Use the same principles as photography and film — lead lines, value contrast, framing. In a game environment, composition must work from multiple camera angles and player positions, not a single viewpoint. Place your focal points along likely movement paths and use lighting to reinforce them.

Design modules at a grid scale that matches the level design requirements. Build in variation at the dressing layer rather than the kit layer — a small number of modular pieces with a larger number of props creates better variety than many unique kit pieces. Identify repetition risks before building and design breakup solutions into the kit from the start.

Define a palette of 3-5 primary surface types and vary them through roughness, age, and damage rather than new base materials. Establish a value range — no pure white or pure black — and keep specular response consistent within the same material family. Tiling textures need visible breakup through detail normals or vertex-painted blends.

Build the environment to work across the full lighting range — avoid relying on directional light to carve out details that need to read at night or in overcast conditions. Use emissives and fill lights to maintain readability across time-of-day cycles. Test your scene at multiple lighting states before final sign-off.

LOD 0 at full production detail, built for the closest player interaction distance. Subsequent LODs aggressively remove detail that doesn’t read at increasing distance — interior geometry, bevels, secondary normals detail. LOD 4+ should be single draw call impostor meshes. Test each LOD at its actual trigger distance in-engine, not at a fixed render distance in your modelling package.

Object placement, wear patterns, light sources, and path logic all carry story information. A broken window with glass spread inward tells a different story than outward. Cigarette ends clustered near a specific spot tell you someone waited there repeatedly. Every prop placement decision is either confirming or contradicting the story the level is trying to tell.

Be honest about the initial reaction and then describe the professional response — understanding the reason for the change (scope, art direction pivot, performance), communicating the timeline impact, and executing the rework without dragging. The answer interviewers want: you separated your ego from your work and delivered.

Describe specifically which engine and what you can own end-to-end: asset import, material instance setup, light placement, post-process volume setup. Studios increasingly prefer environment artists who can dress and light their own work in engine rather than handing it over. Name the engine and be specific about your depth.

Kit style guides, material master documentation, and regular cross-artist reviews. Establish a reference environment that sets the quality and density bar and review new areas against it. Naming conventions and a single prop library prevent the duplication that fractures visual consistency.

Be specific and honest. Choose something real — a texture that tiles poorly at certain angles, a modular kit that required too many unique pieces to achieve visual variety. Showing you can identify your own work’s limitations is more reassuring to interviewers than claiming it’s finished.

Prioritise the primary read surface — whatever the player will see most at game camera distance. Bake detail into the normal map to recover surface information without additional texture budget. Reduce resolution on surfaces facing away from likely player viewpoints. Consider a texture atlas for secondary surfaces that share material families.

A full scene breakdown — not just the hero shot. Wireframes, material breakdowns, in-engine screenshots at game camera distance, and process notes. Studios want to see that you can produce the environment, not just photograph it. A weak portfolio shot of a beautiful environment is less useful than a documented modular environment that shipped in a game.