> ## Knowledge Base Index
> Fetch the complete knowledge base index at: https://help.seelab.ai/sitemap.xml
> Use this file to discover available pages before exploring further.
> Pure-Markdown content can be obtained by appending a '.md' suffix to the content URLs listed in the sitemap (without the trailing slash).

# Create and adapt an ad: from product visual to animated story

![From product visual to formats](https://storage.crisp.chat/users/helpdesk/website/-/6/8/0/3/6803ea00be5f5800/ad-en-flow_iywpyq.jpg =1000xauto)

A display or social [ad campaign](https://www.seelab.ai/en/use-cases/ads) is rarely a single image. You need the visual, its headline and its button, then a dozen formats, and often an animated version. This guide follows one ad from start to finish, for the bakuchiol serum of the fictional brand Axène: created, titled, adapted and animated in Seelab.

# The journey at a glance

| Step | Where | What you get | Observed credits |
| --- | --- | --- | --- |
| 1. The product visual | **Creation**, with your Seelab model | The product in a scene, with free space for the text | 4 per image (1:1, 2K) |
| 2. Text and CTA | **Creation**, with GPT Image 2.6 | The finished ad visual: headline, tagline, button | 8 per image (1:1, 2K) |
| 3. The formats | **Workflow**, **Ads Generator** node | One file per format, pixel-perfect | 7 per format |
| 4. The animation | **Video**, Gemini Omni Flash 1.2 | An animated story, with sound | 25 for 6 seconds (720p) |

# 1. Create the product visual

Everything starts with your product, exact. In **Creation**, pick your Seelab model in the model selector, under **My custom models**: here, **Axene - Bakuchiol Serum**. The model guarantees the bottle, the logo and the label; the prompt only describes the scene.

The advertising reflex: **reserve the space for the text right away**. Say where the product goes and where the headline will sit.

> The Axène Bakuchiol serum bottle standing on a pale travertine block on the right third of the frame, soft golden morning sunlight casting long leaf shadows on a warm beige plaster wall, a few green leaves and seeds scattered at its base. Generous clean empty wall space on the left half of the image for a headline. Premium skincare advertising photography, sharp label, natural colors.

**A few tips:**

* **Choose a versatile starting format**, such as 1:1: it then adapts easily to both vertical and horizontal formats.
* **Keep the text area simple**: a plain wall, a sky, a smooth surface. The calmer it is, the more readable the text stays when the layout changes from one format to the next.
* **Generate several versions** and keep the one whose composition gives the message the most room.

In detail: [Using your model in image and video](https://help.seelab.ai/en/article/using-your-model-in-image-and-video-15ag1ou/).

# 2. Add the text and the CTA

Is the visual approved? It becomes the reference for the next step.

1. **Set it as a reference**: in the feed, open the image menu and choose **Reuse image**. It becomes image 1.
2. **Switch to a model that writes cleanly**: **GPT Image 2.6 Sunburst**, or **Nano Banana 2.1**.
3. **Write the text in quotes**, with its position, its typeface and its color, and ask to keep the photo as it is.

![The prompt, with the visual as image 1](https://storage.crisp.chat/users/helpdesk/website/-/6/8/0/3/6803ea00be5f5800/ad-en-ui-prompt_c0g8fo.jpg =1000xauto)

> Turn image 1 into a finished advertising visual. Keep the photo, the bottle, its label, the stone and the lighting exactly as they are. In the empty wall space on the left, add the headline "Retinol, the plant way." on two lines in an elegant thin serif font, dark olive green. Below it, the line "BAKUCHIOL SERUM" in small spaced uppercase sans-serif letters. Below, a rounded rectangular button in dark olive green with the white text "Shop now". Clean, minimal, premium skincare ad layout with generous margins. All text must be spelled exactly as written.

![Before, after](https://storage.crisp.chat/users/helpdesk/website/-/6/8/0/3/6803ea00be5f5800/ad-en-ba_19yssj9.jpg =1000xauto)

The bottle and its label stayed intact; the headline, the tagline and the button settled into the space planned for them.

**A few tips:**

* **Short text works better**: a headline, a tagline, a button. Keep the rest for the landing page.
* **Check the spelling by zooming in**, especially accents and small characters.
* **One version per language**: only change the text in quotes, the rest of the prompt stays the same.
* **Your brand typography must be exact to the letter?** Keep the text-free visual from step 1 and set the text in your layout tool.

# 3. Adapt the formats with the Ads Generator

A square visual is not enough for a campaign: you need a skyscraper, a half page, a mobile interstitial… In **Workflow**, the **Ads Generator** node recomposes your visual for each format, instead of cropping it.

![The workflow: reference image, generator, output](https://storage.crisp.chat/users/helpdesk/website/-/6/8/0/3/6803ea00be5f5800/ad-en-ui-workflow_lfmwfc.jpg =1000xauto)

1. **A Reference Image node** (1) with your ad visual, set to **One by one**, which this generator requires.
2. **An Ads Generator node** (2): tick the formats you need in **Format selection**, and use the **Local prompt** to remind it what must not change.
3. **An Output node** (3) to review the results and download them as a .zip.
4. **Run all** (4): count 7 credits per format.

