@inproceedings{anantha-ramakrishnan-etal-2026-anchor,
title = "{ANCHOR}: {LLM}-driven Subject Conditioning for Text-to-Image Synthesis",
author = "Anantha Ramakrishnan, Aashish and
Huang, Sharon X and
Lee, Dongwon",
editor = "Liakata, Maria and
Moreira, Viviane P. and
Zhang, Jiajun and
Jurgens, David",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-acl.30/",
pages = "618--638",
ISBN = "979-8-89176-395-1",
abstract = "Text-to-image (T2I) models have achieved remarkable progress in high-quality image synthesis, yet most benchmarks rely on simple, self-contained prompts, failing to capture the complexity of real-world captions. Human-written captions often involve multiple interacting subjects, rich contextual references, and abstractive phrasing, conditions under which current image-text encoders like CLIP struggle. To systematically study these deficiencies, we introduce ANCHOR, a large-scale dataset of 70K+ abstractive captions sourced from five major news media organizations. Analysis with ANCHOR reveals persistent failures in multi-subject understanding, context reasoning, and nuanced grounding. Motivated by these challenges, we propose Subject-Aware Fine-tuning (SAFE), which uses Large Language Models (LLMs) to extract key subjects and enhance their representation at the embedding-level. Experiments with contemporary models show that SAFE significantly improves image-caption consistency and human preference alignment, serving as a practical and scalable solution. The dataset and code will be released upon publication."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="anantha-ramakrishnan-etal-2026-anchor">
<titleInfo>
<title>ANCHOR: LLM-driven Subject Conditioning for Text-to-Image Synthesis</title>
</titleInfo>
<name type="personal">
<namePart type="given">Aashish</namePart>
<namePart type="family">Anantha Ramakrishnan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sharon</namePart>
<namePart type="given">X</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dongwon</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Maria</namePart>
<namePart type="family">Liakata</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Viviane</namePart>
<namePart type="given">P</namePart>
<namePart type="family">Moreira</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiajun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Jurgens</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">San Diego, California, United States</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-395-1</identifier>
</relatedItem>
<abstract>Text-to-image (T2I) models have achieved remarkable progress in high-quality image synthesis, yet most benchmarks rely on simple, self-contained prompts, failing to capture the complexity of real-world captions. Human-written captions often involve multiple interacting subjects, rich contextual references, and abstractive phrasing, conditions under which current image-text encoders like CLIP struggle. To systematically study these deficiencies, we introduce ANCHOR, a large-scale dataset of 70K+ abstractive captions sourced from five major news media organizations. Analysis with ANCHOR reveals persistent failures in multi-subject understanding, context reasoning, and nuanced grounding. Motivated by these challenges, we propose Subject-Aware Fine-tuning (SAFE), which uses Large Language Models (LLMs) to extract key subjects and enhance their representation at the embedding-level. Experiments with contemporary models show that SAFE significantly improves image-caption consistency and human preference alignment, serving as a practical and scalable solution. The dataset and code will be released upon publication.</abstract>
<identifier type="citekey">anantha-ramakrishnan-etal-2026-anchor</identifier>
<location>
<url>https://aclanthology.org/2026.findings-acl.30/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>618</start>
<end>638</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ANCHOR: LLM-driven Subject Conditioning for Text-to-Image Synthesis
%A Anantha Ramakrishnan, Aashish
%A Huang, Sharon X.
%A Lee, Dongwon
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Findings of the Association for Computational Linguistics: ACL 2026
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-395-1
%F anantha-ramakrishnan-etal-2026-anchor
%X Text-to-image (T2I) models have achieved remarkable progress in high-quality image synthesis, yet most benchmarks rely on simple, self-contained prompts, failing to capture the complexity of real-world captions. Human-written captions often involve multiple interacting subjects, rich contextual references, and abstractive phrasing, conditions under which current image-text encoders like CLIP struggle. To systematically study these deficiencies, we introduce ANCHOR, a large-scale dataset of 70K+ abstractive captions sourced from five major news media organizations. Analysis with ANCHOR reveals persistent failures in multi-subject understanding, context reasoning, and nuanced grounding. Motivated by these challenges, we propose Subject-Aware Fine-tuning (SAFE), which uses Large Language Models (LLMs) to extract key subjects and enhance their representation at the embedding-level. Experiments with contemporary models show that SAFE significantly improves image-caption consistency and human preference alignment, serving as a practical and scalable solution. The dataset and code will be released upon publication.
%U https://aclanthology.org/2026.findings-acl.30/
%P 618-638
Markdown (Informal)
[ANCHOR: LLM-driven Subject Conditioning for Text-to-Image Synthesis](https://aclanthology.org/2026.findings-acl.30/) (Anantha Ramakrishnan et al., Findings 2026)
ACL