How do AI headshots work? From one selfie to a new portrait in 5 steps

An AI headshot isn't your selfie with a filter on top. The tool reads your face from a photo, then generates a brand-new portrait with a different background, clothing and light, while trying to keep you recognisable. Here's what happens at each step, why some results look "off", and what it costs.

The short version

Most AI headshot workflows follow the same basic pattern:

1. You upload one clear, recent selfie. Additional source photos are optional.
2. The system analyzes facial structure and image quality.
3. You choose or receive a professional style: background, clothing, lighting, pose, and expression.
4. A generative model creates new portraits from those identity cues and style instructions.
5. You review the results and choose the version that looks natural.

That is why AI headshots are different from normal retouching. Retouching edits an existing photo. AI headshot generation creates a new image based on your face and the requested professional look.

1. You upload one clear, recent selfie. Extra photos are optional and are used as extra references.
2. The system reads your face (shape, proportions, features) as an identity reference.
3. You choose a look: background, clothing, light, pose.
4. A generative model creates new portraits from your identity reference and that look.
5. You review the results and keep the one that still looks like you.

Step 1: Upload a selfie that shows your real face

The input photos matter more than most people expect. A good selfie gives the model enough information to understand your face without guessing.

Use a selfie that is:

• recent
• sharp, not blurry
• evenly lit
• taken from a normal front-facing angle
• free from heavy filters
• not blocked by sunglasses, hands, hats, or hair across the face

You do not need studio photos. A phone selfie is fine when the face is clear. The most common reason AI headshots look wrong is not the camera; it is poor source material.

More tips in our selfie tips for better headshots.

Step 2: The system reads facial structure

Before generating anything, the model needs to understand the identity cues in the uploaded photos. These include face shape, eye spacing, jawline, mouth shape, nose structure, hairline, and overall proportions.

This stage is why clean lighting matters. If one side of your face is hidden in shadow, or if the selfie is softened by a beauty filter, the model receives weaker identity information. It may still produce a polished headshot, but the result can drift away from how you actually look.

Step 3: Style instructions shape the final photo

Once the system has enough identity information, it combines that with style instructions. These instructions define the visible choices that make the image feel professional:

• background
• clothing
• pose
• expression
• crop
• lighting
• overall tone

For LinkedIn, a neutral background and direct head-and-shoulders crop usually works best. For a CV, a slightly more conservative look can be safer. For a portfolio or speaker bio, a warmer expression may fit better.

In Portreya you pick one template pack, such as LinkedIn Shots, and the photos are created in the styles inside it. Want ideas first? Browse headshot examples.

Step 4: The model generates a new portrait

The generated image is not a pixel-by-pixel edit of your original selfie. It is a new portrait guided by your identity and the selected style.

That matters because it explains both the strengths and the weaknesses:

• It can create a cleaner background than your original selfie.
• It can improve lighting and wardrobe.
• It can produce several professional variations quickly.
• It can also make mistakes around details like glasses, jewelry, teeth, hair edges, and hands.

The right way to judge the output is simple: does it look like a believable professional photo of you, or does it look like an over-polished version of someone similar?

Want to try the idea before paying? See what a free AI headshot generator actually gives you and when $5.99 is worth it.

What the AI actually does (in plain words)

Most AI headshot tools are built on diffusion models. A diffusion model learns to turn random noise into an image step by step, removing a little noise at each step until a clear picture is left (Ho et al., 2020). Modern image models usually do this in a compressed "latent" version of the image, which makes it fast enough for everyday use, and they can be guided by an instruction such as "studio headshot, grey background, navy blazer" (Rombach et al., 2022).

On its own, that model would draw a stranger. To make it draw you, tools add your identity in one of two ways:

1. Train a personal model on several photos. The tool fine-tunes a copy of the model on a set of your photos so it learns your face. Research on this approach uses "just a few images of a subject" (DreamBooth, Ruiz et al., 2022); some commercial headshot tools ask for 8–20 photos. It takes extra time before your first result.
2. Use one photo as a face reference. Some tools read identity features from a single photo and guide the model with them while it generates, with no per-person training. The InstantID paper describes this as personalisation "using just a single facial image" (Wang et al., 2024).

