Summary
WithEveryone is a unified framework for identity-preserving group image generation that can incorporate up to ten reference identities in a single image, addressing the reliability problems that arise when multiple specified people must appear in one scene. The system injects each selected identity as an addressing token, predicts a structured identity-layout plan, and renders that plan as visual condition. Its core objective, Layout-Grounded ID Loss, directly supervises target identities using annotated face regions, avoiding unstable embedding-based face matching. An additional technique, ID Representation Forcing, trains predictions for each identity prior to image composition. On benchmarks with disjoint identities, WithEveryone achieves the highest targeted contextual identity similarity: face similarity improves from GPT-Image-2's 0.462 to 0.499, copy-paste artifacts drop from 0.169 to 0.055, and the model covers 97.3% of requested identities with only a 2.8% duplication rate. Paper: arXiv 2608.20336.
Paper Overview
Field: Computer Vision (CV)
Authors: Hengyuan Xu, Qixun Wang, Yiji Cheng, Miles Yang
Published: 2026-08-22
arXiv: 2608.20336
Abstract (translated)
Identity-preserving image generation becomes increasingly unreliable when a scene needs to include multiple specified people. Beyond preserving each identity, the model must bind each reference to a distinct person and location. This paper proposes WithEveryone, a unified framework for generating group images that include up to ten reference identities.
WithEveryone:
- Injects each selected identity as an addressing token
- Predicts a structured identity-layout plan
- Renders the plan as visual condition
Its core objective,
Layout-Grounded ID Loss, directly supervises target identities using annotated face regions, avoiding unstable embedding-based face matching.
ID Representation Forcing trains predictions for each identity before image composition.
Results
On benchmarks with identity-disjoint references, WithEveryone achieves the highest targeted contextual identity similarity:
| Metric | GPT-Image-2 | WithEveryone |
|---|---|---|
| Face similarity | 0.462 | 0.499 |
| Copy-paste artifacts | 0.169 | 0.055 |
It covers 97.3% of requested identities with only a 2.8% duplication rate.
---
*Auto-collected on 2026-08-22*
This page is an English static mirror generated for search and AI citation.
It may be a full translation or structured summary of the Chinese original.
Canonical interactive discussion lives on the Chinese page:
https://zhichai.net/topic/178633808