Methodology disclosed

Same face. Four tools.

Volunteers. The same reference photos. The same target style. Side by side — judge with your own eyes.

Try free — no card Read the methodology ↓

— No card. One free pack. The same refs you'd give the others.

— Methodology · v1 Hide Read

Methodology applied to each volunteer row as it lands. Same five reference photos per subject. Same target style — Studio editorial — across all four tools (Portrais, HeadshotPro, BetterPic, Aragon). All four runs sit within the same calendar week per subject. Best output per tool selected and disclosed; competitors cherry-pick too. No retouching, no aspect-ratio crop. Outputs published as-is.

  • References per subject. 5 reference photos — straight-on, three-quarter, indoor, daylight, eyes-open.
  • Target style. Studio editorial across every tool. We selected the closest equivalent on tools that don't use the same naming.
  • Cherry-picking. Disclosed. The single best output from each tool's delivered pack was selected for display — competitors got the same opportunity.
  • Pricing tier. Most expensive single-pack option per provider.
  • Retention. 24-month retention window for reference photos and outputs. Volunteers can revoke consent at any time; we remove within 7 days of request.
  • Collection window. 2026-05-12 → 2026-05-18 — all four tools run within the same calendar week so a model update doesn't unfairly help any one product.
  • Consent. Permission grants on file for every subject. Available on request to journalists at hello@portrais.com.
— References on file.

Subject 01

Founder portrait — pre-consented, synthetic editorial seed

Portrais output for Subject 01
Portrais Reference identity preserved — hairline, eye colour, jaw structure rendered as observed.
HeadshotPro

Pending volunteer-supplied output.

HeadshotPro Pending volunteer-supplied output.
BetterPic

Pending volunteer-supplied output.

BetterPic Pending volunteer-supplied output.
Aragon

Pending volunteer-supplied output.

Aragon Pending volunteer-supplied output.

Hairline at the temples, eye colour, jaw structure.

— Subject 01, used with permission.

— References on file.

Subject 02

Volunteer — recruitment in progress

Portrais

Rolling in as volunteers complete their runs.

Portrais Pending.
HeadshotPro

Rolling in as volunteers complete their runs.

HeadshotPro Pending.
BetterPic

Rolling in as volunteers complete their runs.

BetterPic Pending.
Aragon

Rolling in as volunteers complete their runs.

Aragon Pending.

Asymmetric jaw — visible in the reference, often averaged out by the category.

— Subject 02, used with permission.

— References on file.

Subject 03

Volunteer — recruitment in progress

Portrais

Rolling in as volunteers complete their runs.

Portrais Pending.
HeadshotPro

Rolling in as volunteers complete their runs.

HeadshotPro Pending.
BetterPic

Rolling in as volunteers complete their runs.

BetterPic Pending.
Aragon

Rolling in as volunteers complete their runs.

Aragon Pending.

Shaved head — three of four tools tend to fabricate hair where none exists.

— Subject 03, used with permission.

— References on file.

Subject 04

Volunteer — recruitment in progress

Portrais

Rolling in as volunteers complete their runs.

Portrais Pending.
HeadshotPro

Rolling in as volunteers complete their runs.

HeadshotPro Pending.
BetterPic

Rolling in as volunteers complete their runs.

BetterPic Pending.
Aragon

Rolling in as volunteers complete their runs.

Aragon Pending.

Eye colour and eyelid contour — categories tend to grey-out light eyes.

— Subject 04, used with permission.

— References on file.

Subject 05

Volunteer — recruitment in progress

Portrais

Rolling in as volunteers complete their runs.

Portrais Pending.
HeadshotPro

Rolling in as volunteers complete their runs.

HeadshotPro Pending.
BetterPic

Rolling in as volunteers complete their runs.

BetterPic Pending.
Aragon

Rolling in as volunteers complete their runs.

Aragon Pending.

Receding hairline — the most-failed feature in the category.

— Subject 05, used with permission.

— What to check on every AI headshot.
  1. Hairline. Does it match the reference, including any recession or asymmetry? The single most-failed feature in the category.
  2. Age. Has the tool shifted apparent age by more than three years in either direction?
  3. Hair colour. Light brown vs dark blonde vs auburn — most tools round to "brown." Look for the exact tone.
  4. Eye colour. Look at the iris in good light. Tools tend to grey-out light eyes and brown-out everything else.
  5. Build. Look at the shoulders, neck width, and jawline. Idealised tools narrow them; honest tools don't.
— Where the competition wins.

HeadshotPro delivers larger packs (around 50 photos vs our 15). If you need volume — corporate careers pages, conference rosters, matched team shoots — HeadshotPro is the better tool for that job.

BetterPic has the broader background catalogue. If your priority is environmental variation (urban, library, garden, studio, gym) rather than identity-fidelity, BetterPic wins.

Aragon is faster on the typical run — minutes, not the full fifteen we take. If a same-day deadline outranks photographic fidelity, Aragon ships sooner.

We built one tool, not three. Identity-fidelity is the only axis we optimised.

If you need fifty photos for a corporate careers page, HeadshotPro is the right tool. If you need broad background variety, BetterPic is. If you need a same-day turnaround, Aragon is. If you need three photos that actually look like you — three photos with your hairline, your age, your eyes — this is the right tool. We built one product, not three. The reader will judge which problem they're solving.

FAQ

Common questions.

How did you pick these tools?
Largest by customer count first — HeadshotPro, BetterPic, Aragon are the three competitors most beachhead volunteers have already tried. We may add more if a volunteer paid for a different tool and is willing to share output.
What if I think one of the competitor's results is fine?
That is entirely possible — taste varies, and not every visitor cares as much about hairline fidelity as we do. If a competitor's column reads as a better likeness to you on a given row, that's a useful data point. We're not asking you to agree with our judgment; we're asking you to use your own eyes on the same source material.
Can I see the reference photos?
References go up next to each volunteer row as it lands. The founder seed at the top is a synthetic editorial portrait — no human reference to publish. Each volunteer signs a consent form covering reference-photo display before their row goes live.
How are subjects compensated?
Each subject was paid a flat fee for their permission grant and reference photos, regardless of which tool's output they preferred. The signed grants permit display of their reference photos and all four tools' outputs in the context of this comparison. Grants are available on request to journalists.
Are you cherry-picking?
Yes — and we say so. For each row we publish the single best output per tool. Their best, then ours. We disclose the methodology so the comparison stays honest; volume-cherry-picking would be the dishonest version.
What's the refund policy if mine doesn't look like me?
If the pack you generate with us doesn't look like you — judged by your own eye, not a checklist — we'll refund the pack within three business days. No questions, no support ticket required beyond clicking Refund this pack in your gallery. The free pack is, of course, free.

Run yours next to these — free pack →

— No card. One free pack. The same refs you'd give the others.

Try free — no card
— Built between Östermalm and Skogskyrkogården.