Bernini-R Style Transfer LoRAs: Anime and Real Life Styles by tarn59

news

tarn59 released two style transfer LoRAs for Bernini-R — anime style and real life style — enabling video-to-video style conversion directly in ComfyUI.

tarn59 released two style transfer LoRAs for Bernini-R, bringing anime and real life style conversion to ByteDance's renderer model. Both LoRAs were shared on July 18 and are available under the Apache 2.0 license.

Anime Style Transfer

The Bernini-R Anime Style Transfer LoRA converts video content into anime aesthetic. Trigger word: Anime style transfer.

Files:

FileSize
bernini_r_anime_style_transfer_high_noise.safetensors~300 MB
bernini_r_anime_style_transfer_low_noise.safetensors~300 MB

Demo:

Anime style transfer demo

HF Repo: tarn59/bernini_r_anime_style_transfer_lora

Real Life Style Transfer

The Bernini-R Real Life Style Transfer LoRA converts animation-style content into photorealistic video. This LoRA comes in two variants:

FileSizeNotes
bernini_r_real_life_style_transfer_v3_high_noise.safetensors~300 MBV3 real life style, high-noise schedule
bernini_r_real_life_style_transfer_v3_low_noise.safetensors~300 MBV3 real life style, low-noise schedule

Demo:

Real life style transfer demo

HF Repo: tarn59/bernini_r_real_life_style_transfer_lora

Usage in ComfyUI

Both LoRAs load through the standard ComfyUI LoRA loader at strength 1.0, applied to the Bernini-R model. The trigger words are simple English phrases:

  • Anime: Anime style transfer.
  • Real Life: Real life style transfer.

Background: Bernini-R

Bernini is ByteDance's unified framework for video generation and editing, combining an MLLM-based semantic planner with a DiT-based renderer. Bernini-R is the renderer-only component, fine-tuned from the Wan diffusion renderer, supporting text-to-video, video-to-video, image editing, and render-to-video tasks. These style transfer LoRAs extend Bernini-R with anime and photorealistic style control, enabling video-to-video style conversion directly in ComfyUI.

Both models are based on ByteDance/Bernini-R and are trained as standard LoRA adapters compatible with ComfyUI and the Diffusers pipeline.

Bernini-R Style Transfer LoRAs: Anime and Real Life Styles by tarn59 | ComfyUI Wiki