The tool sequence designers actually used, by discipline. See how the stack shifts as AI takes more of the execution.
Art direction is the job. The director sets reference, constraints, and rejection criteria. Agents produce variations. Humans pick and finish.
57 synthesized monthsin the data layer. Stage breakdowns (Starter / Scaler / Titan) are available for 2026 only — earlier months show under the All segment but won’t appear under stage filters until the design-context pipeline runs further back.
Graphics splits hard. Volume work goes to agents inside Adobe. Premium work goes to a smaller cohort of hand-craft specialists charging more for being visibly non-AI.
Director sets a visual language, agents generate variations across format, designer picks and edits. Video is a default output, not a specialty.
Art direction by prompt and reference. The designer curates, edits, and decides. Output is high; the bar is taste.
Agents generate variants at volume. The designer's job is selection, editing, and holding a consistent visual point of view. Craft-led illustration gets a premium tier as backlash to flat AI output.
Art director sets a visual reference and constraints. Agent produces variants at volume. Director picks, edits in-canvas, ships.
Graphics designers in Q2 2026 batch-generated visual directions in Midjourney or Firefly — sometimes 50–100 images in a session — then compressed that into a tight 3–5 direction shortlist for client review. The craft value was in the curation and iteration: knowing which generation to extend, which to kill, and how to steer the model toward brand-consistent output rather than generic aesthetics. Motion work (Runway) was increasingly expected even in still-image briefs, as clients wanted social-ready loops alongside print deliverables.
Graphics and generative-image designers ran prompt-to-asset pipelines through Midjourney v7 and Firefly 4 for hero imagery and illustration, then used Krea AI for real-time style refinement and Runway Gen-3 for short motion loops. The production ratio inverted: a single designer could output a week's worth of raw visuals in a morning, meaning curation, retouching for brand consistency, and motion post-production became the majority of billable hours. Buyers increasingly valued art-direction documentation — written prompt frameworks and style references — as a deliverable alongside the assets themselves.
Illustrators and generative artists in Q1 were running multi-model pipelines: Midjourney or Krea for initial image generation, Photoshop (now with conversational AI) for compositing and correction, and Runway for motion passes on still assets. The NVIDIA-Firefly partnership announcement signaled that commercially safe, enterprise-grade generative images are coming; for working graphics creatives, the short-term shift is that clients expect more options faster. The edge is art direction — knowing which outputs are worth extending and why.
Graphics and motion designers ran a generate-refine-animate pipeline in Q1 2026: Midjourney or Firefly for foundational image generation and brand moodboarding, Photoshop Generative Fill for client-asset cleanup and expansion, then Runway or Pika for motion layers—animating stills, adding transitions, or compositing short-form video for social. A typical week included 2–3 client briefs resolved as multi-format asset sets (static + motion), with the Midjourney→Pika pipeline being the dominant fast-turnaround route and Runway reserved for higher-fidelity production outputs. The February model rush meant designers ran bake-offs on the same brief across providers before committing a week's direction.
Graphics and generative image designers in Q1 2026 were recalibrating around the open-weight moment — with models like GLM-Image closing the gap on complex in-image text rendering, the workflow moved toward self-hosted or fine-tuned pipelines for brand-sensitive work rather than defaulting to a vendor API. A typical week involved generating a batch of campaign imagery in Midjourney or Firefly, running legibility and brand-compliance checks in Photoshop, and increasingly using Runway for short motion treatments to make static assets feel alive. The new skill premium was prompt engineering for typography-heavy compositions, given that legible in-image text had been generative AI's most visible weakness.