AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

VLM-Guided Adaptive Negative Prompting for Creative Generation

Shelly Golan, Yotam Nitzan, Zongze Wu, Or Patashnik

ICLR 2026 · Image generation › Text-to-image diffusion · approach: Text generation and structured prediction

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace static GPT-4o negative lists with per-step adaptive VLM negatives (ablation)VLM-Guided Adaptive Negative Prompting · Creative generation (4 categories)relative typicality: 1.043 -> 2.190Improved
apply negatives dynamically instead of replaying accumulated list from step 0 per seed (ablation)VLM-Guided Adaptive Negative Prompting · qualitative bag/buildingcreativity (qualitative)Improved
use per-seed adaptive negatives instead of cross-seed reuse (ablation)VLM-Guided Adaptive Negative Prompting · Creative generation (4 categories)Vendi score: 4.584 -> 5.794Improved
add accumulation of negative prompts across steps (ablation)VLM-Guided Adaptive Negative Prompting · Creative generation (4 categories)GPT novelty score: 0.060 -> 0.370Improved
use stronger VLM (BLIP-2 or Qwen2.5 instead of ViLT) (ablation)VLM-Guided Adaptive Negative Prompting · Creative generation (4 categories)GPT novelty score: 0.1575 -> 0.370 (BLIP-2), 0.401 (Qwen2.5)Improved
limit VLM guidance to first 10-15 of 28 steps (ablation)VLM-Guided Adaptive Negative Prompting · Building, Bag, Jacket (qualitative)creative quality / runtime: Qwen2.5-3B runtime 71 -> 56 sNo clear effect
use adaptive VLM-guided negative prompting versus creative promptingSD3.5-large · User studynovelty (1-5): 1.753 -> 4.550Improved
use VLM-guided method instead of ConceptLabcreative generation · User studyvalidity (1-5): 3.950 -> 4.503Improved
add VLM-guided negative prompting versus standard SD3.5 promptingSD3.5-large · User studyvalidity (1-5): 4.886 -> 4.503Hurt
remove creativity adjective from positive prompt (ablation)VLM-Guided Adaptive Negative Prompting · qualitativecreativity (qualitative)Hurt
add VLM negative prompting; runtime overhead with ViLTSD3.5-large · Runtimeseconds per image: 22 -> 35Hurt
add creativity prompt words (Creative Prompting) aloneSD3.5-large · Creative generation (4 categories)relative typicality: 1.640 -> 1.645No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy