Everything we have published on dev.to

Written for a technical readership: resolution, failure modes, how these models actually behave rather than how they are marketed.

dev.to (67)

• 5 LinkedIn mistakes that filter engineers out before anyone reads the CV
• A photo of your face is not biometric data. A face embedding might be.
• A test harness for image generators that does not fool you
• Almost every comparison in this category earns a commission on your click
• Antes de subir quince fotos de tu cara: los seis documentos que conviene leer
• At forty portraits the metric stops being quality and becomes variance
• Avant de téléverser quinze photos de votre visage : les six documents à lire
• Background removal without a green screen is an underdetermined problem. Here's what the maths is doing.
• Bevor Sie fünfzehn Fotos Ihres Gesichts hochladen: die fünf Dokumente, die es zu lesen lohnt
• Bewerbungsfoto in Deutschland: seit 2006 freiwillig, trotzdem überall
• Building a staff directory: the photo fields you will wish you had added
• Colour management for portfolio sites, and the 4 professions where getting it wrong loses the client
• Cost per usable output is the only pricing metric that survives contact with reality
• Crop ratios and framing conventions across 5 professional photo genres, as a reference table
• Cuatro profesiones donde la foto tiene reglas, y las reglas no se parecen entre sí
• Dein LinkedIn-Profilbild wird bei 48 Pixeln entschieden, nicht auf deinem Profil
• Diagnose before you reshoot
• Evaluating an image tool for 200 people is a different test from evaluating it for 1
• Forty matching portraits is a different problem from forty good ones
• Fotografiar a una empresa entera es un problema de agenda disfrazado de problema de fotografía
• Four categories of tool that all call themselves the same thing
• How AI headshot generators actually work: LoRA, 15 selfies, and the overfitting problem
• How to read a vendor case study when you are the engineer who has to make it work
• How to test whether a generated portrait still looks like the person, without fooling yourself
• I probed GitHub's avatar CDN. Your 4000px profile picture is 3KB by the time anyone sees it.
• In der Medizin ist das Profilfoto Teil der Patientenkommunikation, nicht der Selbstdarstellung
• Is it safe, and the two searches that answer it
• La foto en el CV: obligatoria en España, sospechosa en Londres, y el ATS no tiene nada que ver
• La photo de profil et les applications mobiles : ce qui change entre les deux
• La photo sur le CV en France : usage tenace, obligation nulle, et l'ATS n'y est pour rien
• Latency budgets for batch image generation, and why "20 minutes" is the wrong number to optimise
• Most AI headshot comparisons are testing the wrong thing
• Photographier une entreprise entière est un problème d'agenda déguisé en problème de photographie
• Photographing yourself: the four variables and the fifty that do not matter
• Profilbild, Handy-Apps und Geschäftsfotos: was sich zwischen den drei ändert
• Quatre métiers où la photo obéit à des règles, et où les règles se contredisent
• Qui écrit les comparatifs de cette catégorie
• Quién escribe las comparativas de esta categoría
• Reading this category when almost every source has a commission
• Rolling out headshots for a whole company is a logistics problem, not a photography one
• Rolling out portraits across a company is a queue problem, not a photography one
• Test any generation tool on the hardest face you can find, not the easiest
• The background decides more of your profile photo than your face does
• The checklist I use before sending employee faces to any third party
• The checklist that comes before any quote
• The objections, and why most of them are correct
• The reader is running a classifier and you are choosing the features
• The resolution maths behind AI headshots, and why most tiers are sized wrong
• The same portrait at 5 sizes: what breaks at 1200px, 400px, 96px, 48px and 15mm of print
• The specifications decide more than the advice does
• Three fields where the portrait follows rules that contradict each other
• Tres categorías de herramienta que se llaman igual y no lo son
• Tu foto de LinkedIn se decide a 48 píxeles, y tú llevas semanas mirándola a 400
• Votre photo LinkedIn se joue à 48 pixels, et vous la regardez à 400 depuis des semaines
• Warum dein Handy-Bewerbungsfoto deine Nase größer macht, und wie du das löst
• Was die kostenlose Version wirklich taugt, und die eine Sache, die sie nicht kann
• Wer die Vergleiche in dieser Kategorie schreibt
• What a professional portrait is for, and what it cannot claim
• What "enterprise" actually means when a small tool sells it to you
• What generated portraits are genuinely good at
• What "professional grade" actually adds
• When generated portraits are the wrong tool, from someone who sells them
• Where to get one, ranked by how much trouble it is
• Why companies buy this, in the order the reasons actually decide it
• Your avatar has six render targets and you have tested one
• Your CV photo doesn't break the ATS. The layout you built around it does.
• Zoom, Teams, Meet and Slack all crop your avatar differently. Here are the numbers.

A single index covering every platform is at the BetterPic index. The portrait pages themselves are on BetterPic.