skills/marketing-assistant/references/poster.md

# Create a Poster

When the user requests a poster, create and deliver an actual poster image, not only copy, a brief, or an image prompt.

Never jump directly from the request to image generation. Follow these stages in order and ask only one question in each message.

## Stage 1 — Select the hook

Use the confirmed profile and the poster goal. If the topic, offer, date, or occasion is essential but unclear, ask for that one detail first.

Read and apply [ddd-hooks.md](ddd-hooks.md). Suggest exactly three hooks that pass its DDD and Golden Rule gate. Give each hook exactly one word explaining its main psychological driver.

Ask the user to select 1, 2, or 3. Do not create the visual story or poster before selection.

## Stage 2 — Offer three visual stories

After hook selection, turn the selected hook into exactly three meaningfully different visual-story options. Keep each option to a few vivid lines. Do not describe a poster, layout, typography, text placement, or graphic design. Describe only what happens inside the image as if it were a movie scene.

Each story must specify the person, visible emotion/action, environment, lighting, camera angle, and uncluttered background naturally in prose. Make every option ultra-realistic and physically believable: natural skin texture, authentic expressions, realistic hands and anatomy, true-to-life fabric and materials, plausible lighting, and a candid human moment. Reject plastic-looking skin, doll-like faces, robotic poses, stiff symmetry, exaggerated beauty filters, glossy CGI, or artificial stock-photo expressions.

Use this compact structure:

> **Create the Visual Story**
>
> **Selected hook:** [selected hook]
>
> 1. **[Short story name]** — [cinematic story in a few lines]
> 2. **[Short story name]** — [cinematic story in a few lines]
> 3. **[Short story name]** — [cinematic story in a few lines]
>
> **Which visual story do you select? Reply with 1, 2, or 3.**

Do not create an image yet. After the user selects one story, restate only the selected story and ask:

> **Do you approve this visual story, or would you like another story?**

Continue only after approval. If the user wants another, create three new story options for the same selected hook.

## Stage 3 — Boutique dress and model flow

Use this stage only for a boutique, fashion, clothing, tailoring, or dress-product poster.

1. Ask whether the user wants to upload a dress/product photo so the model can wear that specific dress, or prefers a dress recommended from the confirmed product details.
2. If a dress photo is supplied, preserve its actual design, colour, pattern, neckline, sleeves, fabric character, embellishments, and length. Do not invent or replace important garment details.
3. Before creating the model image, recommend exactly three natural poses matched to the approved story, dress, target customer, and poster space. Briefly explain the benefit of each pose, then ask the user to select one or request another recommendation.
4. Ask whether they have any additional recommendation for the model, styling, pose, modesty, setting, or appearance. Treat “no” as approval to continue.

Never generate the model photo before the dress choice, pose choice, and recommendation question are resolved.

## Stage 4 — Collect poster assets

Ask these separately, skipping only an asset already supplied in the current conversation:

1. Ask the user to upload their logo or reply that they do not want to use one.
2. Ask the user to upload any reference picture or reply that they do not want to use one.
3. Recommend one simple, hook-matched **DM** CTA using a single easy keyword, for example `DM THAILAND`, `DM TREK`, or `DM BRIDAL`. Ask the user to approve it or provide another keyword.

Do not ask for or print a WhatsApp number in the poster workflow. Do not use “WhatsApp to enquire,” “WhatsApp Enquiry,” or “Enquire Now.” Keep the visible CTA short: `DM [KEYWORD]`.

## Stage 5 — Final confirmation and poster generation

Before generating, summarize the approved hook, selected visual story, boutique dress/pose choices when applicable, logo/reference assets, headline, essential offer details, and DM CTA. Ask for final confirmation. Do not generate until the user confirms.

After confirmation:

1. Prepare concise copy: one headline based on the selected hook, the main benefit, essential details, and the approved `DM [KEYWORD]` CTA.
2. Use supplied assets accurately; never claim to include an unavailable asset.
3. Create a clean, modern, mobile-first design suitable for WhatsApp, Instagram, or Facebook.
4. Generate an ultra-realistic image with natural skin, believable emotion, correct anatomy, realistic hands, authentic fabric texture, plausible lighting, and a natural non-robotic pose. Avoid plastic skin, mannequin appearance, stiff posture, glossy CGI, over-smoothing, or artificial expressions.
5. Use the available image-generation capability to generate and deliver the actual image.
6. Inspect spelling, prices, dates, DM CTA, readability, realism, garment fidelity, and consistency before delivery. Correct material errors when possible.

If image generation is unavailable, explain that plainly and provide ready-to-use copy plus a precise design brief. Never pretend an image was created.

After delivering an actual poster, ask only:

> **Do you want a Flow video prompt to animate this poster?**

If yes, create a concise image-to-video prompt that preserves the poster's message and describes natural subject movement, camera movement, atmosphere, and timing. Do not add new claims or change visible brand details. After delivering it, show the main menu. If no, show the main menu.

Do not request personal identity, a testimonial, or permission to save feedback.