#text-to-image
5 briefs tagged #text-to-image.
Research
Researchers propose DiSCO for black-box image safety
The prompt-level defense targets NSFW outputs without model access, retraining or fine-tuning.
arXiv cs.AI+1 outlet·13h ago
Research
Abra study maps scaling laws for diffusion image models
Researchers report predictable text-to-image scaling across 10^19 to 10^22 FLOPs.
HF Daily Papers·1d ago
Research
Researchers advance pixel-space diffusion models
Two HF papers explore pixel-level diffusion for image restoration and text-to-image generation.
HF Daily Papers·2d ago
Research
Study finds faster path for pixel-space diffusion models
Researchers report a latent-to-pixel recipe that matches or beats latent-space image models.
HF Daily Papers·2d ago
Research
CAPEval Separates Caption Coverage From Precision
The benchmark tests how caption detail and factual reliability affect multimodal understanding and generation.
HF Daily Papers·Aug 3