Epic Games Gift Wrappings

The egs_solutions.json JSON file contains:

• the riddles (gift wrappings),
• their solutions (the Steam appIDs of the corresponding games)

for the event which took place:

• at the Epic Games store,
• between December 13, 2023 and January 10, 2024.

The calendar was as followed:

• the 1st game was offered on December 13, for a duration of a week,
• the 2nd game was offered on December 20, for a duration of a day,
• the 3rd game was offered on December 21, for a duration of a day,
• ...
• the 16th game was offered on January 3, for a duration of a day,
• the 17th game was offered on January 4, for a duration of a week.

AppIDs of all the games

The appids.json JSON file is a list of the 95800 appIDs of all the games:

• as downloaded on December 27, 2023 via IStoreService/GetAppList,
• with the filter have_description_language=english.

Steam palette

This is the output of the palette extractor for 95249 Steam games.
The 8 most common colors are stored for each game.

Here are the output files:

• filtered_appids.json contains the list of appIDs for which an image could be downloaded,
• filtered_indices.json contains the list of indices to filter all the 95800 appIDs in order to get filtered_appids.json,
• steam_palette_8.pth is a PyTorch tensor of size NUM_GAMES x NUM_COLORS x NUM_CHANNELS, with RGB palettes,
• steam_palette_8_hsv.pth is a PyTorch tensor of the same size, with HSV palettes, converted from RGB palettes,
• steam_palette_8_lab.pth is a PyTorch tensor of the same size, with CIELAB palettes, converted from RGB palettes,
• steam_palette_8_luv.pth is a PyTorch tensor of the same size, with CIELUV palettes, converted from RGB palettes.

where:

• NUM_GAMES is the number of games, here 95249,
• NUM_COLORS is the number of most common colors which were extracted, here 8,
• NUM_CHANNELS is the number of channels in an RGB image, i.e. 3.

Warning

You may notice differences between the stored palettes and palettes which you would compute on the fly on the images.
This is due to:

• img2dataset saving images to the disk as JPG files (lossy compression). cf. the default value for the encode_format flag,
• compared to mediapy reading the images on the fly, without compressing and saving them to the disk.