A multi-agent GPT-4o pipeline that iteratively refines ad banners through self-critique and style voting outperforms single-shot generative baselines in the paper's evaluations, though the evaluation is partly self-referential and includes manual logo replacement.
Delights. Indulge in our freshly baked, delicious pastries made with the finest ingredients; enjoy a special 20% discount on your first purchase!
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Mirror in the Model: Ad Banner Image Generation via Reflective Multi-LLM and Multi-modal Agents
A multi-agent GPT-4o pipeline that iteratively refines ad banners through self-critique and style voting outperforms single-shot generative baselines in the paper's evaluations, though the evaluation is partly self-referential and includes manual logo replacement.