The local prompt used here:

> Keep the headline, the line below it, the button and the bottle exactly as in the image, with the same fonts and colors. Recompose the layout to fit each format without cropping any text.

**The available formats**, grouped by family: Amazon IAB (300 x 250, 160 x 600, 300 x 600, 828 x 250, 1456 x 180, 600 x 500, 1940 x 500), Billboard 970 x 250, Portrait 300 x 1050, Skyscraper 160 x 600, Medium rectangle 300 x 250, mobile interstitials (640 x 1136, 750 x 1334, 1080 x 1920), and your own **custom formats**.

![Three formats, a single input image](https://storage.crisp.chat/users/helpdesk/website/-/6/8/0/3/6803ea00be5f5800/ad-en-formats_409anp.jpg =1000xauto)

Each format is recomposed: the headline moves above the bottle in the half page, and the scene stretches upward in the interstitial. The final file is then fitted to the exact dimensions of the format, on a white background when the proportions do not match exactly. The node also delivers a summary sheet, with the source and every format side by side, handy for approval.

**A few tips:**

* **Check every format before publishing**, as you would for an adaptation made by hand. Not happy with a format? Rerun the node: each run offers a new composition.
* **A short headline** recomposes more easily, especially in very wide banners.
* **One visual per run**: to adapt the FR version and then the EN version, put the FR visual in the Reference Image node, run, then replace it with the EN visual.
* **Add an Add to collection node** to file each format automatically in the campaign collection.

In detail: [Component reference](https://help.seelab.ai/en/article/component-reference-1qkc00k/) and [Run, rerun and share a workflow](https://help.seelab.ai/en/article/run-rerun-and-share-a-workflow-l3sy53/).

# 4. Animate the visual

The vertical 1080 x 1920 format becomes an animated story. In **Video**, **Medias** tab:

![The Video space, with the vertical visual as start frame](https://storage.crisp.chat/users/helpdesk/website/-/6/8/0/3/6803ea00be5f5800/ad-en-ui-video_1jvewsb.jpg =1000xauto)

1. **Import the vertical visual** as the **Start frame** (1).
2. **Choose Gemini Omni Flash 1.2** as the **AI Engine** (2).
3. **Set up the shot** (3): 9:16, 720p, 6 seconds.
4. **Describe a subtle movement** (4), and say what must not move.
5. **Click Generate** (5): 25 credits for 6 seconds in 720p.

> Slow, gentle camera push-in. The leaf shadows sway softly on the beige wall as a light breeze passes, the warm morning light shimmers slightly through the serum. The bottle, its label, the headline, the line below it and the button stay perfectly still, sharp and unchanged for the whole shot. Calm, premium skincare mood. Soft ambient sound of a light breeze, no voice, no music.

![The animated story](https://storage.crisp.chat/users/helpdesk/website/-/6/8/0/3/6803ea00be5f5800/ad-en-story-small_1f5r24m.gif =240xauto)

![The shot, from the first to the last second](https://storage.crisp.chat/users/helpdesk/website/-/6/8/0/3/6803ea00be5f5800/ad-en-video_8kjgf7.jpg =1000xauto)

The bottle, the headline and the button stay sharp from start to finish; the camera moves in gently and the leaf shadows sway. Sound is generated with the image: here, a light breeze.

**A few tips:**

* **A slow movement** keeps the message readable: a camera push-in, shadows, reflections, a falling leaf.
* **Explore in 360p** (1.4 credits per second), then rerun the shot you keep in 720p or 1080p.
* **Up to 15 seconds** with Gemini Omni Flash 1.2, for a longer story.
* **An animation without text?** Animate the visual from step 1, then add the text in your edit.

In detail: [Video models in detail](https://help.seelab.ai/en/article/video-models-in-detail-ie0uck/) and [Prompting a video: intact product, camera and lenses](https://help.seelab.ai/en/article/prompting-a-video-intact-product-camera-and-lenses-1tcqyiz/).

# Cheat sheet

* **The product first**, with your Seelab model, and free space planned for the text.
* **Then the text**, with GPT Image 2.6 or Nano Banana 2.1, in quotes, with the visual as image 1.
* **The formats in a workflow**: Reference Image set to **One by one**, Ads Generator, Output. 7 credits per format.
* **Check every format** and rerun the node if a composition does not suit you.
* **The animation last**, from the vertical format, with a slow movement and the instruction not to touch the text.

# With the Seelab team

A campaign to roll out across all your products and markets? The Seelab team can model your range, build your [adaptation workflows](https://www.seelab.ai/en/platform/automations) and [produce the visuals with you](https://www.seelab.ai/en/services/production). Talk to your Seelab contact or write to us from the chat.

# Learn more

* [Create and train a model](https://help.seelab.ai/en/article/create-and-train-a-model-82ftkc/)
* [Create a visual from a reference image](https://help.seelab.ai/en/article/create-a-visual-from-a-reference-image-1uc8jlc/)
* [Workflows: from a single visual to series production](https://help.seelab.ai/en/article/workflows-from-a-single-visual-to-series-production-vr51cc/)
* [Animate your product with your video model](https://help.seelab.ai/en/article/animate-your-product-with-your-video-model-1f7ismq/)
* [Video models in detail](https://help.seelab.ai/en/article/video-models-in-detail-ie0uck/)