Chapter 3: Art Test Walkthroughs & Rubric Alignment

Created by Sarah Choi (prompt writer using ChatGPT)

Art Test Walkthroughs & Rubric Alignment

Why rubric alignment wins offers

Shipped games prove how studios judge art: consistency, readability, performance awareness, and documentation quality. Art tests reproduce those pressures in miniature. Candidates who map their work to the studio’s rubric—explicit or inferred—outperform equally talented artists who present beautiful but unsystematic results. This article shows how to reverse‑engineer common rubrics from shipped titles, plan a test day‑by‑day, and present a packet that makes reviewers’ scoring easy.

How to infer a studio’s rubric from shipped work

Start by examining two or three of the studio’s released games. Look for material ID discipline, trim sheet reuse, decal strategies, LOD behavior, damage modules, and customization breadth. These visible choices usually mirror internal scores: Production Readiness, Visual Design, Technical Execution, Communication, and Process. If a title is live‑service heavy with skins, expect rubric weight on modularity and decal‑friendliness. If cinematics drive the brand, expect higher weight on hero readability and camera‑aware composition.

The universal rubric (and common weight ranges)

While weights vary, most vehicle art tests score something like:

  • Visual Design (20–30%): Silhouette strength, proportion, hierarchy, style cohesion with studio IP.
  • Production Readiness (20–30%): Materials/IDs, trim usage, decal strategy, LOD intent, hardpoints.
  • Technical Execution (15–25%): UVs and texel density, naming/pivots, topology and bake (if modeled), performance thinking.
  • Communication & Documentation (15–25%): Orthos with metrics, clear callouts, variant matrix, acceptance criteria.
  • Process & Collaboration Signal (10–15%): Iteration artifacts, rationale, change log, timeboxing honesty.

Write these categories onto your first slide and keep them visible while you work. Design your deliverables to make each box easy to tick.

A day‑by‑day walkthrough (72–96 hours total)

Day 1 — Deconstruct & Decide. Reverse‑engineer the brief into constraints: ID cap, trim approach, hardpoint standard, LOD survival rules, kit scope. Write a one‑page SSOT (single source of truth) with anchors and production rules. Produce silhouettes and three value comps that honor the brief’s gameplay reads. Present one chosen direction with a paragraph of why.

Day 2 — Kit & Metrics. Lock the kit: modules, hardpoints, interface tolerances. Author measured orthos with gameplay metrics (door widths, ramp angles, traverse arcs). Define UV/TD targets and material ID list. Prepare an exploded view showing attachment logic. Keep a change log.

Day 3 — Materials & Presentation. Build trim sheet map and decal plan; outline shader parameters for wear, oxidation, and emissive. Produce a hero render with a quiet plane for typography. Author LOD policy side‑by‑side callouts (LOD0/1/2) with survival statements. Assemble the slide deck with rubric‑visible sections. If modeling is required, timebox a clean LOD0 blockout and a presentable LOD1.

Day 4 (buffer) — QA & Story polish. Run acceptance criteria; ensure IDs ≤ cap, hardpoints named, pivots placed, UVTD within range. Add two variants (role × trim) using the same kit and IDs. Tighten captions for cause→effect reasoning. Deliver on time with a concise handoff note.

Slide deck order that matches how reviewers score

  1. Constraints & Goals (Rubric Overlay). State the rubric categories and how your solution meets each. Mention ID cap, trims, LOD survival, kit scope.
  2. Family & Myth (Silhouette/Value). Three silhouettes, chosen direction, paragraph on anchors and style fit.
  3. Measured Orthos & Metrics. Scaled views with gameplay dimensions and collision guidelines; units declared.
  4. Kit & Hardpoints. Exploded view, dictionary of sockets, interface tolerances, banned swaps.
  5. Materials & Decals. ID list with ranges, trim sheet map, decal strategy, shader parameters.
  6. LOD Policy. LOD0/1/2 callouts with survival statements; expected draw calls or proxy counts.
  7. Variants. Role × trim matrix created from the same kit and IDs; note batching implications.
  8. QA & Change Log. Acceptance criteria, known risks, and dated changes.

Reviewers can now score page‑by‑page without hunting.

Writing acceptance criteria (so QA can ‘pass’ you)

Offer plain tests that mirror shipping pipelines: “Pass if LOD1 preserves prow/canopy/spine anchors; IDs ≤ 6 (Paint, Glass, Rubber, Metal Aux, Lights, Interior); UVTD 512 px/m body, 1024 px/m cockpit; sockets present for exhaust/lightbar/cargo; decals carry all numbers/stripes; variant swaps don’t add IDs; change log updated.” This reduces debate and signals team empathy.

Aligning to AAA vs Indie rubrics

AAA tends to reward documentation depth, shared ID discipline, and modularity that supports skins/damage. Show hardpoint standards, destruction modules, and decal systems. Indie often weights scope literacy and speed: low ID counts, trim‑first detail, and a kit that spawns multiple roles quickly. In both cases, LOD survival rules and clean orthos are non‑negotiable.

