AI Headshot Guide for LinkedIn and Resume (2026)
What a Good Headshot Actually Is
A professional headshot is not a glamour shot. It is a photo that lets someone who has never met you recognize you at a meeting, judge that you look professionally serious, and form a positive first impression in 2 seconds. Recognizability is the load-bearing requirement. Three visual properties make a good headshot. First, the face is sharp, well-lit, and approximately centered. Second, the framing is tight — head and upper shoulders, with comfortable margin around the face. Third, the background is uncluttered: a solid color, a soft blur, or a tasteful environmental wash. What does not matter much: whether you smile (open smile, closed smile, neutral are all fine if natural), whether you wear a suit (industry-context matters more than formality), whether the photo is in color (almost always color in 2026). What is fatal: out-of-focus eyes, harsh shadow on half the face, distracting background, or a photo that doesn't look like the in-person you.
LinkedIn Specs in 2026
LinkedIn profile photo requirements as of 2026: Dimensions: 400x400 minimum, 800x800 recommended for retina displays. LinkedIn supports up to 7680x4320 but anything beyond 1500x1500 is downsampled. Aspect: 1:1 square. LinkedIn auto-crops uploads to square in a circle mask. Center the face accordingly and leave margin so the circular crop doesn't slice ears or chin. File format: JPEG or PNG. JPEG at 85-90% quality is fine and faster to upload. File size: 8MB maximum. Most properly sized headshots are 200-800KB. Frame composition for the circular crop: face occupies roughly the center 60% of the square. Eyes are positioned along the upper third (rule of thirds). Top of head is just inside the circle; chin sits well above the bottom edge. LinkedIn also shows your photo as a thumbnail (60x60px or smaller) in comments, search results, and connection lists. Test your photo at small sizes by previewing in incognito on a mobile browser — if it reads as just a blob of color or you can't make out the face, choose a tighter crop.
Resume Photo Conventions
Whether to include a photo on a resume depends heavily on country and industry. US, UK, Canada, Ireland, Australia: do NOT include a photo on a resume. It creates bias concerns and many ATS systems handle photo-bearing resumes poorly. Recruiters are trained not to evaluate photos. Most of continental Europe (Germany, France, Spain, Italy), Latin America, much of Asia, Middle East: photos are conventional and often expected. Local norms vary; check the specific country. Where a photo IS included, the conventions: passport-photo-style framing (head and shoulders), 35x45mm or comparable, neutral background (white or light gray), conservative wardrobe (jacket or buttoned shirt), neutral or warm-professional expression. Typically placed in the upper right of the resume. AI is useful for cropping a casual photo into the resume format, cleaning the background to white or light gray, and standardizing the lighting. It is not useful for inventing a photo you don't have — recruiters notice.
Capturing a Good Source Photo
If you don't have a recent headshot, here's how to capture one in 10 minutes. Location: indoors, near a window. North-facing window in mid-morning or mid-afternoon gives soft, even directional light. Avoid direct sunlight (creates harsh shadows and squinting). Background: stand 4-6 feet in front of a plain wall. Any solid color works; off-white, light gray, or a deep neutral like navy are all fine. The AI editor can swap this if needed. Wardrobe: solid colors, no busy patterns. Avoid pure white shirts (they blow out under window light). Avoid logos. Jacket optional but adds structure. Camera: use a phone with the rear (not front) camera — higher resolution and better dynamic range. Prop the phone at eye level on a stack of books or a tripod. Use the 2x telephoto if you have it (avoids the wide-angle face distortion). Framing: head and shoulders visible, eyes roughly 1/3 from the top of the frame, comfortable margin all around. Take 20-30 frames with small expression and pose variations. You'll edit the best 1-3. Expression: think about something pleasant for a beat before the shot. A relaxed micro-smile reads as warm and confident. A forced grin reads as forced.
What AI Should Do (and Not Do)
What AI editing should do for a headshot in 2026. Clean and replace the background. Swap an apartment kitchen for a neutral gray or a soft office blur. This is the most useful single edit. Even out the lighting. Lift shadows on one side of the face, reduce a hot spot. "Brighten the left side of the face slightly, reduce the hot spot on the forehead, keep everything else identical." Fix wardrobe issues. Recolor a too-bright shirt to a more conservative tone. Remove a wrinkled collar. Add a jacket over a t-shirt (works most of the time, occasionally introduces small artifacts at the collar). Tighten the crop and frame. "Crop to 1:1 with my face filling 60% of the frame." Remove blemishes or distractions. Stray hair, a single blemish, a piece of food on a tooth. Light corrections only — over-edited skin reads as fake. What AI should NOT do. Generate a headshot from a non-headshot photo (a full-body party shot transformed to a corporate headshot). The output usually looks plausible but doesn't match the in-person you. Change your apparent age, weight, hairstyle (beyond minor cleanup), or facial structure. These are identity changes. Co-workers and recruiters notice. Replace your face with a generated face. Tools that offer "AI-generated professional headshots" from a few selfies sometimes drift into inventing a similar-but-not-you face. Use the original photo as the base.
