Liblast 3D asset pipeline (lgm2026)

The Liblast 3D asset pipeline is a workflow and template for developing and (re-)using 3D game assets. The pipeline stems from identifying worrying trends in 3D game asset life cycles. Chase of highest fidelity has been driving game production cost, team size and completion time up, while driving all the other qualities down. This has been excelently put by Isaque Sanches in his 2022 presentation “The Case Against Visual Fidelity.” Ongoing issues: High memory/bandwidth: assets require a lot of storage to produce and reproduce. Simply put model and texture project and shipping file sizes are out of control Artists burn a lot of time on repetitive, automatable tasks. Most work done on 3D assets is not that creative, and can be separated and automated, freeing artist time to be better spent than fixing blemishes in baked normal maps. Copy-paste effect: with traditional pipelines, each asset instance has exactly the the same wear and dirt patterns. This kind of visual repeating pattern is what humans are really good at picking up, and it breaks immersion. Iterating or changing art direction is extremely costly. Simply changing a single detail on a “hero” asset model could require updating (sculpted) highpoly, lowpoly models, updating rigging, re-baking high poly maps, re-generating painted textures, re-painting/drawing parts of textures, updating mesh and texture data in game engine and consequently game builds Scaling fidelity of assets down to accomodate low-performance hardware is very limited. Minimum system requirements are cutting off a large part of the population Our pipeline aims to remedy a lot of the above, making decent visuals cheaper to produce, ship, update, easier to iterate on, pivot art direction and to scale fidelity up or down based on available system performance, broadening your game audience while holistically reducing asset cost. The asset production cost is higher upfront, but after a baseline for your art direction is down, making new assets is becoming much cheaper while fidelity and stylistic coherence remain high. Licensed to the public under https://creativecommons.org/licenses/by/4.0/ about this event: https://pretalx.c3voc.de/lgm-2026/talk/LABVSB/