Adults only. Which apps render a university-age AI girlfriend and a believable campus scene, the words that ruin it, and the prompt we start from.
Adults only. This page is about university students, meaning adult women in their twenties: final-year undergraduates, postgraduates and researchers. Every app listed requires you to be eighteen or over. The technical point that makes this category its own thing is that it is a scene problem rather than a face problem. What people actually want here is a setting, a wardrobe and a rhythm of life, and settings are where image models fall apart: lecture-hall architecture warps, book spines and signage come out as nonsense lettering, lanyards and laptops mangle, and the moment you put more than one person in a campus scene, hands and faces in the background collapse.
There is a second trap, and it is the important one. The bare word "student" is ambiguous in training data and drags renders toward school iconography, which is both the wrong image and the fastest way to get a prompt refused. The fix is to say "university" or "postgraduate" and to state an adult age in every prompt. Below: the apps that handle scenes and wardrobe best, our verdict, the vocabulary that keeps the setting on campus and the character unambiguously adult, and the exact settings and prompt we would start from.
Best image score of the eight, which matters more here than in face-only categories because backgrounds and props are what break first. Character locking keeps her the same person as you move her between scenes.
Equal image quality at $19.99. Backgrounds are model-specific in a way faces are not, so a second platform is a genuinely different roll on the same campus scene.
Free tier and strong scores across the board. Scene work is iterative by nature, so somewhere to burn attempts before paying is worth more here than usual.
$12.99, strong memory, natural voice. The one to pick if what you want is a companion whose week has a shape to it rather than a photograph of a library.
Different model in the same band, and background handling varies between models more than almost anything else. Worth a pass.
Same score band as Lovel and Xotic. Useful as a third attempt when two platforms have both given you a bookshelf full of gibberish lettering.
Image score of 9 with a large realistic library, so you can browse an existing character and spend your effort on the scene instead of the face.
Lowest image score here at 7 and romance-led rather than image-led. Fine for the relationship, weak for a category whose value is largely visual.
These scores are our editorial judgement of the category, not of any single app and not measurements. Every character discussed is an adult university student. The scores describe what tends to go right and wrong when you build one anywhere.
| Criterion | Rating | What it means for you |
|---|---|---|
| Setting reads as university, not school | The core problem. "Student" alone pulls the wrong institution and the wrong age. You have to say university, postgraduate or a named subject, plus an explicit adult age, every single time. | |
| Background detail holding up | The worst score on this page. Book spines, noticeboards, whiteboards and signage all render as convincing-looking nonsense, and lecture-hall geometry bends. Keep backgrounds shallow and out of focus. | |
| Everyday wardrobe | The category's strength. Hoodies, jeans, tote bags, layered coats and trainers are extremely well represented and come back clean. | |
| Ease of setup | The character is easy. The scene takes several attempts, and multi-person campus shots take more than several. | |
| Filter friction | The highest friction of the age-related categories, because the word "student" is ambiguous and moderation treats it cautiously. Stating an adult age and writing "university" or "postgraduate" clears most of it. If you find yourself reaching for softer wording to get a prompt through, stop. | |
| Conversation and persona | Strong. Deadlines, a dissertation, a part-time job and a shared flat give a language model plenty to work with, and none of it depends on the image. |
We revise these as the underlying image models change. If your experience differs, tell us. We would rather correct a page than defend it.
A starting point, not a rule. Build the character first against a plain background and lock her, then move her into scenes. Doing it the other way round means every scene attempt also re-rolls the face.
| Setting | Suggested value |
|---|---|
| Stated age | Always an explicit adult number, twenty-two or twenty-four. This is the setting that keeps the render on the right side of both the age band and the moderation layer. |
| Institution wording | Say "university student", "postgraduate" or "in her final year". Never the bare word "student", which pulls school imagery you do not want. |
| Course | Name a subject: architecture, medicine, comparative literature. It quietly supplies wardrobe, props and conversation topics for free. |
| Wardrobe | Ordinary layers. Hoodie or knit, jeans, trainers, a canvas tote, a coat. This is the part the models do best, so let it carry the scene. |
| Background depth | Shallow depth of field, background thrown out of focus. It is the single most effective fix for warped shelving and nonsense lettering. |
| Scene choice | One person, one location. Library corner, a cafe table, a stairwell, her own room. Avoid crowds, lecture halls full of people and anything with legible signage. |
| Personality | Write the week: what is due, what she is avoiding, who she shares a kitchen with. It is the half of this category that never breaks. |
Prompt to copy
A 23-year-old university student, photorealistic. Dark wavy hair tied back loosely, minimal makeup, natural skin texture. Wearing an oversized navy hoodie, jeans and a canvas tote bag over one shoulder. Sitting at a cafe table with a laptop and a coffee. Warm afternoon light through a window, background thrown well out of focus. Moderate realism, shallow depth of field, no legible text in frame.
