Instruction-based Image Manipulation by Watching How Things Move

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Cao, Mingdeng, Zhang, Xuaner, Zheng, Yinqiang, Xia, Zhihao
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910747827634176
author Cao, Mingdeng
Zhang, Xuaner
Zheng, Yinqiang
Xia, Zhihao
author_facet Cao, Mingdeng
Zhang, Xuaner
Zheng, Yinqiang
Xia, Zhihao
contents This paper introduces a novel dataset construction pipeline that samples pairs of frames from videos and uses multimodal large language models (MLLMs) to generate editing instructions for training instruction-based image manipulation models. Video frames inherently preserve the identity of subjects and scenes, ensuring consistent content preservation during editing. Additionally, video data captures diverse, natural dynamics-such as non-rigid subject motion and complex camera movements-that are difficult to model otherwise, making it an ideal source for scalable dataset construction. Using this approach, we create a new dataset to train InstructMove, a model capable of instruction-based complex manipulations that are difficult to achieve with synthetically generated datasets. Our model demonstrates state-of-the-art performance in tasks such as adjusting subject poses, rearranging elements, and altering camera perspectives.
format Preprint
id arxiv_https___arxiv_org_abs_2412_12087
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Instruction-based Image Manipulation by Watching How Things Move
Cao, Mingdeng
Zhang, Xuaner
Zheng, Yinqiang
Xia, Zhihao
Computer Vision and Pattern Recognition
This paper introduces a novel dataset construction pipeline that samples pairs of frames from videos and uses multimodal large language models (MLLMs) to generate editing instructions for training instruction-based image manipulation models. Video frames inherently preserve the identity of subjects and scenes, ensuring consistent content preservation during editing. Additionally, video data captures diverse, natural dynamics-such as non-rigid subject motion and complex camera movements-that are difficult to model otherwise, making it an ideal source for scalable dataset construction. Using this approach, we create a new dataset to train InstructMove, a model capable of instruction-based complex manipulations that are difficult to achieve with synthetically generated datasets. Our model demonstrates state-of-the-art performance in tasks such as adjusting subject poses, rearranging elements, and altering camera perspectives.
title Instruction-based Image Manipulation by Watching How Things Move
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2412.12087