# Visual credits

The project logo is the same original AI-and-game-world mark used by the survey manuscript, keeping the paper, website, and repository under one visual identity.

The game-world illustration used on the website and README was generated specifically for this project. It is concept artwork, not a research result or an image of a functioning system. The README's six role icons and navigation cards reuse the website's role iconography and colors; its venue chips are original interface assets.

The ten images in `survey-map/` are web renditions of the manuscript's six-role overview, six chapter maps, selected-work timeline, and two-page knowledge map. They are explanatory figures, not screenshots of evaluated systems. The website contains raster renditions only; it does not contain the paper PDF.

The research gallery reproduces illustrative material from the following public projects. These images are not covered by the repository's MIT software license. Copyright and other rights remain with their respective creators and owners; inclusion does not imply endorsement.

| Image | Project and source | Context |
| --- | --- | --- |
| `gallery/sophy.webp` | [Gran Turismo / GT Sophy announcement](https://www.gran-turismo.com/us/news/00_4957003.html) · [original image](https://www.gran-turismo.com/images/c/i1F8TIJBDJ0MJ8E.jpg) | Official Gran Turismo 7 racing image. This is not a figure reporting the original 2022 research evaluation. |
| `gallery/gamengen.webp` | [GameNGen](https://gamengen.github.io/) · [source demonstration](https://gamengen.github.io/static/videos/e1m1_t.mp4) | Frame from a model-generated DOOM demonstration, approximately two seconds into the video. |
| `gallery/mariogpt.webp` | [MarioGPT](https://github.com/shyamsn97/mario-gpt) · [source animation](https://raw.githubusercontent.com/shyamsn97/mario-gpt/main/static/example_interactive.gif) | Frame from the project's text-conditioned level-generation animation. |
| `gallery/gamecraft.webp` | [GameCraft-Bench](https://tongxuluo.github.io/gamecraft-bench-website/) · [source demonstration](https://tongxuluo.github.io/gamecraft-bench-website/static/videos/seele02pro__strategy-signal-rail-dispatcher.mp4) | Seele02-pro's Signal Rail Dispatcher example, approximately four seconds into the video. A selected gallery artifact, not a benchmark-wide result. |
| `gallery/nights.webp` | [1001 Nights](https://www.1001nights.ai/) · [source artwork](https://static.wixstatic.com/media/08a4c9_bb44d5b9a3204ea78ca30125df7bfe3d~mv2.png) | Official project artwork, not a player-study screenshot. |
| `gallery/ea-testing.webp` | [EA SEED research announcement](https://www.ea.com/seed/news/cog23-challenges-deploying-rl-agents-game-testing) · [research paper](https://arxiv.org/abs/2307.11105) | Battlefield 2042 helicopter navigation from Figure 3(c), including the original testing overlay. |

The gallery files are WebP conversions of the source-attributed images listed above.

The nine files in `playable/` are screenshots of this project's browser games: *Cloud Sling*, *Sprout Guard*, *Neon Kart*, *Moonlight Garden*, *Aurora Outpost*, *Lulu Snow Night*, *Snow Dragon Rescue*, *Snow Fox Survival*, and *Revised Platform Route*. The first three form the Classic Arcade collection; *Moonlight Garden* and *Aurora Outpost* use AI-generated rule kernels; the remaining four are AI-created or AI-revised prototypes.

For attribution corrections or removal requests, please [open an issue](https://github.com/Eurekaleo/awesome-ai-for-games/issues).