"Technical artist" is eight jobs wearing one badge. Nobody does all eight - you'll start as a generalist, gravitate toward two, and get hired for one. Here's the map: what each actually does all day, whether you'd like it, and the exact place to start.
For each specialization: the one-line pitch, a real Tuesday task, the core skills, "you'd love this if...", and where on this site (or beyond) to start. Demand notes are honest generalizations - every studio weights differently. Don't pick from the couch: try the starter link for each one that intrigues you, and notice which one you keep coming back to after dinner.
Make surfaces lie beautifully to the eye at 60fps.
The look-development engine room: master materials, stylized water, dissolves, snow accumulation, fabric sheen - all while staying inside a millisecond budget.
"The AD wants wet streets after the rain system triggers - build a material function every surface can opt into, without doubling permutations."
You've ever lost an evening to "why does the highlight look wrong" - happily. High demand, prices near engineering bands.
Shaders & HLSL section -> Materials 101 -> build one effect end-to-end for the portfolio.
Delete a thousand hours of clicking, one script at a time.
Exporters, validators, batch processors, the glue between DCCs and the engine. The least visible specialization and the one studios quietly can't function without.
"Character exports break whenever someone renames a bone - build validation that catches it at submit time, not at 6pm before the milestone."
Automating a chore gives you genuine joy and you like your users three desks away. Highest demand-to-supply ratio of all eight.
First Pipeline Script -> the Rosetta Stone -> rebuild one of the tools inside a DCC.
Build the puppet so well the animator forgets it's a puppet.
Skeletons, skinning, control rigs, deformation systems, corrective shapes - where anatomy meets linear algebra and both must ship.
"The shoulder candy-wraps at extreme poses. Add a twist-bone chain and correctives - without breaking the 60 animations already published."
You watch dance videos thinking about clavicles. Scarce skills, loyal demand - rigging TAs rarely job-hunt long.
Rigging with Python, then rig and break something bipedal, repeatedly.
Explosions, spells, smoke - art direction at 200 particles per frame.
Real-time effects in Niagara/particle systems, shader-driven motion, sim-to-realtime bakes (VATs, flipbooks). The most immediately gratifying specialization and the most budget-scrutinized.
"The ultimate ability reads as a gray smudge on the Steam Deck. Rebuild it: fewer, smarter cards, shader displacement instead of particle count."
You pause games during explosions. Portfolio is everything here - three great effect breakdowns outweigh any resume.
Recreate one game effect you love in Niagara; document every trick you reverse-engineered. Then the noise generator becomes your best friend.
Don't make the forest. Make the thing that makes the forest.
HDAs, scattering systems, terrain pipelines, buildings-from-curves - content that scales without headcount. The deepest learning curve on the map and the biggest lever once you're up it.
"Level design changed the canyon again. Good thing the rock placement, erosion, and foliage all regenerate from the new spline in eleven minutes."
Systems delight you more than instances; you'd rather grow a tree than model one.
Houdini section -> build a curve-driven fence/pipe/road tool - the classic first HDA.
Everything between "animation approved" and "animation in game."
Retargeting, state machines, ragdolls, cloth and hair sim, animation pipelines. Sits between rigging and gameplay engineering; often shares a desk with both.
"Mocap for the new enemy retargets fine except the fingers, which explode. Find out why the source skeleton disagrees with ours about knuckle twist."
Motion quality bothers you at a physical level and you like debugging things that move. Undersupplied everywhere character games are made.
Rigging first (Maya path), then Unreal's animation blueprints - retarget mocap onto your own rig and fix what breaks.
Make it gorgeous, then make it gorgeous at framerate.
Lighting pipelines, lightmap wrangling, GI systems (Lumen and friends), exposure standards, and the tooling that lets lighting artists iterate without eating a rebake.
"Night levels hitch when the player torches the village. Profile whether it's shadow-casting light count or Lumen updates, then build the debug view that shows lighting artists their own cost."
You take photos of parking garages "for the light." Smaller niche - usually a senior TA's second specialization rather than a first job.
Lighting & Shadows, then relight one scene three ways (noir / golden hour / horror) with the same geometry and profile all three.
All of the above, Tuesday-dependent - the indie and mid-size studio reality.
At a 15-person studio, "the TA" is the shader person, pipeline person, and occasional rigger simultaneously. Broad beats deep here - and it's the best possible first job, because you discover which specialization pulls you.
Morning: fix the Blender exporter. Lunch: why is the water pink on Switch. Afternoon: teach an artist P4. Evening (denied): the AD "just quickly" wants fireflies.
Variety energizes you and "I'll figure it out" is your resting state. See also Indie Dev - generalist TA is the natural indie co-founder skillset.
The whole Start Here path - it deliberately builds a generalist before specializing.
Three honest heuristics. One: the specialization you keep tinkering with at 11pm has already chosen you; this page is just paperwork. Two: if nothing has chosen you yet, pick pipeline - it's the most learnable from scratch, the most in-demand, and it teaches you every other specialization's problems from the inside, so switching later is cheap. Three: your first job will be 60% not-your-specialization anyway. That's not a failure of planning; that's how the map gets filled in.
The Technical Artist