Visual Design & Composition

01Rework one cluttered flyer or web hero into three content-driven alternatives: one using a column grid, one using a different structure, and one that deliberately breaks the grid. Define the intended reading order, then compare scanning and message recall with three people.

02Design three poster variants for the same event: optimize one for urgency, one for trust, and one for delight. Run a five-second recall test with three people, then select the variant that best supports the intended response.

03Take one layout and produce three responsive variants (mobile, tablet, desktop) keeping hierarchy intact as the grid reflows.

01Photograph the same subject to communicate three intentions - calm, tension, and scale. Use a composition technique only when it supports the intention, and include one frame that deliberately breaks a familiar convention.

02Shoot the same still-life three ways - changing only the light direction (front, side, back) - and write what each lighting choice does to mood and form.

03Demonstrate exposure control: shoot one scene at three deliberate settings (motion-blur vs frozen, or shallow vs deep focus) and explain the tradeoff.

01Design a 6-icon set for one domain (e.g. weather) on a consistent grid and keystroke weight, then test legibility at 24px and 16px.

02Reverse-engineer three Material Symbols icons by redrawing their keylines, stroke logic, and optical balance, then design one new symbol that belongs to the same family.

03Take one ambiguous icon (users guess wrong) and repair it through two redesign passes, testing recognition with 3 people each round.

01Build three button-state microinteractions (hover, active, loading) with intentional easing and duration, and document the timing values.

02Recreate one well-known microinteraction (e.g. the like-button burst or a toggle) in code or prototype, matching its feel as closely as you can.

03Add a reduced-motion variant: take one animation and provide a prefers-reduced-motion fallback, proving it respects accessibility settings.

01Take one flat, underexposed photo and do a full non-destructive correction pass (exposure, white balance, contrast, crop), keeping a layer/history record.

02Retouch a portrait: remove blemishes and even skin tone using healing/clone while preserving texture (no plastic skin).

03Process one RAW product shot for a clean white background, fixing color cast and exposure, and write down the recipe so it is repeatable across a set.

01Lay out a double-sided A5 flyer with correct bleed (3mm), safe margins, and CMYK color, then export a print-ready PDF/X.

02Build a one-page prepress checklist (bleed, color mode, resolution, fonts embedded, overprint) and use it to catch errors in a deliberately broken file.

03Repair an RGB, low-res, no-bleed file into a print-correct version and document each fix and why it matters.

01Typeset one long-form article using a modular type scale, defining sizes/line-heights for h1-h3, body, and caption, then justify each choice.

02Pair two typefaces for a poster and prove the pairing works by showing it at display and body sizes; explain the contrast (serif/sans, weight, mood).

03Diagnose a typographically broken layout by naming observable failures in hierarchy, scanning, readability, and wrapping. Make two revisions, then test which version better supports a stated reading task.

01Design one product screen for a named user, task, target device, and density. Define a spacing scale that supports that context, then show default, hover, keyboard-focus, disabled, loading, and error states for its interactive components.

02Reverse-engineer a screen from a well-known app: infer its spacing tokens, type hierarchy, components, states, and responsive behavior. Adapt that system to different content and add error and empty states.

03Choose a clunky interface, name one target user and task and one observable failure (missed action, slow scan, or input error), then create two redesign alternatives and test both with three people.

01Produce an AI-assisted visual direction board for a brief (mood, palette, 6-8 generated frames) with a written prompt/process log and a human critique of each frame.

02Take one AI-generated image and finish it by hand - fix anatomy/text/edges, composite, and color-grade - then write a rights/source note covering tool, model, and licensing.

01Build a mini identity for a fictional brand: logo, color, type, and one application (business card or social avatar), tied to a one-line brand strategy.

02Redesign one weak existing identity (before/after) and write the rationale connecting each visual change to a brand value or audience need.

03Template it: produce a reusable mini-brand starter (logo lockup slots, color tokens, type roles) others can fork, plus a 1-page usage guide.

01Build the smallest real design system: a one-page style guide (color tokens, type scale, spacing) plus three working components with states, and a contrast report.

02Document the system in Storybook (or a shared doc) so each component has usage rules, props/variants, and a do/don't, making it teachable.

03Stress-test the system: hand the file to one person, have them build a new screen using only your tokens/components, and log every gap or ambiguity they hit.

01Illustrate one flat-style scene or character built entirely from clean vector shapes and a limited palette, mastering the pen tool and boolean operations.

02Build a small reusable vector asset kit (e.g. 8 spot illustrations on a shared style) others can drop into projects, exported as optimized SVG.

03Repair a messy vector file (overlapping nodes, non-optimized paths, off-grid points) into clean, production-ready artwork and note what you fixed.

01Complete a structured calligraphy drill sheet (basic strokes, then a full lowercase alphabet) and photograph the progression across pages.

02Letter one short phrase (3-5 words) by hand, then digitize it as clean vectors, refining curves and consistency.

01Composite a subject into a new background so light direction, color temperature, and edges all match; use masks and adjustment layers non-destructively.

02Create one surreal/concept composite from 3+ images and document your blending recipe (masks, color match, shadows) as a teachable mini-tutorial.

01State one audience question and intended takeaway for a real dataset, then build three honest encodings in Datawrapper. Test which version gives three people the best answer accuracy and completion time, then select the chart that is fit for that goal.

02Design a single-topic infographic (one dataset, one clear takeaway) combining a chart, a stat callout, and a short narrative, on a grid.

03Repair a misleading chart (truncated axis, wrong type, dual-axis trick) into an honest version and annotate exactly what was deceptive and why.

01Design a simple carton (e.g. a tuck-end box) on a real dieline with bleed and fold marks, applying brand graphics across all panels.

02Validate the design in 3D: fold a paper mockup (print + cut + fold) or build a quick 3D box in Blender to check how graphics wrap and read on shelf.

01Light one product model in a three-point or HDRI studio setup, set realistic materials, and render a hero shot; document lighting positions and material values.

02Produce a small case study: render the same product under three lighting moods (studio, dramatic, soft daylight) and explain the tradeoffs and render-time costs.

01Paint one original illustration from a value study through to finished color, using layers for blocking, rendering, and adjustments in Krita.

02Build a small custom brush set and a process timelapse for one piece, making your technique teachable (and pre-empting AI-plagiarism doubts with proof-of-process).

01Design a coherent display typeface for a limited character set (uppercase A-Z + digits), keeping consistent stems, terminals, and proportions, then space and export a working .otf in Glyphr Studio.

02Do a spacing/kerning pass on a problem word (e.g. AVA, To, Wa), adjusting side bearings and kern pairs, and document the before/after metrics.