Reverse‑engineering case notes from shipped games

  • Customization‑heavy racers: Low geometric micro‑detail; livery systems and decals carry identity; panel cadence favors large, skin‑friendly fields. Rubric likely penalizes sculpted ornament that fights skins.
  • Tactical shooters: Panelization aligns with damage and door logic; sockets named for lightbars, bumpers, racks; emissive capped to exposure budgets. Rubric likely rewards predictable detach points.
  • Space sims: Hardpoint/station systems and interior‑exterior continuity drive the look; materials unified for batching. Rubric likely weights kit logic and LOD at extreme distance.

Self‑critique rubric (use before you submit)

  • Design: Are anchors memorable in gray? Is hierarchy clean? Does it fit the IP?
  • Production: IDs capped? Trim/decal strategy clear? Hardpoints sensible? LOD policy written?
  • Technical: UVTD consistent? Pivots named/oriented? Topology/bakes clean if modeling included?
  • Communication: Orthos measured? Captions causal (because→therefore)? Change log complete?
  • Process: Did you show iteration and timeboxing? Known risks called out?

If any answer is “no,” cut detail and fix the rubric gap first.

Common failure modes (and how reviewers phrase them)

  • “Beautiful, but not production‑ready.” (No ID/trim plan; vague LODs.)
  • “Unclear how this assembles.” (No hardpoints or tolerances.)
  • “Will break skins.” (Sculpted ornament in paint zones.)
  • “Great render, weak metrics.” (Orthos lack units/scale.)
  • “Pipeline risk.” (Too many unique materials or hero geo everywhere.) Translate these into corrective actions before submission.

Tactics that quietly score points

  • State units and axes on every orthographic page.
  • Keep a visible change log with dates and impacts.
  • Use one neutral HDRI angle for all comparisons; move the camera, not the lights.
  • Show one slide of rejection: an idea you cut for rubric reasons, with why.
  • Provide a variant that proves the kit—designers love to see extensibility.

Walkthrough example (condensed)

Brief. “Design a law‑enforcement interceptor hovercar.” Rubric overlay. IDs ≤ 6; LOD1 preserves wedge/canopy/spine; decal‑friendly; sockets for lightbar, bullbar, winch. Day 1. Silhouettes → pick wedge‑prow, stepped canopy, dorsal spine. SSOT written. Day 2. Orthos with door width, intake clearance; kit with HP_ROOF_A, HP_NOSE_B, HP_REAR_C; tolerances for bullbar/lightbar. Day 3. Materials: Paint, Glass, Rubber, Metal Aux, Lights, Interior; trim sheet MECH_A; decal atlas for numbers/chevrons; LOD callouts. Day 4. Variants: Patrol (Base), Pursuit (Plus), Riot (Pro); QA page + change log. Submit deck.

Communicating risk like a teammate

Flag constraints up front: “Photo mode requires higher gloss; gameplay uses matte. I tuned paint for gameplay and included a photo preset.” Or, “If damage is needed later, panelization aligns with modules; detach seams are ready.” This reassures reviewers you anticipate production realities.

What to model (if modeling is optional)

If time allows, block out LOD0 and a clean LOD1 that demonstrates topology discipline, pivots, and UVTD. Prioritize readable forms and kit interfaces over micro detail. Provide a single Marmoset/GLTF viewer link or frame that matches deck lighting for consistency.

The handoff note that gets forwarded

Two short paragraphs:

  1. Design summary: anchors, kit logic, culture/style fit, and how skins/damage are supported.
  2. Production summary: ID list, trims/decals, LOD survival, hardpoints/pivots, QA checklist. Close with availability and source package organization.

Closing: design to their scorecard

An art test is not only a portfolio piece—it’s a simulation of how you’ll work tomorrow. Reverse‑engineer the studio’s scorecard from shipped games, then build your packet to be easy to grade. When every slide answers a rubric box, reviewers stop guessing and start checking “hire.”

Appendix A — One‑page SSOT template (text)

  • Anchors: 2–3 silhouette invariants + quiet plane
  • IDs: list with acceptable ranges and shader parameters
  • Trims: shared trim sheets with usage rules
  • Decals: atlas/placement logic; blocked zones
  • Hardpoints: names, axes, tolerances, banned swaps
  • LOD: survival statements for LOD0/1/2
  • Metrics: gameplay dimensions; scale/units
  • QA: acceptance criteria; change log link

Appendix B — Reviewer rubric sheet (printable text)

  • Visual Design (0–5): silhouette, hierarchy, IP fit
  • Production (0–5): IDs, trims, decals, LOD, hardpoints
  • Technical (0–5): UVTD, pivots, topology/bake
  • Communication (0–5): orthos, captions, structure
  • Process (0–5): iteration, timeboxing, change log
  • Total /25 (weights applied per studio)

Appendix C — Acceptance criteria (ready to paste)

“Pass if: IDs ≤ 6; LOD1 preserves anchors; UVTD 512 px/m (body), 1024 px/m (cockpit); hardpoints HP_* named/oriented; decals carry all numbers/stripes; variants created from same kit/IDs; change log updated with dates/impacts.”