Why your profile photo matters more than you think
A recruiter opening your LinkedIn profile decides in a fraction of a second whether to keep reading, and the photo is the first thing they see. LinkedIn's own figures have said for years that profiles with a photo receive many times more views and connection requests than profiles without one, and hiring managers admit, when asked, that a blurry holiday snap or a cropped group picture makes them doubt the rest of the page. None of this is fair — competence has nothing to do with lighting — but it is how first impressions work, and a professional headshot is the cheapest way to control one.
The problem is that most people do not have a professional headshot. Booking a photographer means finding one, paying for a session, choosing a day, dressing up, driving there and waiting for the edited files. For a freelancer changing positioning, a graduate building a first profile, or an employee who simply wants to look presentable before a job search, that is more friction than the task deserves. The result is that millions of profiles carry a picture taken at a wedding, at the beach, or by a colleague in a corridor with a phone.
PixGenia's AI headshot generator is designed for exactly that gap. You upload two to four ordinary selfies, you receive two free low-resolution previews in under a minute, and if you like what you see you can buy a pack of twenty high-definition portraits in five backgrounds and four outfits for €24.90, delivered in minutes to a private link. No account, no subscription, no studio. This guide explains what a good professional portrait looks like, how to take selfies the model can work with, how the generation actually works — including its limits — where your photos go, what the pack contains, and when a real photographer remains the better choice.
What makes a good LinkedIn or CV portrait
Before talking about AI at all, it helps to know what you are aiming for, because a generated portrait obeys the same rules as a photographed one. The first rule is framing. A professional headshot is a head-and-shoulders picture: the top of the head sits near the top of the frame, the crop ends around the chest, and the face occupies roughly a third to a half of the image. Wider than that and your face becomes a dot in the LinkedIn circle; tighter and it feels like a passport photo. The subject is slightly off-centre or square to the camera, shoulders turned a few degrees, chin very slightly forward — the posture photographers coax out of everyone because it defines the jawline and reads as attentive.
The second rule is light. Almost every bad profile photo is bad because of light: a lamp overhead that digs shadows into the eye sockets, a window behind the subject that turns them into a silhouette, a phone flash that flattens the face and reddens the eyes. Good portrait light is soft, comes from the front and slightly to one side, and leaves a small catchlight in each eye. In a studio this is a large softbox; outdoors it is open shade or an overcast sky; at home it is a window with the subject facing it.
The third rule is expression, and it is the one people worry about most. Recruiters and clients respond to faces that look approachable and confident at the same time: a relaxed, closed-mouth or slightly open smile, eyes engaged with the camera, no tension in the forehead. A broad laugh can work in creative fields but reads as unserious in finance or law; a completely neutral face reads as cold in almost every field. The fourth rule is background: uncluttered, out of focus or plain, and in a tone that separates you from it. A grey studio backdrop, a softly blurred office, a city street reduced to shapes — anything that says nothing, so that the face says everything.
Finally there is the question of what recruiters actually expect. Surveys of HR professionals consistently rank the same things: a recent photo that still looks like you, professional but not stiff clothing appropriate to the sector, no sunglasses, no hat, no other people, no obvious filters, and a resolution high enough that the picture is not pixelated on a large screen. A good AI headshot ticks every one of those boxes — provided the selfies you feed it give it something to work with.
How to take selfies the tool can work with
The generated portraits can only be as faithful as the reference photos. The model needs to see your actual face clearly, from more than one angle, in honest light, so that it can reproduce your features rather than invent them. That means the best selfies for an AI headshot generator are, paradoxically, the least flattering ones you would normally take: no beauty mode, no portrait blur, no colour filter, no wide-angle distortion from holding the phone too close. Two selfies are the minimum, four is the maximum, and three is the sweet spot: one straight on, one turned slightly left, one turned slightly right.
Light matters as much for the references as for the final picture. Stand facing a window in daylight, or outdoors in open shade, and keep the light source in front of you rather than behind. Avoid mixed lighting — a warm lamp on one side and a blue screen on the other — because the model will faithfully reproduce the colour cast on your skin. Hold the phone at eye level, at arm's length or slightly further, and look at the lens, not at the screen. If you wear glasses every day, wear them; if you normally do not, leave them off. The tool keeps what it sees.
