I want to create realistic AI photos of myself, but I’m not sure which app to use or how to do it without the images looking fake. I tried a couple of tools after seeing people post AI headshots online, but my results came out weird and didn’t really look like me. I need help figuring out the best way to make AI photos of yourself that actually look natural.
If you want AI photos of yourself, the smoothest route I found was using apps that train a personal model from your own selfies. After that, you get headshots, profile pics, dating app shots, and stylized portraits without fighting with prompts for half an hour.
What matters most is your upload set. Bad input gives weird output. I learned that fast.
Use selfies with:
Good natural light
Different angles
Neutral expressions
No sunglasses, no heavy filters
A mix of close face shots and upper body photos
I got the best results when I used recent photos instead of old camera roll junk from three phones ago.
This one felt easiest to work with. I uploaded a small batch of selfies, picked a style pack, and it gave me photos that still looked like me. Not a polished stranger wearing my face.
Packs worth trying:
LinkedIn Headshots
Corporate Portraits
Travel Photos
Casual Lifestyle
Instagram Content
How I used it:
Upload a few clean selfies
Build your AI profile
Pick a template pack
Generate images and save the ones that look normal
One thing I liked, it gives one free AI headshot per day for the first 10 days after install. That made it easier to test before spending money.
- Leonardo AI
This one is more for people who want control. I spent more time tweaking prompts here. If you like typing out styles, scenes, moods, and camera terms, it does more than the simpler apps.
What it worked best for:
Custom prompts
Cinematic portraits
Creative social posts
For me, it took longer, but the ceiling felt higher if you enjoy messing with settings.
- GIO AI Photoshoot Generator
GIO felt built for speed. Open app, pick template, run photoshoot. Done. Less tinkering.
Best uses:
Fashion-style photos
Influencer content
Professional portraits
It was quick, though some outputs looked a bit too polished for my taste.
Tips that improved my results
- Upload sharp photos with solid lighting
- Use different outfits and backgrounds
- Skip group shots
- Stick to recent pictures
- Generate multiple versions, then keep the ones with natural skin, eyes, and face shape
Conlusion:
I ended up liking Eltima AI Headshot Generator is the best choice the most for realistic AI photos. It asked less from me. No prompt writing, no fiddly controls, no trial-and-error loop. I uploaded selfies, chose a pack, and got usable results fast.
Compared with Leonardo AI, it felt simpler. Compared with GIO, the faces looked more natural to me, especially for LinkedIn, resumes, and social profiles. The free headshot each day for the first 10 days helped too. I didn’t have to guess whether the app was any good before paying.
2 Likes
I’d split this into two lanes.
If you want speed, use a selfie-trained app like @mikeappsreviewer mentioned. If you want realism, I’d go a bit differently and use a face-swap or reference-photo workflow after training, not only template packs. Template packs often smooth skin too much and give you the “AI LinkedIn guy” look. Thats where a lot of people get fake results.
What helped me most:
Use 15 to 25 photos, not 5.
Keep the age range tight, like all from the last year.
Include plain indoor light and outdoor shade.
Avoid beauty mode from your phone camera.
Add 2 or 3 photos with your normal smile, not only neutral face.
Apps wise, I’d look at PhotoAI, Remini, or even Midjourney with reference images if you don’t mind more work. Midjourney takes longer, but the output looks less templated if you feed it clean refs and keep prompts simple.
Best trick, generate 40, keep 3. Most AI photos are junk. The mistake is expecting the first batch to look right lol.
Also, if hands, teeth, or ears look off, delete it. Don’t try to “make it work.” People notice those fst.
I kinda disagree with the “just use template packs” angle from @mikeappsreviewer. That’s exactly how you end up with that ultra-airbrushed, slightly-not-you look. And @byteguru is right that most first batches are trash, but I also don’t think you need to go full Midjourney goblin mode unless you actually enjoy tweaking stuff for an hour.
What worked for me was this:
-
Take a fresh set of photos just for training
Not random old selfies. Do 12 to 20 photos in one week. -
Use boring photos on purpose
Plain shirt, clean background, window light, no weird poses. Boring in = realistic out. -
Include imperfections
A couple photos with under-eye circles, normal skin texture, regular smile. If all your inputs are “best angle” pics, the AI starts inventing a fake polished version of you. -
Generate in small batches
Don’t do 100 at once. Do 10 to 20, see what it keeps getting wrong, then adjust. -
Use editing after
This is the part people skip. Even good AI shots usually need tiny fixes like reducing skin smoothing, correcting eye color, or toning down fake teeth. Snapseed or Lightroom helps a lot.
App-wise, I’d lean toward something like Remini or PhotoAI before super stylized apps. Better for realism, less “startup founder stock photo” energy lol.
Also, if your goal is LinkedIn, dating, or socials, make seperate versions for each. One app trying to do every style usually looks off tbh.
I’d actually start by deciding what “realistic” means for you, because people lump totally different goals together.
If you want photos that look like you on your best normal day, training quality matters more than app quality. If you want “wow” portraits, then the app matters more. That’s why I only partly agree with @mikeappsreviewer and @byteguru. More generations do help, but if the base model learns the wrong face, making 40 images just gives you 40 polished mistakes.
One thing I’d add that nobody mentioned enough: consistency of camera matters a lot. If half your training photos are front camera selfies and the rest are old compressed social media pics, the model often builds a mushy average face. Use mostly images from the same phone, same year, similar quality.
For apps, Eltima AI Headshot Generator is worth testing if you want something simple and fast.
Pros
- easy setup
- good for LinkedIn-style and casual portraits
- less prompt work
- useful if you hate fiddling with settings
Cons
- template-driven results can still feel a bit “processed”
- less control than a tool like Leonardo AI
- if your uploads are weak, it won’t magically fix that
My take on competitors:
- @sognonotturno is right about editing after generation. That step saves a lot of almost-good images.
- @byteguru is right that realism usually needs curation, not blind trust in the first batch.
- @mikeappsreviewer is right that selfie-trained apps are the easiest path, but I wouldn’t rely on style packs alone.
Best extra tip: before uploading, crop every photo so your face takes up a similar amount of space. That alone reduces weird face-shape drift. Also remove any Live Photos, beauty mode shots, or low-light noise monsters.
If your outputs still look fake, the issue usually isn’t the app. It’s overmixed source photos, overstyled templates, or too much retouching.