Written by the BetterPic team. Disclosure: we sell generated headshots, which is relevant here because posing is the one thing a generator cannot do for you.
Search for headshot poses and you get galleries of forty variations. Almost all of them are the same four poses with different arms, and the arms are rarely in frame. Here are the four, what each one signals, and the two variables that matter more than any of them.
Square to camera. Shoulders parallel to the lens, head straight. Reads as direct, formal, a little confrontational at close crops. This is the standard for law, medicine, finance and anywhere authority is the point. It is also the least flattering for most faces, which is a trade the register is asking you to make.
Shoulders angled, head to camera. Turn your body about thirty degrees and bring your face back to the lens. This is the default professional pose and it is the default for a reason: the angled shoulders narrow the frame, the returned gaze keeps the directness, and it flatters almost everyone.
Angled with a slight head tilt. Same body, head tipped a few degrees. Reads warmer and more approachable. Correct for therapy, teaching, hospitality, customer-facing roles. Overdone it reads as ingratiating, and the line is about five degrees.
Leaning in. Weight forward, shoulders toward the lens. Reads as engaged and energetic. Common in tech and in creative work, wrong almost everywhere formal.
That is the whole vocabulary. Everything else in those galleries is a variation on where the arms go, and in a head-and-shoulders crop the arms are not visible.
Chin position, which nobody photographs themselves correctly. Push your chin slightly forward and down. It feels absurd and it removes the softness under the jaw that a lens exaggerates. This single adjustment changes more than any pose does, and it is the thing a good photographer says within the first minute.
Where your eyes are focused. Look at the lens itself, not at the screen or at the photographer. The difference is a couple of centimetres and it is the difference between engaged and distracted. On a phone, that means looking at the small dot, not at your own face.
The camera height. Below eye level looks up your nose; above it makes you look small and is disproportionately recommended to women. Eye level is the option that does not editorialise.
The lens. A wide lens at portrait distance widens the nose and flattens the face, and no pose corrects for it. Step back and use a longer focal length, or shoot wide and crop in.
The expression. This is the hardest variable and it is not a pose. Think of something specific a second before the shutter rather than holding a smile, and take far more frames than feels reasonable — fifty is not excessive.
Generated headshots do not let you choose a pose. The model produces what it produces, and the poses it favours are the two middle ones above, because that is what dominates its training data.
What you can control is the input. A selfie set with varied head angles — slightly left, slightly right, straight on, chin up, chin down — gives the model the information it needs to build your face in three dimensions. A set that is all one angle teaches it one angle, and the output looks flat.
So the posing work moves from the output to the input, and it is still worth doing. There is a free tier of up to 10 images if you want to see what a given set produces before paying, and the longer guide covers what to shoot.
For the framing and crop work afterwards, the free profile picture editor.