How to Generate Selfie Style AI Images — Complete Guide
A bedroom at the end of the evening. One warm lamp, a mirror that hasn't been wiped in a while, a crop that sits slightly off-centre — and a phone held at chest level, plainly visible in the glass. That is the entire category in one image. Selfie style on AINudeGirls produces nude images that read as self-taken rather than photographed, and the phone is what sells it.
For background on the subject itself, see Selfie on Wikipedia.
Two ways in: the generator if you want to make one now, or the selfie style collection if you'd rather see the range first.
What Selfie Style Actually Looks Like
The rule is short. Her phone has to be visible — either gripped in an extended hand and angled back toward her, or caught in the mirror she is standing in front of. The angle follows from that: slightly foreshortened, taken from arm's length, closer and more intimate than anything a photographer across the room could compose. Casual amateur framing, not studio framing.
Everything else in the category is downstream of that defining look. The light is whatever the room happens to have. The pose is whatever her arm allows. Nothing is styled.
There are two ways it plays out. In the hand-held version her arm is out and the frame pulls tight, because that is genuinely what arm's length looks like. In the mirror version you see her standing at the glass with the phone raised, and the device reaches you secondhand, as a reflection. Both count. What does not count is a frame where the phone has slipped out of shot, hidden behind a hand or a fold of fabric — at that point the image has quietly become a straight nude portrait.
Six from the gallery, same composition, different rooms and moments:
selfie style example 1
selfie style example 2
selfie style example 3
selfie style example 4
selfie style example 5
selfie style example 6
All six live on the selfie style category page, and the realistic cut narrows the same set down to the photoreal ones.
Prompt Anatomy
You never type any of this. Picking the category is what produces it — but it is worth seeing what the category actually asks for, because it explains why some outputs land and others drift.
The base description reads like this:
```text
a young woman taking an intimate nude mirror selfie with her phone, exposed body fully visible in the reflection, casual amateur framing with warm bedroom lighting
```
Four things are doing work there. *Taking an intimate nude mirror selfie with her phone* pins the device in frame — that is the non-negotiable part. *Fully visible in the reflection* keeps the body readable in the glass rather than lost to a bad angle. *Casual amateur framing* is what stops the composition tidying itself into a portrait. *Warm bedroom lighting* sets the room.
The category also carries alternate settings, so a batch isn't six versions of the same wall. This one stays in the bedroom but goes further on detail — hand position, expression, the specific instruction that she is alone in frame:
```text
intimate mirror-selfie view of a young woman alone in a softly-lit bedroom, phone held in her left hand at chest level, completely nude with her bare breasts and completely shaved smooth pink pussy visible in the reflection, hair tousled, mischievous smile and bitten lip, warm bedroom lighting, casual amateur framing, female alone in frame, no other people
```
And this one moves the whole scene into a bathroom after a shower, which changes the light, the skin and the props while keeping the phone exactly where it needs to be:
```text
intimate mirror-selfie view of a young woman alone in a bathroom after a shower, phone held in her left hand at chest level, completely nude with wet hair and water droplets on her bare body in the reflection, towel pooled at her feet, mischievous smile, warm bathroom lighting, female alone in frame, no other people
```
Notice how much of each variant is about the room rather than the body. That is deliberate. The pose is fixed by the phone; the setting is where the variety comes from.
Running a Selfie Style Generation
1. Open the Generate page and pick selfie style from the category list. You can pick more than one category — that's how the platform composes a scene.
2. Start the generation. Composition, phone placement and lighting are assembled from the category itself, so there is nothing to describe.
3. The image lands in your gallery as a WebP file at 1312x1968 — a portrait ratio that suits how close and vertical selfie framing naturally is.
A single image costs 15 credits. Sets are cheaper per image: 68 credits for five, 120 for ten, which is the sensible way to buy if you want a few settings to compare. Video runs at 250 credits.
If you want the same woman recognisable across a whole batch rather than a new one each time, pick the same AI model. It gives the generation a fixed face to hold onto from output to output.
Frequent Problems
Picking selfie POV by mistake. The two sound alike and both involve a phone, but they are opposites. POV puts you behind her eyes, which means the phone is standing in for your own viewpoint and never appears. Selfie style puts the device in the picture as an object. If you want to see the phone, this is the category; POV will hand you something else.
Anatomy going soft at the angle. Foreshortened shots are demanding. An arm bent back to hold a phone, a torso twisted toward a mirror — these are exactly the poses where a spine can read subtly wrong, a shoulder can look dislocated, a limb can break, or proportions can drift in ways a straight-on portrait would never produce. The generation pushes against twisted spines, dislocated joints, broken limbs and distorted proportions as a matter of course, which is what keeps the body believable at close range.
Output that looks too good. Perfectly centred, evenly lit, no visible device — that is a portrait wearing a selfie's clothes. Uneven light and an awkward crop are the point here, not defects. If the phone is barely legible or angled out of shot, the image has left the category.
Selfie Style vs Its Neighbours
The categories people most often confuse with this one, or deliberately pair it against:
Category What it actually is
--- ---
selfie style (this guide) self-taken phone selfie: her PHONE is visibly in her extended hand (or visible in the mirror of a mirror-selfie), casual amateur framing
all fours on hands and knees, on all fours
bent over standing and bent forward at the hips, ass pushed out toward the viewer or partner
dancing she dances sensually, sinuous movement
Playing Well With Others
Selfie style belongs to the scene group — the same set as 69-position, against-wall, ahegao, all-fours, amazon-position and anal. Only one scene applies to an image. Pick two and just the first one you selected shows up; the second is quietly ignored rather than blended in. So if you want selfie style, select it and leave the other scenes alone.
Company is handled separately. By default she is on her own in the frame, which is what the category is built around, and the male-partner toggle is a plain on/off that sits outside the scene grouping. Switching it on changes who is in the picture, not how the picture is framed — phone still visible, framing still casual.
Quick Answers
Q: Does the phone have to be in a mirror, or can she just hold it up?
Either works. A mirror shot shows her at the glass with the phone reflected back; a hand-held shot has her arm extended with the device tilted toward her. The only requirement is that it stays visible.
Q: Why do my images keep switching between a bedroom and a bathroom?
Because the category ships with more than one setting. A bedroom mirror and a post-shower bathroom are both built in, which is what gives a batch some range instead of six near-identical frames.
Q: My result looks like a polished nude portrait. What went wrong?
The composition tidied itself up and the phone got lost. Selfie style depends on the device being legible somewhere in frame — in her grip or in the glass. Without it, there is nothing left to distinguish the shot from a standard portrait.
Q: What does one image cost?
15 credits for a single image, 68 for a set of five, 120 for a set of ten. Every one comes back as WebP at 1312x1968.
Continue Exploring
Guides for the categories that sit closest to this one:
- How to generate all fours AI images
- How to generate bent over AI images
- How to generate dancing AI images
The rest are on the blog index.