PAPER_TITLE

FIRST_AUTHOR_LAST, FIRST_AUTHOR_FIRST; SECOND_AUTHOR_LAST, SECOND_AUTHOR_FIRST

SIGMA-GEN: Structure and Identity Guided Multi-subject Assembly for Image Generation

Oindrila Saha¹, Vojtech Krs², Radomir Mech²,
Subhransu Maji¹, Kevin Blackburn-Matzen^2*, Matheus Gadelha^2*

¹ University of Massachusetts Amherst
² Adobe Research

Paper Code

SIGMA-GEN enhances controllability of text-to-image workflows by allowing users to prescribe both structure and subject identity.

Abstract

We present SIGMA-GEN, a unified framework for multi-identity preserving image generation. Unlike prior approaches, SIGMA-GEN is the first to enable single-pass multi-subject identity-preserved generation guided by both structural and spatial constraints. A key strength of our method is its ability to support user guidance at various levels of precision — from coarse 2D or 3D boxes to pixel-level segmentations and depth — with a single model. To enable this, we introduce SIGMA-SET27K, a novel synthetic dataset that provides identity, structure, and spatial information for over 100k unique subjects across 27k images. Through extensive evaluation we demonstrate that SIGMA-GEN achieves state-of-the-art performance in identity preservation, image generation quality, and speed.

Unified control modality — SIGMA-GEN enables unified control over image generation at varying levels of granularity including 2D, 3D boxes, 3D objects with a single model.

SIGMA-GEN can enable single and multi-subject insertion in one pass.

SIGMA-GEN can leverage different amounts of depth information.

SIGMA-GEN can be used for subject reposing.

SIGMA-GEN can handle mask shapes not seen during training.

SIGMA-GEN can handle different granularity levels of condition in single generation.

SIGMA-GEN can be used with changing styles of subjects using text.

More coming soon

BibTeX

@misc{saha2025sigmagen,
    title={SIGMA-GEN: Structure and Identity Guided Multi-subject Assembly for Image Generation},
    author={Oindrila Saha and Vojtech Krs and Radomir Mech and Subhransu Maji and Kevin Blackburn-Matzen and Matheus Gadelha},
    year={2025},
    eprint={2510.06469},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}