Your face must be fully visible in every shot: no hand on the chin, no hair across an eye, no mask, no sunglasses, no cap. Tie hair back if it usually hides your jaw. Wear something plain in a neutral colour: the outfit will be replaced in the generated portraits, but a busy pattern, a high collar or a scarf can confuse the boundary between your neck and the clothes. Finally, use recent photos taken on the same day if possible, so that hair length, beard and skin tone are consistent from one reference to the next.
- Daylight from the front, never a window behind you, no flash
- Face completely uncovered: no glasses unless you always wear them, no sunglasses, no hat, no hand
- Three angles: straight on, slightly left, slightly right, all at eye level
- No filters, no beauty mode, no portrait-mode blur, no heavy crop
- Plain clothing in a neutral colour, hair as you normally wear it
- Photos from the same day, sharp, with the face at least a quarter of the frame
How the AI actually works — and what it cannot do
It is worth being precise here, because the market is full of vague promises. PixGenia's portrait tool uses reference-guided image generation from OpenAI's image model. Your selfies are sent to the model as reference images together with a detailed written instruction: reproduce this person's face, skin tone, apparent age, eye colour, hairstyle, facial hair and glasses exactly, place them in a head-and-shoulders portrait with a specified background, outfit and studio-style lighting, looking at the camera with a light, natural smile. The model then generates a new image that has never existed — it is not a retouch of one of your selfies, it is a synthesis guided by them.
That distinction explains both the strengths and the limits. The strength is that the result is a genuinely new photograph-like image with professional light and a clean background, in a pose you never held, wearing clothes you may not own. The limit is that the model is reconstructing you from a handful of references, and reconstruction is never perfect. In practice, the face is recognisably yours in the great majority of portraits, but individual images may drift: the nose a fraction narrower, the smile a touch different, a mole missing, the hairline slightly moved, the glasses frame subtly reshaped. The more consistent and well-lit your selfies are, the smaller the drift.
What the tool deliberately does not do is change who you are. The instructions forbid altering apparent age, ethnicity, body shape, skin tone, glasses or hairstyle. There is no slimming, no whitening of teeth, no smoothing that erases the lines of your face. The goal is the picture a good photographer would have taken of you last week, not an idealised version of you. If you want a different haircut or a beard you do not have, the honest answer is that this is a job for the barber first and the camera second.
The model also cannot work miracles on poor references. If it receives two dark, blurry selfies where half the face is in shadow, it will guess the missing half, and the guesses are where likeness breaks down. That is why the previews exist.
Why we show two previews first, and why the pack has twenty portraits
Reference-guided generation is probabilistic. Run it twenty times on the same selfies and you get twenty different images, most excellent, a few merely good, and occasionally one where the likeness is off. No AI headshot generator can promise that every image will be perfect, and any that does is not telling you the truth. PixGenia's answer is to let you see before you pay: two free previews, low resolution and watermarked, generated from your own selfies with the background and outfit you selected. If the previews look like you, the pack will too, because it uses the same references and the same method. If they do not — usually because the selfies were too dark or too few — you can retake them and try again rather than spending money.
The pack contains twenty portraits for the same reason. Twenty is enough for every combination of the five backgrounds and four outfits, so you get real variety rather than twenty near-identical frames. It is also enough that you can discard the two or three you like least and still have a large set to choose from: a formal one for LinkedIn, a warmer one for the company website, a neutral one for the conference programme. Professional photographers work the same way — they shoot dozens of frames and hand you a selection — and twenty is roughly the number of usable frames a short session produces.
Each portrait in the pack is delivered at 1024 by 1536 pixels, a vertical two-to-three ratio, as a PNG file without watermark. That is more than enough for LinkedIn, a CV, a website team page or an email signature, and it survives cropping to a square without becoming soft. The previews, by contrast, are 512 pixels high with a visible tiled watermark: they are for judging likeness, not for use.
The five backgrounds and four outfits: which to choose for your field
The pack covers five backgrounds and four outfits, and the point of the matrix is that different audiences expect different things. A corporate lawyer and an indie game developer both need a professional headshot, but not the same one. The five backgrounds are a neutral grey studio backdrop, a bright modern office softly out of focus, a blurred city street in daylight, a solid colour wall, and a warm bokeh of golden lights out of focus. The four outfits are a tailored blazer over a shirt, a plain collared shirt, a fine-knit sweater, and a smart casual look (a dark unstructured jacket over a plain t-shirt).
