This one almost always comes down to photo quality and consistency during training.
Even small differences in your photo set can make a huge impact on how realistic your results look. Here are a few things that could explain it:
📸 Their photos were clearer or more consistent
If they uploaded high-res pics with good lighting, natural angles, and consistent styling, their AI had an easier time learning how they look.
👤 They followed the upload tips more closely
Photos with hands blocking the face, poor lighting, or flipped images (like mirror selfies) can reduce accuracy.
😐 They showed more useful expressions
Your AI can only replicate the expressions you gave it — if your friend uploaded smiling, neutral, and expressive shots, their AI has more range.
🧠 They may have experimented more with prompts and settings
Sometimes it just takes a bit of tweaking — try adjusting your character flexibility or rewriting prompts for better results.
The good news? You can always retrain your AI with better photos to level things up.