EUVIMEDEUVIMED entdecken
Crossref

CoherenDream: Boosting Holistic Text Coherence in 3D Generation via Multimodal Large Language Models Feedback

Chenhan Jiang, Yihan Zeng, Dit-Yan Yeung

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (7) · 5369-5377

DOI: 10.1609/aaai.v40i7.37453 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Chenhan Jiang und 2 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Score Distillation Sampling (SDS) has achieved remarkable success in text-to-3D content generation. However, SDS-based methods struggle to maintain semantic fidelity for user prompts, particularly when involving multiple objects with intricate interactions. While existing approaches often address 3D consistency through multiview diffusion model fine-tuning on 3D datasets, this strategy inadvertently exacerbates text-3D alignment degradation. The limitation stems from SDS's inherent accumulation of view-independent biases during optimization, which progressively diverges from the ideal text alignment direction. To alleviate this limitation, we propose a novel SDS objective, dubbed as Textual Coherent Score Distillation (TCSD), which integrates alignment feedback from multimodal large language models (MLLMs). Our TCSD leverages cross-modal understanding capabilities of MLLMs to assess and guide the text-3D correspondence during the optimization. We further develop 3DLLaVA-CRITIC - a fine-tuned MLLM specialized for evaluating multiview text alignment in 3D generations. Additionally, we introduce an LLM-layout initialization that significantly accelerates optimization convergence through semantic-aware spatial configuration. Our framework, CoherenDream, achieves consistent improvement across multiple metrics on TIFA subset.As the first study to incorporate MLLMs into SDS optimization, we also conduct extensive ablation studies to explore optimal MLLM adaptations for 3D generation tasks.

Zitieren

APA

Jiang, C., Zeng, Y., & Yeung, D. (2026). CoherenDream: Boosting Holistic Text Coherence in 3D Generation via Multimodal Large Language Models Feedback. Proceedings of the AAAI Conference on Artificial Intelligence, 40(7), 5369-5377. https://doi.org/10.1609/aaai.v40i7.37453

Vancouver

Jiang C, Zeng Y, Yeung D. CoherenDream: Boosting Holistic Text Coherence in 3D Generation via Multimodal Large Language Models Feedback. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(7):5369-5377. doi:10.1609/aaai.v40i7.37453

BibTeX
@article{Jiang2026,
  author = {Jiang, Chenhan and Zeng, Yihan and Yeung, Dit-Yan},
  title = {CoherenDream: Boosting Holistic Text Coherence in 3D Generation via Multimodal Large Language Models Feedback},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {7},
  pages = {5369--5377},
  doi = {10.1609/aaai.v40i7.37453}
}
Kostenlos starten →