In the Blink of an Eye: Instant Game Map Editing using a Generative-AI Smart Brush

TL;DR

The paper tackles the challenge of efficiently editing high-detail game maps without compromising artistic control. It introduces a hybrid Smart Brush powered by two backbones, BrushGAN and BrushCLDM, coupled with a context-aware, multi-chunk stitching pipeline to produce seamless, high-fidelity textures. Quantitative evaluation on a World of Tanks map dataset shows that BrushGAN delivers sharper, more detailed textures while preserving context, with BrushCLDM offering strong structural coherence; together they enable per-chunk edits in roughly 2–6 seconds, accelerating production workflows. The approach preserves artist agency while delivering AI-assisted automation and demonstrates practical potential for production pipelines, with future work toward richer context and map-from-scratch generation.

Abstract

With video games steadily increasing in complexity, automated generation of game content has found widespread interest. However, the task of 3D gaming map art creation remains underexplored to date due to its unique complexity and domain-specific challenges. While recent works have addressed related topics such as retro-style level generation and procedural terrain creation, these works primarily focus on simpler data distributions. To the best of our knowledge, we are the first to demonstrate the application of modern AI techniques for high-resolution texture manipulation in complex, highly detailed AAA 3D game environments. We introduce a novel Smart Brush for map editing, designed to assist artists in seamlessly modifying selected areas of a game map with minimal effort. By leveraging generative adversarial networks and diffusion models we propose two variants of the brush that enable efficient and context-aware generation. Our hybrid workflow aims to enhance both artistic flexibility and production efficiency, enabling the refinement of environments without manually reworking every detail, thus helping to bridge the gap between automation and creative control in game development. A comparative evaluation of our two methods with adapted versions of several state-of-the art models shows that our GAN-based brush produces the sharpest and most detailed outputs while preserving image context while the evaluated state-of-the-art models tend towards blurrier results and exhibit difficulties in maintaining contextual consistency.

Figures (18)

• Figure 1: Examples of real map environment from World of Tanksgame:wot.
• Figure 2: A WoT map structurally consists of a 2D grid of chunks, usually divided into $10 \times 10$ chunks inside the playable area.
• Figure 3: Each chunk of the map consists of 8 tile masks with their corresponding texture materials. The tile masks define the blending ratios of the materials.
• Figure 4: Examples of extracted map components: The top row shows a subset of tile masks, the second row their corresponding materials, and the third row provides examples of object masks (water, trees, roads, and buildings). The bottom row displays the global AM (left image) as well as a height map (right image).
• Figure 5: Left: Final map visualization obtained by blending the materials based on their tile masks for a chunk (white stippled line). Right: Rough visualization of the dominant materials.