Llama Learns to Direct: DirectorLLM for Human-Centric Video Generation
Source: https://arxiv.org/abs/2412.14484
Structured fields
{
"checked_at": "2026-10-17T00:00:00.000Z",
"usefulness": "An LLM (Llama 3) plans human poses that condition a video renderer; relevant as a 'director plans, renderer executes' split, but the planning is pose-level, not shot-grammar-level.",
"source_type": "paper",
"availability": "available",
"source_owner": "Song et al. (Meta)",
"next_check_at": "2026-10-31T00:00:00.000Z",
"upstream_updated_at": "2025-09-08T00:00:00.000Z"
}Evidence claims
- [upstream_claim] LLM plans human poses that guide UNet/DiT renderers; reports gains in motion, prompt adherence and realism. · locator: abstract source
- [sourced_fact] arXiv non-exclusive license; no code link on the page. · locator: license and links source
- [upstream_claim] Research review verdict at export: reviewed (as of 2026-10-17). Recorded by the research pipeline, not a hub review outcome; evidence.review stays unreviewed until an independent reviewer records one.
Rights
| component | license | hosting | redistribution | modification | training | Evidence and notes |
|---|---|---|---|---|---|---|
| documentation | arXiv-nonexclusive-distrib-1.0 | restricted | restricted | unknown | unknown | Paper only. |