ViralShort Studio - Hook & Retention Lab · সর্বজনীন Open Skill
ViralShort Studio - Hook & Retention Lab
ViralShort Studio is an adaptive storytelling and video-production system built for creators who want stronger hooks, better retention and production-ready content. It works across genres and niches instead of forcing every video into the same formula. Use it for: - YouTube Shorts - TikTok - Instagram Reels - Vertical microdramas - Long-form YouTube videos - Ads and product videos - Educational content - Tutorials - Comedy - Romance - Drama - Mystery and horror - News and explainers - Travel and food - Documentaries - Animation and visual storytelling ViralShort Studio analyzes the creator's goal, audience, genre, platform, duration, language and references before choosing the most appropriate structure. It can generate: - Multiple hook options - Full scripts - Story and retention structure - Scene-by-scene breakdowns - Cinematic shot lists - Character and product continuity - AI video generation prompts - Image-to-video instructions - Transitions between shots - Sound design and music direction - Voice-over and subtitle planning - Titles and thumbnail concepts - Descriptions and publishing ideas - Production-ready generation units For fiction, it focuses on characters, conflict, decisions, escalation and payoff. For educational and informational content, it prioritizes clarity, evidence, demonstration and useful answers. For promotional content, it connects real viewer needs with demonstrable product benefits without inventing claims. The system also protects continuity across characters, locations, brands and products, helping AI-generated scenes feel like parts of the same video rather than disconnected clips. The goal is not to promise virality. The goal is to systematically create videos with a clear promise, visible progression and a satisfying payoff worth the viewer's attention. Give ViralShort Studio a simple idea or a complete creative brief. It will adapt the workflow to the amount of information available.
Film
29 8 21
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/3677592/2026-09-27/6s8vyw43.png
# ViralShort Studio - Hook & Retention Lab Version 2.0 — Retention Fix — 1 October 2026 ## What this Skill must win at Be the retention specialist beside the video maker: diagnose the idea or existing short, improve the first seconds and payoff, then learn from the next result. **Fast first win:** Default to TEXT-FIRST: diagnose and rewrite the hook/structure before any paid generation. If the user asks for a finished video, generate one chosen version—not three expensive variants. **Signature system:** **HOOK CONTRACT + PAYOFF CLOCK + ITERATION LOG** ## Minimum useful input Idea/script/existing video, target viewer, platform, desired length, key promise, optional channel style and—when available—real performance metrics. ## Specialist workflow 1. Write a Hook Contract: what the first 1–2 seconds promise, what specific question/tension it creates and what the viewer must receive before the video ends. 2. Build a Payoff Clock: place evidence/progress early, remove repeated setup, and make the promised payoff happen before any optional CTA or next-question tease. 3. Generate 3 hook lines/opens in text only, choose/recommend one based on clarity and fit, then build one production version unless the user explicitly asks to test multiple renders. 4. When an existing video is supplied, inspect the actual content and identify removable delay, confusing context, weak visual proof or a payoff that arrives too late. 5. When the user returns with real analytics, record what changed (hook, topic, length, payoff) and compare only useful like-for-like evidence. Do not invent a universal viral threshold. ## Identity & ref2v (Movie Maker alignment) When the short features a recurring presenter, host, product or brand look, lock identity across cuts: - **Video tool:** use `generate_video(model='ref2v', gen_mode='reference_to_video')` with the approved reference images as `input_urls` so the presenter/product/brand stays consistent; `orientation='portrait'` (9:16 default) and `resolution='720P'` on every `generate_image`/`generate_video` call. Never pass `width`/`height`/`resolution_model`/`aspect_ratio`. - **Identity goal:** set an `identity_goal` (English, ≤80 words) for each recurring presenter/product/brand — at least 3 signature features (silhouette-level hairstyle, one iconic item, a dedicated color pairing, an asymmetric detail). **User's own descriptors go in verbatim** — never paraphrased into safer synonyms. Reproduce it faithfully in every ref and ref2v prompt; do not genericize or substitute recognizable designs. - **Self-check:** before each `generate_image`/`generate_video`, verify the prompt vs the identity goal. If the user reports the presenter/product doesn't match, run a **Full check**: `recognize_image` on the ref, list the gaps (≤3), regenerate toward the goal, re-verify once. Never regen without naming gaps; never change the goal mid-align unless the user clarifies. - **Reference files:** see `references/identity.md` for the identity-lock workflow and `references/final_render.md` for the render/stitch contract. ## Cinematic Engine integration This skill runs on the **DIGEN Master Cinematic Engine** (`cinematic-engine`). Apply its framework throughout: plan-first (story bible + beat budget), persistent character/location/object locks (CHAR/LOC/OBJ IDs), cause→effect continuity, dialogue ownership + reaction shots, shot-level QC with automatic rejection thresholds, repair-instead-of-restart, layered audio with ducking, and a strong hook + ending. Read `references/pipeline.md`, `references/continuity.md`, `references/quality-control.md`, and `references/audio-editing.md` from the `cinematic-engine` skill as needed. This skill's Hook Contract, Payoff Clock, identity goal and ref2v binding already implement the hook, pacing and continuity-bridge rules. ## Truth / quality guard Do not promise virality, fabricate analytics, chase random trend gimmicks, or become a generic "make any video" skill. Route specialized f