For traditional corporate sectors — finance, law, consulting, insurance, senior management — the safest choice is the grey studio or the bright office with the blazer. It is what colleagues and clients will see everywhere else on the firm's website, and it signals that you understand the codes. For technology, product and start-up environments, the blurred office or the solid colour background with a shirt, sweater or smart casual look reads as modern without being sloppy; the dark jacket over a t-shirt in particular has become the informal uniform of founders and designers. Creative professionals — designers, writers, marketers, architects — can afford the city street or the warm bokeh, which add personality, paired with the sweater or an open-collar shirt.
Healthcare, education and public service sit in between: a bright, clean background with a shirt or a soft sweater conveys competence and approachability at once, which is exactly what patients, students and citizens want to feel. Sales and customer-facing roles benefit from the warmer backgrounds and a smile that is a little broader. Whatever your field, one rule holds: the background should be quieter than your face, and the outfit should be one you could plausibly be wearing at the meeting the viewer is imagining.
- Grey studio + blazer: finance, law, consulting, executives, formal CVs
- Bright office + shirt: healthcare, education, HR, administration, most office jobs
- Solid colour + smart casual or sweater: tech, product, start-ups, founders
- Blurred city + shirt or sweater: sales, communication, freelancers, real estate
- Warm bokeh + sweater: creative fields, coaching, culture, small businesses
Where to use your portrait, and how to crop it for each place
LinkedIn displays the profile photo as a circle and recommends an upload of at least 400 by 400 pixels, with a maximum file size of 8 MB. Your portrait is 1024 by 1536, so crop it to a square centred on the face — head near the top, crop just below the shoulders — before uploading, or use LinkedIn's built-in cropper. Check the result at the small size shown in comments and search results: if the face is too small in the circle, tighten the crop. The same square crop works for X, Instagram, Slack, Microsoft Teams, Zoom and any tool with a round avatar.
On a CV, a photo is customary in France, Germany, Spain, Italy and much of continental Europe, and discouraged in the United Kingdom, the United States, Canada and Ireland where anonymous screening is the norm — follow the convention of the country you are applying in. When you include one, use a small vertical crop, roughly 3.5 by 4.5 centimetres at print size, placed top-right or top-left, and choose the most formal portrait in the pack. A company website team page usually wants a consistent format across the whole team; ask the person who maintains it whether they use squares, circles or portrait rectangles and pick the background that matches the rest of the team.
Email signatures need a tiny image, typically 80 to 150 pixels wide, so crop tightly to the face and export a small JPEG so that the signature does not weigh down every message. Conference organisers, event programmes and press kits usually ask for a high-resolution square or portrait file, often 1000 pixels or more on the short side: the full 1024 by 1536 PNG covers this directly. For a printed badge or a poster, 1024 pixels prints cleanly at around 9 centimetres at 300 dpi, which is plenty for a name badge and fine for a programme; for a large banner you would want PixGenia's image upscaler on top.
One recommendation applies everywhere: use the same portrait, or portraits from the same pack, across every platform. Consistency is part of what makes a person look professional online, and a pack that shares light, expression and style makes it effortless.
Privacy: where your selfies go, and when they are deleted
Photos of your face are sensitive data, and you are entitled to know exactly what happens to them. Here is the full path. In your browser, each selfie is reduced to 1024 pixels on its long edge and re-encoded as JPEG before anything is sent — the full-resolution original never leaves your device. The reduced file travels over an encrypted connection to PixGenia's server, hosted on Vercel, where it is stored under an unguessable address for the duration of the job. When you request previews or the pack, the server sends the selfies and the written instruction to OpenAI's API in the United States, which generates the images and returns them. The generated portraits are stored on the same server so that your private link can serve them.
