MAGREF
Back

MAGREF
0 1 0

Characterization

ACG design

MAGREF

在传统视频生成中,只提供一张人像参考图,往往会导致模型在生成过程中ID丢失,人物面貌或服饰在每一帧中“逐渐跑偏”。只需一张人像参考图,MAGREF 生成的视频即可精准保留人物身份特征,同时根据多样文本提示,演绎出跨场景、跨风格的动态内容,真正实现“一图千面”的数字表达。MAGREF 的强大之处在于:无论输入是谁、来自哪个时代、风格是否抽象,它都能精准复刻其身份特征,在生成视频中保持一致性,同时根据文本指令渲染出多变的动作、环境和光影氛围。

This model is sourced from an external transfer (transfer address: https://github.com/MAGREF-Video/MAGREF# ),if the original author has objections to this transfer, you can click,
Appeal
We will, within 24 hours, edit, delete, or transfer the model to the original author according to the original author's request

user_ttecxhtm

user_ttecxhtm

Characterization

ACG design

Model Information

Original author:
Yufan Deng
Model Type:
Unet
Basic Model:
WAN2.1

在传统视频生成中,只提供一张人像参考图,往往会导致模型在生成过程中ID丢失,人物面貌或服饰在每一帧中“逐渐跑偏”。只需一张人像参考图,MAGREF 生成的视频即可精准保留人物身份特征,同时根据多样文本提示,演绎出跨场景、跨风格的动态内容,真正实现“一图千面”的数字表达。MAGREF 的强大之处在于:无论输入是谁、来自哪个时代、风格是否抽象,它都能精准复刻其身份特征,在生成视频中保持一致性,同时根据文本指令渲染出多变的动作、环境和光影氛围。

This model is sourced from an external transfer (transfer address: https://github.com/MAGREF-Video/MAGREF# ),if the original author has objections to this transfer, you can click,
Appeal
We will, within 24 hours, edit, delete, or transfer the model to the original author according to the original author's request