This article is about the story of Photographe AI (written by myself, still a human as of today).

✨ Intro

I don't know about you, but I have only 3 portraits of myself that I actually like:

• the single time a professional photographer took a photo of me during an entrepreneurship salon (back in the early days of Twake),
• one random photo on a white background (no idea when it was taken),
• and one fun shot from 11 years ago.

I used them for every post and document until just a few weeks ago.

I needed a new professional photo. But the idea of booking a session, dressing up, posing… felt like too much.

Let's face it, getting a great headshot is expensive, time-consuming, and honestly a bit awkward. You need a good photographer, the right outfit, the right lighting, the right mood, and the time to do it (probably the most difficult for me). And you still might not like the result, because you know you never like photos of yourself anyway.

So I built my own tool — one that could turn any image into a clean, professional portrait and even more — as I'm slowly losing my hair I also need to checkout original and new hairstyles — Movie poster? Meme? Celebrity shot? No problem.

That's how Photographe AI was born.

📸️ What you can do with Photographe.ai

Generate photos in any situation

The base feature of Photographe.ai, you describe what you want, and you get an image out of it. Astronaut, PhD in biology, an army of you manifesting? You've got it!

Put yourself in an existing photo

This is probably the most fun feature, upload a movie poster, or virtually any image, and put yourself as the main character in one click. You can reuse exactly the image as it is, or use the general setup and create a new one (perfect for memes).

Try new hairstyles

One of my sister's favorite features, try new hairstyle and see how good it suits you. Just describe your new haircut or use a reference photo.

Try out outfits

Trying outfit from a Vinted link or just from an image is super easy. I used it to prepare myself for a friend wedding. But it works for basically any outfit.

🚀 What's up next

As we can see on the home page of https://photographe.ai/, there is 3 main focus for the future of the app:

• Point and click editing: I don't know yet if the future of photo editing is text requests or point and clic, usually we like to have some visual and spacial tools and I want to try to keep this (in addition to text editing!). So be prepared for point and click to edit any image.
• New hairstyle, from an image, try outfits: Show a list of popular reference images, or celebrities you could impersonate, not just yourself, why not friends too? Sky is the limit!
• Video: Generative Video is the next chapter planned for 2025, there are some very anticipated open-source models I can't wait to implement in photographe.ai

✍️ My take on generative AI

I strongly believe our digital "self" is becoming more real through AI. We all carry multiple shells: the one we show in private, who we are with close friends, how we behave at work — and finally, how we present ourselves online.

This last shell — our digital presence — has become increasingly fragile. It's worn thin by constant emails, notifications, and the exhausting need to perform — to always be "on," to always look polished in front of a faceless, endless audience.

We need a new kind of shell — and it makes sense that a parallel version of us exists on the internet, not fake, but chosen. One that serves as the first point of contact with other humans on this same World Wide Web.

This evolution has already started. It began with choosing the perfect photo before posting it, then moved to Snapchat filters, and now to beauty filters enabled by default on phone cameras.

But those tools flatten us.

Generative AI can do something different: it can bring our inner image to life — not just a reflection, but a vision. What better tool to express the self we carry inside?

We've all edited reality. But what if AI helped us express it instead?

⚙️ How it works (for tech enthusiasts)

Building Photographe.ai wasn't just about slapping Stable Diffusion on a website and calling it a day. We needed full control, precision, and above all, consistency — so I decided to train original models using Flux1D, a custom pipeline that lets me create specialized LoRAs that create portrait perfectly matching the real person behind.

To improve the prompts and avoid chaos in the output (you know, extra hands, wonky faces, haunted eyes…), we use OpenAI and other LLMs to rewrite prompts dynamically. It helps stabilize the result, fine-tune the mood, and even introduce some subtle personality traits based on the target pack or visual. It also allows the user to describe the situation in their own language. No need to write long paragraphs, just one word is enough and our LLMs will generate the full image prompt.

In some cases — and that is super useful for poster face replacement or hairstyle change — we need to generate an initial image that doesn't look like the user, and then we replace the original face using MediaPipe for detection and OpenCV for clean, sharp reintegration. I actually documented this process here.

After a few roadblocks, broken pipelines, and very cursed outputs, we finally started getting results that were not just good — and they were superb.

Conclusion

Building Photographe.ai has been an incredible journey from solving my own problem to creating a tool that helps others express their digital identity. The technology behind AI portraits is fascinating, but what excites me most is how it democratizes creative expression.

Whether you need a professional headshot for LinkedIn, want to see yourself as a movie character, or just want to have fun with friends, AI photography is opening up possibilities we couldn't imagine just a few years ago.

The future of digital identity is not about replacing reality — it's about expanding our ability to express who we are and who we want to be.

Try it out, I'd love to get your feedback!

Visit Photographe.ai and let me know what you think. The future of photography is in your hands — literally.

(not) me with a parachute !

Try it out, I'd love to get your feedback!