Papers/2610.31734
🧪 Test?View on arXiv

DirCast: Live Esports Directing and Commentary with a Single Streaming Video-Language Model

Not provided in the abstract

multimodalstreamingvideo-language modelesports
2610.31734
Builder Relevance
80%
1h ago

Abstract

DirCast is a model that simultaneously directs and comments on live esports broadcasts using raw camera feeds.

Reality Card

Core Claim

DirCast outperforms both rule-based and zero-shot directors, and its commentary is significantly better than that of existing streaming commentators.

Method / Result

DirCast's commentary wins 29.2% of comparisons against GPT-4o under the Proact-VL protocol.

Limitations

The dataset is limited to 27 professional Dota 2 matches, which may affect generalizability.

Paper to code

Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.

No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.
← Back to all papers