Photography and Imagery · 8 min read
Photography and Imagery in UX
Choosing, directing and treating the images in a product: seeing like a photographer, AI-generated images, grading and blend modes, text over images, and shipping them fast and accessible. With labs.
Seven steps in order, then five labs and a part for people who lead design teams. Start at the beginning, or go straight to the step you need. Every claim links to its study or standard, and every photo is openly licensed and credited.
Step 1 · Why images
- What an image does in a productEvery image in a product has a job, or it is decoration. The five jobs an image can do, what the research says about first impressions and trust, and why a smiling face is not a trust button.
- The photos nobody looks atEye-tracking shows that people skip generic, feel-good photos and study the ones that carry information. How to tell decoration from content, the one-question test, and what to do when a screen 'needs a photo'.
- Photo, illustration or nothingA photo says 'this is real', an illustration says 'this is an idea', and nothing says 'the task comes first'. How to choose between them, where generated images fit, and what public design systems decide.
- Audit the images in your productA one-hour audit that turns the ideas of Step 1 into a list you can act on: every image, its job, its source, its licence and its alt text, with a keep, fix or remove decision for each.
Step 2 · See like a photographer
- CompositionComposition decides where the eye goes first and where it rests. The tools photographers use (the frame, the subject's place, lines, balance and empty space), what the research says about the rule of thirds, and how to leave room for words.
- Light and what it saysLight decides the mood of a photo before anyone notices the subject. Hard and soft light, direction, backlight and low key, warm and cool, and how to choose light that matches what a screen has to say.
- Faces and gazePeople look at faces, then where those faces look. Three eye-tracking studies show that a face looking at the product leads attention to it, and a face looking at you keeps attention on itself. How to use that in a layout, honestly.
- Crop and aspect ratioThe same photo has to work as a wide banner, a tall phone screen and a square card. How aspect ratios change a photo, why cropping from the centre loses subjects, and how a focal point keeps them, in Figma and in CSS.
Step 3 · Art direction and choosing
- The image brief and the shot listMost bad product photos start with a bad request. How to write an image brief that a photographer, a stock search or an image generator can act on, and a shot list that turns it into the photos the product needs.
- Library, shoot or generatedThree ways to get an image: license one from a library, commission a shoot, or generate one. What each costs, what each promises, where each goes wrong, and a simple way to choose.
- One look for the whole productGood photos from different sources rarely look like one product. How to make a set consistent (light, colour, grade, distance, people), how design systems write it down, and how to start an image style guide.
- People in imagesEvery person in a product photo is someone. Who appears and who is missing, how to avoid stereotypes, and the permissions you need: model releases, consent, children, public places and generated people.
- Licences without fearCopyright, public domain, CC0, Creative Commons and library licences, in plain words: what each lets you do, what each asks of you, and how to credit an image so anyone can check where it came from.
Step 4 · AI-generated images
- What changes when the image is generatedA generated image looks like a photo and records nothing. What that changes for trust, for truth and for the work of a designer, where generated images are useful, and where they quietly mislead.
- Bias in image generatorsAsk a generator for 'a CEO' and 'a nurse' and see who comes back. Why generators repeat stereotypes, what the research found at scale, why better prompts don't fully fix it, and how to check your images before you ship them.
- Labels and provenanceHow to tell people an image was generated, and how an image can carry its own history. Label wording, what labels cost, C2PA and Content Credentials, invisible watermarks, and what the EU AI Act asks since August 2026.
- Image libraries and AIGetty, Shutterstock and Adobe Stock took three different positions on AI-generated images. What each one decided, what it means when you search a library, and the questions to ask before you use a generated image you didn't make.
Step 5 · Treating images
- Grade, black and white and duotoneTreating an image starts with its colour. What a grade changes in how a photo feels, when black and white helps and when it hides, how duotone makes different photos read as one set, and how to do all three in CSS without touching the file.
- Blend modes and the maths behind themMultiply, screen, overlay and the rest are not magic presets: each is a short formula applied to every pixel. What the main modes do, the maths in plain numbers, why mid grey disappears in overlay, and how the same modes work in Figma, Photoshop and CSS.
