The training brief: text, PDF or both
The training images show your product. The brief gives the Seelab model everything the images can't say: what you want to do with the visuals, the product details that must never be missed, your brand's world. It's what anchors your brand elements and product characteristics in the model: the richer it is, the more the model knows.
In the creation form, Import a brief accepts a text, a PDF, or both. Example with the Monolithe speaker from the demo brand Nord Audio.

Text, PDF or both?
You have… | What you can add |
|---|---|
Few documents about the product | A fairly detailed text brief: uses, points of attention, style |
A product sheet, brand guidelines | The PDF, plus a text saying what you want to do with the visuals |
A complex product or a highly codified brand | Both, as complete as possible |
The text brief: your intentions
The text brief takes a few sentences, more or fewer depending on whether you add a PDF. It can cover:
- What you want to do with the visuals: e-commerce packshots, lifestyle scenes, campaigns, social media, planned formats (1:1, 4:5, 9:16…).
- Points of attention on the product: what must stay exact (logo, label, color, proportions, a signature detail), what is often reproduced wrongly, what could be confusing.
- The overall style: mood, lighting, brand world, what you want to avoid.
- Variants: if your images show several colors or finishes, say so and explain how to tell them apart.
Example for the Nord Audio speaker, with a PDF:
Nord Audio Monolithe, portable wireless speaker. Use: e-commerce packshots, lifestyle scenes in Scandinavian living rooms, social media visuals (4:5 and 9:16). Points of attention: keep the cylindrical shape, the grey woven fabric, the cork base and the small NORD AUDIO logo engraved on the top ring exactly as in the reference images. Style: soft natural light, warm minimalist interiors, calm and premium. Technical details and brand guidelines: see the PDF brief.
The PDF brief: what you already know about the product and the brand
The PDF gathers the information you already have: product sheet, technical sheet, brand guidelines. It feeds the model and saves you from repeating it in every prompt: the composition of a fragrance, the ingredients of a food product, the dimensions of a sofa, the exact texts on a label.
There's no required format: start from your existing documents. What you can put in it:
Section | Examples |
|---|---|
Product sheet | Exact name, features, materials, colors, ingredients or composition, dimensions and weight |
Product texts | Label, engraving, markings, copied exactly |
Reference views | Captioned images (front, side, top), close-ups of details, scale reference |
Brand guidelines | Official logos, typefaces, colors, tone, photo style, don'ts |
Your uses | Examples of expected visuals, formats, channels |
The more complete the PDF, especially on the brand (style guide, logos, colors), the more the model can do.
A few tips
Images
- Name the files by view (MONOLITHE_FRONT.jpg, MONOLITHE_detail_logo.jpg): simple, and it removes ambiguity.
- Add close-ups of what matters: logo, lettering, textures, finishes.
Size and scale
- A close-up makes a small element look big: give the real size of details, for example "the logo is 3 cm wide".
- For furniture or a large object, give the dimensions and, if possible, a familiar reference: a person, a door, an A4 sheet.
Texts and logo
- Copy the product texts exactly and use the official logos rather than redrawn versions.
Special cases
- Several colors or finishes: one model per variant gives the most faithful results.
- A similar product not to be confused with (same range, other version): say what sets them apart.
- People in the photos: crop out faces when possible, otherwise the model learns the person along with the product.
Test the model
After training, test the model on a few very different scenes. As with any prompt using a Seelab model, describe the scene, not the product:
The product on a light oak sideboard in a Scandinavian living room, soft morning light, 50 mm lens
Packshot on a pure white background, soft shadow, front view
If a detail drifts, complete the brief or the images and run the training again with Retrain from the model page.
Where to add the brief in Seelab
In the model creation form, after importing your images and marking the 3 key images 👍, click Import a brief: paste the text, add the PDF, or both.

In short
- Sharp, varied images, ideally named by view.
- A text brief that says what you want to do with the visuals, what matters on the product and the style you're after.
- A PDF with what you already have: product sheet, technical sheet, brand guidelines.
- The more complete the brief, the more faithful the model to your product and your brand.
With the Seelab team
Preparing the brief is part of our onboarding and support: the Seelab team can start from your product sheets and brand guidelines to write the brief, prepare the images, then train and test the model with you. Contact us from the Custom models space, or via the chat.
Learn more
Updated on: 12/11/2026
Thank you!