Some tools then run a separate restoration or upscaling step to sharpen detail; Real-ESRGAN is a well-known example of that kind of model (Wang et al., 2021).

How Portreya does it: Portreya doesn't train a personal model on your face. It uses generative image models from third-party AI providers and passes your selfie to the model as reference images while it creates each photo. After generation, an automatic refinement step tidies face detail; there's no manual editing. One clear selfie is enough; extra photos are optional and are used as extra references.

This is also why errors happen where they do: the model re-draws fine details such as glasses, teeth, earrings and hair edges instead of copying them, which is why Step 5 checks exactly those areas.

Two ways to get your face into the picture

	Train a personal model	One selfie as a reference
Photos you upload	Several (some tools ask for 8–20)	One clear selfie
Per-person training	Yes	No
Wait before the first photo	Longer (training first)	Generation only
Portreya	—	Uses your selfie as reference images while it creates each photo; no per-person training; about a minute per photo

Step 5: Review the result like a human

Do not pick the image that looks most perfect. Pick the one that looks most credible.

Check these areas before using an AI headshot:

• eyes and eye direction
• teeth and smile shape
• glasses and reflections
• hairline and hair edges
• ears and earrings
• collar, lapels, and clothing edges
• skin texture
• overall likeness

Small imperfections are less important than identity. A headshot can be beautifully lit and still be a bad choice if friends or colleagues would not recognize it as you.

Not sure what "good" looks like? See what to check in an AI headshot before/after.

How long it takes and what it costs

Each photo is ready in about a minute once your selfie is uploaded and your options are set. Pricing is one-time, with no subscription:

Plan	Price (USD, one-time)	Photos	Edits included per photo	Template packs
1 Pack	5.99	2	1	Choose one included pack
Try-On	27.49	10	2	Included packs
Studio	34.99	15	3	Included packs

Prices checked on Portreya pricing on 31 Oct 2026. Every purchase has a 14-day money-back guarantee. Creating an account is free; every photo needs a paid pack. Comparing with a studio? See what a professional headshot photographer costs.

What happens to your selfie

According to Portreya's privacy policy (revised 28 Oct 2026), a new selfie "is uploaded and analysed only after you select 'Approve & Save' on the review screen", and "We do not train a custom face model for your account." You can "delete individual selfies and generated images, or all your images in account settings." Uploaded selfies and generated photos "may remain while your account exists", and you can request deletion via support. The policy also notes that AI providers "may retain data under their own terms and the configuration used for a request". For the wider picture, read are AI headshots safe? and the UK regulator's guidance on biometric data.

What affects AI headshot quality most?

Clear lighting

Balanced lighting helps the system preserve face shape and skin tone. Harsh shadows, backlighting, or yellow indoor light can make the generated output less reliable.

Recent photos

Use photos that match how you look now. If the uploaded photos are several years old, the result may look polished but outdated.

Consistent identity cues

If every source photo shows a very different haircut, beard, makeup style, or weight, the model has to average conflicting information. Use selfies that represent the version of yourself you want the final headshot to match.

Simple professional styles

Clean backgrounds, realistic clothing, and natural expressions usually outperform complex cinematic looks. For profile photos, clarity beats drama.

AI headshots vs traditional photography

Question	AI headshot	Traditional photographer
Need a photo quickly?	About a minute per photo	Depends on booking and editing schedule
Want in-person direction?	Limited	Strong
What does it cost?	From $5.99 for 2 photos (1 Pack)	See our photographer price guide
Need full creative control on set?	Limited	Strong
Need a LinkedIn/CV profile photo?	Often enough	Also strong

How to get better AI headshots

Use this checklist before generating:

• Take your selfie near a window or in soft daylight.
• Keep the camera close to eye level.
• Avoid extreme angles.
• Use a neutral expression and one natural smile.
• Remove sunglasses, hats, and heavy filters.
• Use current photos, not old favorites.
• Review the output at small profile-photo size before choosing.

If you are making a LinkedIn profile photo specifically, also check the LinkedIn headshot size guide so the crop works inside LinkedIn's circular profile frame.

Next step

If you want to test this with your own face, start with the AI headshot generator. For profession-specific examples, browse AI headshots by profession, including LinkedIn, business, actor, real estate, and executive profile use cases.