Prompt Patterns That Work
Effective headshot prompts share a structure: state the edit, name what to preserve, and constrain the lighting. Background replacement: "Replace the background with a soft out-of-focus office interior in neutral gray-blue tones. Keep my face, expression, hair, clothing, and pose exactly as in the original. Match the lighting direction." Lighting evenness: "Even out the lighting on the face. Lift the shadow on the left side by about 30%. Reduce the bright highlight on the forehead. Do not change my face, expression, or clothing." Wardrobe upgrade: "Replace my t-shirt with a dark navy collared shirt. Keep my face, expression, hair, pose, and the background identical." Crop and frame: "Crop to 1:1 square. Position my face slightly above center, eyes on the upper third line. Maintain a comfortable margin around the head." Full cleanup combo: "Clean up this headshot for LinkedIn. Replace the background with a soft neutral gray. Even out the facial lighting. Tighten the crop to 1:1 with my face filling about 60% of the frame. Do not change my facial features, expression, hair, skin texture, or apparent age." Notice the constraint pattern in every prompt: an explicit list of what should not change. This is what fights identity drift.
Preserving Your Identity
The biggest risk in AI headshot editing is identity drift — output that looks 95% like you but is subtly someone slightly different. Co-workers and recruiters absorb this faster than you might expect. The drift comes from how diffusion models work. They reconstruct the entire image during editing, including the face. Without an explicit identity-preservation constraint, the model nudges your features toward an average attractive face. The result is technically still recognizable but reads as a vague, off version of you. Four controls reduce this. Use an explicit identity constraint in every prompt: "Do not change my facial features, expression, skin texture, hair, or apparent age." Most current models respect named constraints. Limit the scope of each edit. Background change is safe. Lighting tweak is safe. Cropping is safe. Combined edits ("clean this up for LinkedIn") in one pass have higher drift risk than three separate small edits. Compare against the original carefully. Open the original and the edit side-by-side. Look at the eyes, the bridge of the nose, the jawline. If anything looks different, redo with tighter constraints. Use tools that explicitly support identity preservation. Some AI editors offer an "identity lock" mode that uses face-embedding to constrain the diffusion process. Where available, use it for any photo where recognizability matters.
Wardrobe and Background Choices
Wardrobe for a LinkedIn headshot in 2026 follows the industry rule. Finance, law, consulting, corporate: jacket, possibly with tie. Tech, startups, creative: collared shirt or a smart casual top. Healthcare: clinical white or scrubs context, or a smart professional top. Default if uncertain: dark solid color, no logos, no busy patterns. Avoid pure white shirts on a white background (the figure-ground separation disappears). Avoid loud branded clothing. Avoid neckline so low that it dominates the frame. Background choices. Solid neutral (light gray, beige, deep blue) is the safest. Slightly out-of-focus indoor environment (office, library, clean apartment wall) reads warmer and more human. Outdoor backgrounds (foliage, brick wall) work for creative industries. What to avoid in backgrounds: anything with text or a recognizable logo (a Starbucks behind you is a distraction and a brand association), other people, anything that looks like a personal-life clue (a kitchen, a bed). "Where is this person?" should be answerable as "in a professional setting" without further specifics.
Three Worked Headshot Edits
Example 1: Phone selfie at home, kitchen background, harsh overhead light. Subject is a 35-year-old engineer. Prompt: "Replace the kitchen background with a soft out-of-focus indoor office in muted gray-blue tones. Even out the facial lighting — lift the shadow on the lower left, reduce the bright spot on the forehead. Crop to 1:1 with the face centered slightly above the middle. Do not change facial features, expression, hair, or apparent age." Result in 1-2 passes: professional-looking headshot suitable for LinkedIn. Example 2: Vacation photo, full body, casual t-shirt, palm trees in background. Prompt: "Crop this to a head-and-shoulders headshot. Replace the t-shirt with a dark navy collared shirt. Replace the background with a soft neutral gray. Even out the facial lighting. Do not change my face, expression, hair, or apparent age." Result: serviceable LinkedIn photo from a vacation shot. Quality is slightly lower than a from-the-start headshot capture because the cropped resolution is lower. Example 3: Old corporate headshot from 4 years ago, person now wears glasses, slightly different hair. Recommendation: don't edit this — re-shoot. Trying to age-update a headshot in AI introduces identity drift. The five-minute investment in a fresh capture pays off. Where re-shoot is impossible (the person is recovering from surgery, etc.), use the old photo with a current date disclaimer in the profile bio.
Capture a head-and-shoulders source photo
Phone rear camera, north-facing window light, 4-6 feet from a plain wall.
Pick the best 1-3 frames
Eyes sharp, expression natural, head not tilted weirdly.
Run a single AI cleanup pass
Background + lighting + crop, all with explicit identity preservation constraint.
Compare against original, check identity drift
Side-by-side. Eyes, nose, jawline. Redo if anything changed.
Export 800x800 sRGB JPEG
85% quality. Upload to LinkedIn.
Ethics and Disclosure
Where do AI headshot edits cross the line? The principle is recognizability. Light retouching (lighting, background, wardrobe color, crop, minor blemish) is fine and standard practice — professional headshot photographers have done this kind of retouching for decades. What crosses the line: changing your apparent age substantially, changing your weight, generating a face that isn't yours, removing visible identifying features (a scar, distinctive features) that a colleague would expect to see. LinkedIn's terms of service require profile photos to be of you, the account holder. Heavily AI-generated faces violate this. As of 2024-2025, LinkedIn has begun automated detection of obviously generated profile photos and warns or removes them; expect that detection to improve through 2026. A practical test: would you be comfortable if your CEO or a potential employer saw both the original photo and the edited one? If yes, you're in the safe zone. If you'd rather they not compare, you've crossed it.
Frequently Asked Questions
Polish your LinkedIn headshot
Upload a real photo of you. Clean the background, even the lighting, crop to LinkedIn spec. No identity drift.
Edit My Headshot
Free to try · no sign-up · photos are never shared
Release to upload