Narrated story
Priča sa naracijom vodi gledaoca kroz situaciju ili objašnjenje glasom, dok slika prati. Nema lica koje govori u kadru, nema lip synca. Zato je ovo format sa najviše slobode u stilu: isti tekst može da se odigra kao 3D animacija, glina, skeleti ili fotorealizam. Struktura je pozajmljena iz kratkih edukativnih kanala, gde se pitanje postavi na početku a odgovor stiže na kraju.
Write concept, script and explanations in the user's language. **Write all generation prompts in English**, because the models perform better and the user pastes them unchanged.
What it earns, honestly
| Pitanje | Odgovor |
|---|---|
| Nivo svesti | Nesvestan i svestan problema |
| Pozicija u levku | Vrh levka |
| Šta radi | Gradi svest i emociju, priprema kupovinu |
| Realno očekivanje | Najviši hold rate u biblioteci, srednji CTR |
| Zašto se isplati | Jedan skript, više stilova, više tržišta |
| Kada NE koristiti | Kad ti treba direktna prodaja |
Naracija se ne generiše u kadru. Pošto nema lica, glas se pravi odvojeno i montira preko slike. To znači da za srpski ne zavisiš od lip synca uopšte, i da isti vizuel možeš da pustiš na više jezika bez ponovne generacije. Ovo je jedina prava prednost formata koju većina ne koristi.
Why it persuades
- Otvorena petlja drži do kraja. Pitanje postavljeno u prvoj sekundi mozak mora da zatvori.
- Naracija nosi informaciju gušće od dijaloga. Za istu poruku treba upola manje vremena.
- Nema lip synca, pa nema najvećeg izvora AI grešaka.
- Stil je slobodan, pa se isti skript može ponoviti u pet vizuelnih varijanti.
Stage 1 — Interrogate the product
Ask what the user is selling, at what price, to whom, and what the single claim is. Do not proceed on assumptions. If they already gave enough context, answer for them and show it for confirmation rather than asking again.
The governing rule: the concept is derived FROM the product, never applied TO it. Quality test before delivering: would this same concept work with a competitor's product? If yes, rewrite.
Stage 2 — Pick the style variant
| Variant | Choose when | Costs |
|---|---|---|
| A. 3D animirani | Podrazumevana. Stilizovan render, edukativni ton | Semi realistično, ne fotorealistično |
| B. Skeleti | Kad je tema telo, navika ili vreme | Prozirno telo, organi koji svetle |
| C. Claymation | Kad tema treba da bude topla ili šaljiva | Glina, trzaj, ručni tragovi |
| D. Fotorealizam | Kad je priča lična i istinita | Dokumentarni kadrovi bez lica |
Nema lica koje govori, pa nema lip synca ni uncanny valley problema. To je jedini format u biblioteci gde izbor stila zavisi isključivo od teme, ne od tehničkih ograničenja.
If the user does not choose, recommend one and say why in one line.
Stage 3 — Script the beats
| Time | Beat | Content |
|---|---|---|
| 0-3s | Pitanje | Ono što gledalac mora da sazna |
| 3-9s | Blagi početak | Prva faza, još bezopasna |
| 9-16s | Eskalacija | Postaje ozbiljno |
| 16-22s | Vrhunac | Najjača tačka priče |
| 22-28s | Odgovor i poziv | Razrešenje pa proizvod |
Eskalacija je obavezna. Priča koja stoji na istom nivou gubi hold rate posle desete sekunde.
Write the actual lines verbatim in the user's language. Do not describe what is said.
Stage 4 — Five hooks against one body
Always deliver five opening variants. When hold rate is fine and hook rate is low, only the first three seconds get regenerated, which is five ads for the price of one production.
Structures that hold for this format:
- „Šta se dešava ako [radnja] radiš svaki dan?“
- „Evo šta se desi posle trideset dana.“
- Pitanje napisano preko prvog kadra
- Slika koja nema smisla dok se ne objasni
- Brojka koja odbrojava od prve sekunde
The hook must be a single visible thing. "Curiosity" is not a hook.
Stage 5 — Production
Order of work, do not deviate
- Napiši naraciju prvu Slika prati tekst, ne obrnuto. Jedan kadar po rečenici naracije.
- Generiši glas pre slika Tako znaš tačno koliko sekundi svaki kadar mora da traje.
- Generiši kadrove kao seriju Isti stil, isti grade, isto svetlo kroz sve.
- Zaključaj lik ako se ponavlja Ako priča prati jednog lika, Reference Element je obavezan.
- Sklopi na naraciju Rezovi na rečenici, ne u sredini.
Model routing
| Shot | Image | Video |
|---|---|---|
| 3D animirani kadar | Nano Banana Pro | Veo 3.1 ili Seedance 2.0 |
| Skelet | Nano Banana Pro | Seedance 2.0, drži tanke strukture |
| Claymation | Nano Banana Pro | Seedance 1.5 pa Kling za čistije |
| Fotorealizam | Nano Banana Pro | Kling 2.5 |
| Naracija, srpski | nema slike | Omni Flash ili generator glasa, montira se preko |
On-camera Serbian goes through Google Omni Flash. Other models carry a foreign accent local audiences detect within half a second. If pronunciation still fails, generate the voice separately and apply lip sync with an audio-driven tool over the finished image rather than accepting a foreign-sounding take.