Four directions the same category can go in. Copy one, paste it into the character builder, and change the parts you disagree with.
The classic scene and the hardest one to get clean. Blur the shelves hard or you will get walls of invented lettering.
A 24-year-old postgraduate student, photorealistic. Light brown hair in a loose bun, glasses, no makeup. Grey knit jumper, sleeves pushed up. Sitting at a desk with an open notebook, library shelves far behind her and heavily out of focus. Warm lamp light, shallow depth of field, no legible text. Natural skin texture, moderate realism.
Outdoor, single figure, architecture kept distant. The most reliable way to get a campus feel without a campus rendering.
A 22-year-old university student walking outdoors, photorealistic. Dark hair loose, wind-blown. Green raincoat over a hoodie, jeans, white trainers, canvas tote over one shoulder. Autumn trees and distant blurred stone buildings behind her, overcast daylight. Shallow depth of field, no signage, no other people in frame. Natural skin texture, moderate realism.
Interior, controlled, no institutional architecture at all. The version that regenerates most consistently.
A 23-year-old student in her room at night, photorealistic. Hair up in a messy bun, glasses, oversized university sweatshirt. Sitting cross-legged on a bed with a laptop, mug beside her. Warm desk lamp as the only light source, cluttered shelf out of focus behind. Shallow depth of field, no legible text. Natural skin texture, moderate realism.
No background to break. Generate this one first, lock the character, then move her into the scenes above.
A 22-year-old woman, photorealistic studio portrait. Dark wavy hair tied back, thick eyebrows, a small mole on her cheek, minimal makeup. Plain grey sweatshirt. Soft key light from the left, plain neutral backdrop. Natural skin texture, moderate realism, sharp focus on the face, no beauty smoothing.
Two lists in one. The upper rows are the words that put an image model on a university campus with an adult in the frame. The lower rows are the words that quietly send it somewhere else, which is the commonest reason a prompt in this category either looks wrong or is refused outright.
| Term | What you actually get |
|---|---|
| University student | The correct base term. Anchors the institution and the adult age band in one phrase. Use it in every prompt. |
| Postgraduate, PhD candidate, researcher | The strongest options available. They rule out any ambiguity about age and bring an adult setting with them. |
| Named course | Architecture, medicine, law, comparative literature. Supplies props, wardrobe and conversation without you asking for them. |
| Campus, quad, lecture hall, seminar room | Location tokens for adult higher education. Use them sparingly and keep them out of focus. |
| Library stacks | Evocative and treacherous. Book spines are the single least reliable thing an image model renders. Blur them. |
| Hoodie, tote bag, layered coat, trainers | The wardrobe vocabulary that works. Well represented, stable, and it carries the whole category when the background will not. |
| Shallow depth of field | Your main tool. Everything the model cannot render correctly should be out of focus. |
| The bare word "student" | Ambiguous in training data. It pulls toward school imagery and younger renders, and it is the leading cause of refusals in this category. Always qualify it. |
| "Schoolgirl", "uniform", "high school", "prom" | Do not use these. They are the wrong institution, they push the render out of the adult age band, and they will be blocked. There is no version of this category that needs them. |
| Legible text of any kind | Signage, whiteboards, book titles, lanyards. Assume every one of them will come out as invented lettering, and compose so none of them are readable. |
Yes. Every companion described here is an adult university student in her twenties, and every app listed requires you to be eighteen or over. Prompts that could read as anything else are refused by the platforms and are not something this guide will help you with.
Because image models cannot render text or complex architectural repetition. Book spines, noticeboards and building facades come back as plausible-looking nonsense. Use shallow depth of field, keep buildings distant, and compose so nothing readable is in frame.
It is ambiguous in the training data and pulls toward school imagery rather than higher education, which also drags the apparent age down. Write "university student", "postgraduate" or "in her final year", and state an explicit adult age in the same prompt.
More than in most categories, yes. Moderation treats the word "student" cautiously for good reason. Explicit adult ages and university-specific wording clear most of the friction. If a prompt is being refused, the answer is to make the adult framing clearer, never to soften the wording to sneak past it.