (2) AI analyzes camera angles, lighting, motion, audio, pacing (3) Generates JSON with structured breakdown of every element (4) You edit to swap product or subject while keeping the viral structure (5) Render your version in minutes. Real Reddit example: User found a viral coffee ASMR video (3M views), uploaded it, got the JSON, changed product to smoothie bowl, generated 10 variations, posted one that hit 500K views in 2 days. You can extract: shot sequence (0-2s close-up, 2-5s pour motion, 5-8s reveal), camera specs (35mm lens, handheld, slight dolly), color grading (warm tones, high contrast), audio structure (ASMR tapping, ambient café sounds). Use cases: competitor analysis, trend replication, style matching. Note: Only works with AI-generated or copyright-free videos.
Preview with Veo 3.1 Fast: 60-90 seconds. High quality with Veo 3.1: 2-4 minutes. First-time workflow: Pick template (30 seconds), customize fields (2-3 minutes), generate (60 seconds), review and tweak (1-2 minutes). Total: 5-7 minutes for your first production-ready video. After you've got templates: Duplicate template (10 seconds), change 2-3 fields (1 minute), generate (60 seconds). Total: 2-3 minutes per video. Batch workflows with API: Upload 100 product JSONs (5 minutes), queue renders automatically, all 100 videos done in 2-3 hours overnight. Per-video time: about 2 minutes each. Compare to traditional: Script plus storyboard (2 hours), shoot plus lighting setup (4 hours), editing plus color grading (6 hours). Total: 12+ hours plus equipment costs. Reddit quote: I used to shoot $500K pharma commercials. I made this for $500 in Veo 3.1 in less than a day.
For professional creators.
Advanced features for professional teams.
Enterprise Service
1600 credits for batch creation
800 credits for batch creation
300 credits for batch creation