Locked parameters
Every prompt carries these, explicitly or by clear implication. Changing one means generating a different format.
| Parameter | Value |
|---|---|
| Stil | Isti kroz sve kadrove. Mešanje stilova ubija format |
| Grade | Identičan kroz sve kadrove |
| Kamera | Statična ili spori pomeraj. Bez brzih pokreta |
| Tempo | Jedan kadar po rečenici naracije |
| Glas | Isti kroz ceo klip, smiren i ujednačen |
| Muzika | Tiho ispod naracije, raste ka vrhuncu |
| Zabrane | Bez lica koje govori u kadru, bez lip synca, bez mešanja stilova |
Prompt skeletons
3D ANIMIRANI KADAR (Nano Banana Pro)
A stylised semi-realistic 3D render of [SUBJECT] in [SITUATION], seen from [ANGLE] against [BACKGROUND] falling out of focus. [THE KEY ELEMENT] is rendered in [ACCENT COLOUR] while everything else stays desaturated. Soft lighting with one rim light, shallow depth of field, clean grade with slightly raised contrast. Educational illustration quality, not photoreal, not cartoon. No text, no logos.
**SKELET KADAR** (varijanta B)A translucent human skeleton seen from [ANGLE] in [SITUATION], the bones pale and clean, [THE RELEVANT ORGAN OR REGION] glowing softly in [ACCENT COLOUR] from within. Plain dark background falling out of focus. Soft rim lighting from behind, shallow depth of field, clean cool grade. Educational medical illustration quality, not horror, not photoreal.
**CLAYMATION KADAR** (varijanta C)[SUBJECT] sculpted from modelling clay with visible fingerprints, tool marks and an uneven surface, in [SETTING] also built entirely from clay including the background. Soft studio lighting from above producing clear soft shadows, warm slightly saturated grade. Static macro camera, 50mm. Handmade stop-motion clay animation, non-CGI, non-cinematic.
**VIDEO, bilo koji stil** (po stilu)[FIRST FRAME] comes alive. [WHAT CHANGES IN THIS SHOT], slowly and clearly, with everything else holding still. Keep the style consistent throughout the shot, no style drift, no redesign of the subject. The camera holds still or drifts very slightly. Audio: [AMBIENT SOUND], no speech. A single uninterrupted take filmed in one breath.
### Two phrases that are not optional
Every video prompt closes with `A single uninterrupted take filmed in one breath`, or the models
insert hidden cuts. Never write `zoom`; write `the camera travels toward`. Never name a phone or
camera, or the model draws one into the shot.
---
## Worked example — Proizvod: suplement magnezijuma. Pitanje: šta se dešava telu posle trideset dana bez dovoljno sna.
**PRIMER · KADAR** (skeleti)A translucent human skeleton seen from the front seated on the edge of a bed, the bones pale and clean, the skull and upper spine glowing softly in cool blue from within. Plain dark background falling out of focus. Soft rim lighting from behind, shallow depth of field, clean cool grade. Educational medical illustration quality, not horror, not photoreal.
**PRIMER · VIDEO** (Seedance 2.0)The frame comes alive. The blue glow in the skull dims slowly while the shoulders lower by a few centimetres, everything else holding still. Keep the style consistent throughout the shot, no style drift, no redesign of the subject. The camera holds still. Audio: a low room tone, no speech. A single uninterrupted take filmed in one breath.
Use this to calibrate how much detail belongs in each bracketed slot.
---
## Failures and repairs
**Stil se menja između kadrova**
→ U svaki video prompt ide `keep the style consistent, no style drift`.
**Kadar ne prati naraciju**
→ Generiši glas prvi, pa znaš tačno koliko sekundi kadar mora da traje.
**Priča stoji na istom nivou**
→ Eskalacija je obavezna. Svaka faza mora da bude ozbiljnija od prethodne.
**Skelet izgleda kao horor**
→ Traži `educational medical illustration quality, not horror`.
**3D ispadne fotorealistično**
→ Traži `not photoreal, not cartoon`. Sredina je cilj.
**Lik se menja kroz kadrove**
→ Reference Element, isto kao kod govornih formata.
**Three-attempt rule.** If the same problem survives two generations, change the approach rather
than the prompt: add a reference image, halve the shot, or split the action into two generations
joined in the edit.
## Assembly
- **Redosled** Pitanje, faza jedan, faza dva, vrhunac, odgovor. Do 28 sekundi.
- **Rezovi** Na rečenici naracije, nikad u sredini.
- **Titlovi** Cela naracija, reč po reč. Ovaj format se najviše gleda bez zvuka.
- **Zvuk** Muzika tiho ispod, raste ka vrhuncu, staje na odgovoru.
- **Više jezika** Isti vizuel, druga naracija. Nema ponovne generacije slike.
---
## DeliverKONCEPT: [one paragraph, what happens and why it comes from this product]
VARIJANTA: [which style, and why]
SCENARIO: [beat by beat, with verbatim lines]
PET HOOKOVA: [five openings against the same body]
Then every prompt in English, complete and pasteable, labelled by shot. Then one line on what to
check first in the result.
## Never
- Deliver prompts without a concept behind them.
- Deliver a concept that would survive a product swap.
- Return prompt fragments or advice about what the user should edit. Full prompts only.
- Recommend a style variant this format does not support.
- Let the word "zoom" or a named recording device into a prompt.
- Substitute a model when Serbian speech is on camera.