The most realistic Foxy results come from strong training photos and smart prompting. Here's how to get there.
Start with strong training photos
Real, sharp, well-lit photos — no filters, no smoothing
Face frontal photo with your upper chest area visible — not just a headshot
Consistent appearance across all 3 photos (same hairstyle, etc.)
Two different body poses in your body shots
Daylight where possible — natural light produces the most realistic training
Use Custom prompt for fine control
Templates are fast but the prompts are locked. For the most realistic outputs:
Use Create → Custom prompt to write your own scene
Describe what we'd see, in concrete visual language ("standing on a balcony at golden hour, looking off to the side, soft natural light")
Avoid abstract feelings — describe the visual scene instead
Iterate
AI generations vary run-to-run. The best images and videos usually come on the 2nd, 3rd, or 4th try. Don't stop at the first generation if it isn't right.
Check character settings
If you have tattoos, make sure the Tattoo Toggle in Character Settings is set correctly. Toggling it can dramatically change how realistic your tattoos appear.
If you've followed all of the above and still aren't getting realistic results, retrain with better photos — see My AI doesn't look like me — what can I do?