The transfer to the United States is covered by the EU-US Data Privacy Framework, to which OpenAI is certified, and by OpenAI's API data-processing terms: images sent through the API are not used to train their models, are not shown to other users, and are retained only for the short period needed to run the service and screen for abuse. PixGenia never uses your photos for training either, does not sell or share them, and does not build any face database. The only other processors involved are Stripe, which handles payment and sees your card details, not your photos, and Resend, which sends the delivery email containing your link.
Deletion is automatic and does not depend on you remembering to ask. Your selfies are deleted from PixGenia's server within 24 hours of the generation being completed — in practice, as soon as the pack is delivered. The generated portraits remain available on your private link for 30 days so that you can download them at your leisure, and are deleted after that. The order record itself — amount, date, the email address the link was sent to — is kept for as long as accounting rules require, without any image. No account is created and no password is stored: the private link, and the email that carries it, are your only access, so keep them.
A note on where the tool is available: because some jurisdictions treat facial images under specific biometric statutes with additional requirements, the portrait service is not offered in a small number of US states. Everywhere else, you can also exercise your data-protection rights — access, erasure before the automatic deadline, objection — by writing to the contact address in the privacy policy.
The AI-generated label, and being honest about your portrait
Every portrait in the pack is an AI-generated image, and PixGenia says so. The delivery page and the delivery email carry an explicit AI-generated label, and the technical content credentials embedded by the image model — the C2PA metadata that identifies an image as synthetic — are preserved in the files you download. This is not only good practice: the transparency provisions of the European AI Act, in force since August 2026, require providers of generative systems to mark synthetic content, and we would rather be clearly on the right side of that line.
Does that mean you must announce it on LinkedIn? No — the label is about the image's provenance, not about your obligations. A generated headshot that faithfully shows what you look like is, in practice, comparable to a retouched studio photo: the light and background are constructed, the person is real. What matters ethically and professionally is that the portrait does not misrepresent you. That is why the tool refuses to change your apparent age, ethnicity, weight or features, and why we ask you to give it selfies that show how you really look, glasses and grey hairs included.
There are situations where telling people is simply sensible. If an employer, a professional body or a photographer's release form asks whether the photo is a studio photograph, say that it is AI-generated from your own selfies. If your company's brand guidelines specify photographed portraits, ask before using a generated one on the corporate site. And if you are in a profession where your image is part of the trust relationship — medicine, law, teaching, care — make sure the portrait you choose is the one that looks most like the person clients will meet. A good rule: if a colleague who has not seen you in a year would recognise you from the picture at once, you are fine.
One more line we will not cross: you may only generate portraits of yourself, or of an adult who has given you written permission. The consent step before upload asks you to confirm that the face is yours or that you have that permission, and that you are at least 18. Generating a headshot of someone else without their agreement is a violation of their image rights and of our terms, and it is something we will not knowingly do.
Price, what the pack contains, refunds and the right of withdrawal
The previews are free: two watermarked low-resolution images, generated from your selfies, with no account and no card. The pack costs €24.90, including VAT, paid once through Stripe by card or a supported wallet. There is no subscription, no credits to top up and no hidden fee. For that price you receive twenty portraits at 1024 by 1536 pixels, as PNG files without watermark, across the five backgrounds and four outfits, plus a zip archive of all twenty and a private download page, reachable from the link shown after payment and from the email sent to the address you enter at checkout. The link is valid for 30 days.
Generation starts the moment payment is confirmed and typically completes within a few minutes; the download page updates itself as portraits arrive, so you can start looking at the first ones while the rest are being made. If the service cannot produce the full set — an outage, a moderation refusal on a specific image — it keeps trying, and in the rare event that a pack cannot be completed you are told and refunded rather than left with a partial order.
Because the pack is digital content delivered immediately, European consumer law allows a seller to ask you to waive the usual 14-day right of withdrawal in exchange for immediate delivery, and that is what happens at checkout: you confirm that you request immediate performance and acknowledge that you lose the withdrawal right once delivery has begun. PixGenia nonetheless offers its own guarantee on top: if you have not downloaded any of the twenty portraits, you can ask for a full refund within 14 days of the purchase, no questions asked. Once a portrait has been downloaded, the pack is considered used, and the refund guarantee no longer applies — which is why the previews exist, so that you never pay for something you have not seen.
A professional photographer, another AI app, or PixGenia?