- Texture, grain and noiseGrain makes digital images feel physical, hides banding in gradients and can turn into mush on a phone. What grain and texture do, how to make noise in the browser with an SVG filter, how Figma does it, and how much is enough.
- Masks, clip paths and scrimsShowing only part of a photo, without cutting the file: clip paths for hard shapes, masks for soft fades, and scrims so text can sit on an image and still pass contrast. With the CSS for each and the checks that keep them honest.
- Treatment with a purposeEvery grade, blend, grain and mask should earn its place. How to tell a treatment that helps a photo do its job from one that's decoration, the switch-off test, when treatment hides information, and the extra rules for photos that are records and for generated images.
Step 6 · Text over images
- Measuring contrast on a photoText on a photo has no single background colour, so it has no single contrast number. What WCAG asks for text over images, how to find the spot that decides, and why a component has to pass on its worst photo, not on the one in the mockup.
- Scrims, safe areas and focal pointsHow to make a text-over-image component pass on every photo and every screen: a scrim tuned for the worst image, a safe area where text may go, and a focal point that keeps the subject in the crop. With the CSS for each.
- Images in dark modeA dark theme changes how every image on the page looks: white backgrounds glow, bright photos shout, dark line art disappears. What the guidelines say, what one study found, what's still unknown, and the CSS and HTML to handle each case.
Step 7 · Images on the web
- Formats and compressionJPEG, PNG, WebP, AVIF and SVG: what each is for, what compression does to a photo, and how to choose a format and a quality by looking and measuring, not by habit. With real files and their real sizes.
- Responsive images and art directionOne photo, many screens: how srcset and sizes let the browser download the right file, and how picture lets you change the crop for small screens. What designers need to specify so developers can build it.
- Images and page speedImages are usually the biggest thing on a page and the slowest to arrive. What LCP and CLS measure, why the hero photo decides them, and four habits that make pages faster: dimensions, eager heroes, lazy everything else, and fetchpriority.
- Alt text that helpsAlt text is the image for people who can't see it. How to tell decorative, informative, functional and complex images apart, what to write for each, and why the same photo can need different alt text in different places.
Labs
- Write an image briefA guided lab: pick one product screen, name the image's job, write a one-page brief and a shot list, and find three references. About an hour, with paper or any notes app and a free image library.
- Build a treatment systemA guided lab: take four photos from different sources, design one grade and one grain that make them a set, build it in Figma or any editor, then rebuild it in CSS and compare. About an hour and a half.
- Audit a hero for legibilityA guided lab: take a real text-over-image component, collect its hardest photos, measure the contrast at the worst spot by hand, and fix it with one change that works for all of them. About an hour, with a browser and a free contrast checker.
- Art-direct a responsive heroA guided lab: take one hero photo, set its focal point, plan the crops for phone, tablet and desktop, decide where a different crop is needed, and build it with object-position and picture. About an hour and a half.
- Alt text workshopA guided lab: write alt text for ten openly licensed images, each in a different place, then compare with suggested answers and listen to your own page with a screen reader. About forty-five minutes.
For leads
- An image style guide in your design systemFor leads: how to turn this course into rules your team can use. What an image style guide contains, why it belongs in the design system next to colour and type, how to write rules people follow, and how to keep it alive.
- Running an image libraryFor leads: where a team's images live, what each record must hold, who owns it, and how to keep licences, credits, alt text and generated-image flags with the image instead of in someone's memory. With the IPTC fields that carry them inside the file.
- A team policy for generated imagesFor leads: how to write a short, checkable policy for AI-generated images, covering where they're allowed, where they never are, how they're labelled and recorded, and who decides, with a template you can adapt.
Reference
- Glossary130 terms defined in this course's lessons, A to Z.
- Frameworks10 frameworks and techniques this course's lessons use, by area.
- Library107 books, articles, videos and courses cited in this course, grouped by the step they appear in.
Saved only on this device.