Bernini-R: Latent Semantic Planning Video Renderer by ByteDance

Bernini-R is ByteDance's video diffusion renderer from the Bernini framework, repackaged with multiple precision options including fp16, fp8, and int8 quantizations.

B

Bernini-R

Video GenerationDiTByteDanceComfyUI

ByteDance's Bernini Renderer (Bernini-R) — the DiT-based video diffusion component of the Bernini framework. Combines an MLLM-based semantic planner with this renderer for video generation and editing. Available with high and low noise schedule variants in fp16, fp8, int8, and MXFP8 quantizations.

Hugging FaceGitHubProject PagePaper
DeveloperByteDance
Release Date2026-06
ArchitectureDiT-based Video Diffusion Renderer (WAN 2.1/2.2 backbone)
LicenseApache-2.0
Variantswan2.1 (1.3B) · wan2.2 high/low noise · fp16/fp8/int8/MXFP8
CapabilitiesText-guided video generation and editing

Guides and workflows related to this model series.

No articles found.