Findomestic / Digital & Print Campaign

Creating Findomestic’s AI-Powered Visual Campaign

Overview

In partnership with TBWA and Findomestic, we developed a full-scale advertising campaign composed of 45 high-resolution images — generated entirely with AI. No cameras. No sets. No post-production reshoots.

Just real people, custom-trained models, and a tightly managed generative workflow that delivered production-ready visuals with remarkable creative control and speed. This is not a proof of concept. This is finished work — built with AI, directed by humans, and approved by clients.

The Challenge: Authenticity Without Photography

Findomestic wanted to represent the warmth, diversity, and reality of an Italian family — across print and digital platforms. The campaign required a mix of emotional, everyday moments and professionally composed scenes:
-Family meals and interactions
-Financial consultations with Findomestic advisors
-Urban and domestic environments
-A mix of group and single-character shots

The brief demanded realism, nuance, and continuity — the kind of attention to detail usually handled with a full photo production. Our goal: achieve all of this without a single photoshoot.

Casting and Character Training

To maintain authenticity, we used real people, not AI-generated avatars or composites.
Each actor gave informed consent to have their likeness reproduced using AI. We then captured over 1,000 photos per subject, covering:
✔️A range of poses, angles, and expressions
✔️Multiple outfits and lighting conditions
✔️Isolated backgrounds for clean model training

These images were used to fine-tune private AI models — one per cast member — stored securely and used exclusively for this campaign.

Image Creation Workflow

Creative Brief & Visual References

For each image, TBWA provided visual guidelines: framing, environment, emotion, pose, styling. These served as the backbone of every generation.

Prompting & Generation

We used a hybrid of text prompts, actor models, and reference images to generate low-resolution drafts — typically 50–100 per scene.

Selection & Refinement

From each batch, we shortlisted 20–30 candidates for internal and agency review. The selected images were then refined using inpainting and manual edits to resolve anatomy, composition, or clothing issues.

6K Upscaling & Post-Production

Final selects were upscaled to 6K resolution using proprietary AI tools and then lightly retouched in Photoshop to match print standards. No synthetic textures or facial smoothing — just quality control.

Tools & Ethics

The workflow combined multiple generative tools — including custom-trained models, prompt-based generation, inpainting, and AI upscaling. But more importantly, we prioritized:

✔️Creative direction over randomness
✔️Human identity over avatar fiction
✔️Ethical data usage with informed consent and secure storage

Results

In 10 weeks, we delivered:

✔️ 45 unique, production-ready visuals
✔️ Full character and visual continuity across all assets
✔️ Zero traditional shoots, but full creative flexibility
✔️ A new benchmark for scalable, responsible AI production

This wasn’t faster instead of better — it was faster, with control.
The outcome matched agency and client expectations while proving that generative workflows can scale, adapt, and produce emotionally resonant brand content.

Conclusion

Inside the Image is more than a tagline — it's a shift in how visual campaigns can be created. By combining human creativity, ethical casting, and AI-assisted production, we opened a new chapter in advertising execution.
This is not about replacing craft. It's about redefining it.