Recreating the Lensa App with AI Tools (Because Why Not?)
Overview
A few years ago, Lensa took the internet by storm. You’d upload some selfies, wait an eternity (12 hours, but still), and receive a batch of stunning AI-generated portraits in different artistic styles. People suddenly looked like sci-fi heroes, fantasy warriors, or dramatic movie poster icons.
Naturally, I thought, why wait 12 hours when I can overcomplicate things and do it myself? So, I set out to recreate Lensa’s magic using Replicate and Hugging Face—because if AI can turn me into a futuristic cyberpunk astronaut, I’m all in.
Implementation
Step 1: Training the Model (a.k.a. Teaching AI to Recognize My Face)
To make AI generate images of me instead of, say, a random cat in space, I needed to train a model. I used this Replicate training toolkit, which allowed me to upload a bunch of photos of myself. The more variety, the better—because AI needs to learn my face in different angles, lighting, and expressions (yes, even that one awkward selfie from 2018).
Why Use a Trigger Word?
This is where the magic (or science) happens. I assigned a trigger word—"LDBKLC"—so that whenever I used it in a prompt, the AI would know, Oh, you want to generate an image of this specific person! It’s like giving the AI a secret code to unlock my digital alter ego.
Step 2: Storing the Model (Because AI Needs a Home)
Once the training was done, the model produced a set of weights—essentially, the AI’s knowledge of what I look like. I uploaded these weights to Hugging Face, which is like Dropbox but for AI models. Now, anytime I want to generate images, I can just reference this trained model instead of retraining from scratch (because who has time for that?).
Step 3: Generating the Images (Finally, the Fun Part)
With the model trained and stored, I used this Replicate model to generate images. All I had to do was write prompts like:
- LDBKLC as a futuristic cyberpunk hero
- LDBKLC in a Studio Ghibli-style animation
- LDBKLC as a medieval knight, dramatic lighting, cinematic
Each prompt transformed my likeness into a unique, artistic vision—because why settle for reality when AI can make you look legendary?
Results
The results? Absolutely worth it. I got AI-generated portraits of myself in different styles, and the best part? It took minutes instead of 12 hours.
Would I do it again? Absolutely.
Would I recommend it to others? 100%.
Would I abuse this power and generate an endless stream of AI avatars for no reason? …No comment.
If you want to try it yourself, grab some photos, pick a trigger word, and get experimenting. Who knows? You might just become the next AI-generated superstar. 🚀