Let the photographs
do the talking.

The brief

A photography business sells with its photographs, so the website had one rule from day one: the photographs are the only background. No cream sections, no white cards, no stock textures — every screenful is a real photograph with the words sitting on top of it.

That rule creates hard problems. A photograph stretched across a wide laptop screen loses most of its height; the same photograph on a phone loses most of its width. Heads get cut off, text becomes unreadable over busy backgrounds, and a site with hundreds of large images can easily become slow. The site also had to work for couples and families who speak different languages, and to be something a search engine can actually index.

A photography site is only as good as the worst-cropped photograph on it.

The process

The site is generated by a small set of Python scripts rather than a website builder. AI did much of the drafting — code, copy, translations, structured data — and every result was checked on real screens before it shipped.

The stack

No CMS and no framework to keep upgrading — the generator writes plain HTML and CSS, and Netlify serves it.

Where it stands

The site is built and running as a preview while the portfolio is finalised. It is not yet open to search engines, so there are no traffic or ranking figures to report — this case study describes how it was built, not how it has performed.

What this does not mean

AI does not choose, grade or retouch the photographs. Every photograph is edited by hand in Lightroom and chosen by the photographer before it reaches the site, and the pipeline applies no automatic colour correction. The tools that check for soft focus or near-duplicate frames only flag photographs for review — they never remove anything on their own.

The code and copy were accelerated by AI. The decisions — what the site should feel like, which photographs represent the work, what to leave out — were human calls.