Let us be fair to the photographer first, because for some needs nothing replaces one. A studio session gives you a person who can direct your posture, watch your expression, adjust the light for your face and take a hundred frames until the right one appears. It produces images at very high resolution suitable for a billboard, and it can cover far more than a headshot: full-length portraits, images with your team, environmental portraits in your workshop or office, pictures with a product. If you are a partner in a firm, a public figure, or you need a coherent set of images for a rebrand, book a photographer. The generated portrait is a complement to that reference, not a substitute for the whole shoot.
What the AI headshot generator does better is everything around the simple, single, professional head-and-shoulders picture: it is available tonight, from your sofa, in the country you are in, without an appointment; it costs a fraction of a session; it delivers in minutes; and it gives you a variety of backgrounds and outfits that would take a real shoot several hours and several changes of clothes to produce. For the vast majority of LinkedIn profiles, CVs and team pages, that is precisely the need.
Compared with other AI headshot apps, the differences are practical rather than magical — the underlying technology is broadly similar across the market. PixGenia asks for two to four selfies rather than a dozen or more, so you can start immediately. It shows you previews before you pay, so you are never buying blind. It delivers in minutes rather than hours. It requires no account, no app installation and no subscription. It deletes your selfies within 24 hours and never trains on them, in writing. And it labels the output as AI-generated, which not every service does. Where other tools may offer larger packs, more styles or team features, the choice depends on what you need — for one person who needs one good portrait quickly, twenty variations is plenty.
Step by step: from selfies to portrait in a few minutes
The whole process is designed to be done on a phone, because that is where the selfies are. Open the tool, read the consent notice and tick the boxes confirming that the face is yours or that you have permission, and that you are 18 or older. Then follow the steps below.
- Take three fresh selfies in daylight, face uncovered, straight on and turned slightly to each side, without filters.
- Upload them: the browser resizes each one and shows a thumbnail; add a fourth or replace one if a thumbnail looks dark or blurry.
- Choose a background and an outfit for your two free previews — pick the combination you are most likely to use.
- Wait under a minute and examine the previews at full screen: is the likeness right, is the light flattering, does the expression feel like you?
- If not, retake the selfies with better light and try again; if yes, click to unlock the pack and pay €24.90 through Stripe.
- Watch the download page fill with the twenty portraits, then download them one by one or as a zip, and keep the email with your private link — it works for 30 days.
- Crop a square version for LinkedIn and your avatars, keep the vertical original for CVs and websites, and use the same set everywhere.
Frequently asked questions
Will the portraits really look like me? In the large majority of cases, yes, recognisably and immediately — provided your selfies are sharp, well lit and show your whole face from three angles. Individual images can drift slightly, which is why you get twenty and why the previews come first. If the previews do not look like you, do not buy; retake the selfies.
How long does it take? Previews arrive in under a minute. After payment, the twenty portraits are generated in parallel and typically all appear within a few minutes; you are emailed the link and can close the page in the meantime.
What file format and size do I get? Twenty PNG files at 1024 by 1536 pixels, vertical, without watermark, plus a zip of all of them. Previews are 512 pixels high JPEGs with a watermark and are not meant to be used.
Can I use the portrait on LinkedIn, on my CV, on my company's website? Yes, for any personal or professional use, for as long as you like. You own the right to use the images; we keep none of them after 30 days.
Do I need an account? No. There is no sign-up, no password and no app. Your private link and the delivery email are your access, so keep them safe; anyone with the link can see the portraits until it expires.
What happens to my selfies? They are reduced in your browser, sent to our server and to OpenAI's API in the US under the Data Privacy Framework, used only to generate your images, never used for training, and deleted from our server within 24 hours of generation. The portraits are deleted 30 days after purchase.
Can I get a refund? If you have downloaded none of the portraits, yes, in full, within 14 days of purchase. Once you have downloaded one, the digital content is considered delivered and used, in line with the withdrawal waiver you accept at checkout. If we fail to deliver the pack at all, you are refunded automatically.
Can I change my hair, lose ten years or add a beard? No, and that is deliberate. The tool reproduces you as you are in your selfies. It changes the light, the background, the clothes and the pose — the things a photographer would change — and nothing about your face. Upload your selfies, look at your two free previews, and see what a professional portrait of the real you looks like.