Football’s Stars in Anime Worlds.

Mbappé, Messi and Yamal lead the cast. Blue Lock, Naruto and One Piece dominate the captions, while Seedance is the most-mentioned AI tool. Across Instagram accounts, much of the same visual material appears again and again.

Erling Haaland

A Viking in Norway’s number 9 shirt.

Lionel Messi

One Piece-inspired swordsman, surrounded by red energy.

Lamine Yamal

One Piece’s Luffy, reimagined in Spain’s shirt.

Mbappé leads the cast

Kylian Mbappé appears in more videos than any other identified figure, followed by Lionel Messi and Lamine Yamal. Beyond the leading names, appearances fall sharply: Bruno Fernandes shows up in just three videos.

Videos featuring each figure · select a name to see their faces

Kylian Mbappe

305 videos · 1,204 face observations

See the images ↗
Kylian Mbappe, saved appearance 1
Kylian Mbappe, saved appearance 2
Kylian Mbappe, saved appearance 3

A video can feature more than one person. Original figure ↗

Original player figures
Videos featuring each figure. The ten most frequent are shown separately; the rest are grouped together. A video can feature more than one person. Figure 2 ↗

Three franchises dominate the captions

Blue Lock, Naruto and One Piece account for six in ten anime mentions in captions and hashtags. Only one of those three—Blue Lock—is about football.

Franchise mentions

672 caption and hashtag mentions

Blue Lock195 mentions · 29.0% of the totalSource ↗

Yamal and One Piece are the most common player–franchise pairing: 22 videos feature him and name the franchise in their captions. The pairing does not show whether he takes on a particular character or role.

Player pairings and original franchise figures
Videos where a visually identified player and a caption-named franchise appear together. A pairing does not establish a character role or visual style. Figure 4 ↗

Seedance and Higgsfield lead the tool mentions

Those two names account for 71 of 193 tool mentions. Naming a tool in a caption does not prove it made the video.

Tool mentions

193 caption and hashtag mentions

Seedance37 mentions · 19.2% of the totalSource ↗
Tool origins and original model figures
193 explicit tool mentions. These are caption references, not verified records of which system made each video. Figure 14 ↗

Messi leads in total attention.
Per video, the leaders change.

Videos featuring Messi collect the most views, likes and comments overall. But Vozinha and Rayan Cherki rank among the highest when engagement is averaged per video. Appearing more often gives a player more chances to accumulate attention.

Total views · 10 most frequently appearing figures

Lionel Messi85,587,495 views185 of 253 videos have recorded views.

Averages use videos with recorded values; missing values are excluded. A video can feature several players. Original figure ↗

The same divide shows up among named AI tools. Posts mentioning Seedance lead in total engagement. Fish Audio leads on averages, across just eight videos. A few successful posts can produce a high average.

Compare tool engagement
Total engagement above; per-video averages below. Counts in parentheses show the number of videos naming each tool. Figure 6 ↗

Nearly half the videos share material

Repeated visual material appears in 772 of the 1,548 videos checked for reuse. Those videos form 124 groups; 71 groups span multiple known accounts.

Shared material connects 167 accounts

A dense central network sits alongside smaller groups and isolated pairs. The links show who posted related material earlier and later, without establishing who copied whom.

Each dot is an account. Arrows connect earlier- to later-posting accounts that share detected material; thicker links represent more shared groups. These are not verified copying paths. Figure 8 ↗

Much of the sharing happens quickly. Among 96 groups with enough timestamps, 40 span no more than a day. The longest observed span is 166 days.

Attention drops after the first upload

Median views per day per follower · first observed post = 100%

Second: 29% of the first observed post’s normalized views. Source and confidence intervals ↗

The earliest post in the collection may not be the original. The drop also does not prove that reposting causes lower engagement: accounts, timing and audiences differ.