AI Paper Atlas

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

Holistic Evaluation of Text-to-Image Models

Tony Lee, Michihiro Yasunaga, Chenlin Meng, Yifan Mai, Joon Sung Park, Agrim Gupta, Yunzhi Zhang, Deepak Narayanan, Hannah Benita Teufel, Marco Bellagente, Minguk Kang, Taesung Park and 6 more

NeurIPS 2023 · Image generation › Text-to-image diffusion · approach: T2I evaluation and controllable adapters · 18 citations (OpenAlex)

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
Apply Promptist prompt engineering to Stable Diffusion v1-4Stable Diffusion v1-4 · HEIM aestheticshuman-rated aestheticsImproved
Apply Promptist prompt engineering to Stable Diffusion v1-4Stable Diffusion v1-4 · HEIM alignmenttext-image alignmentNo clear effect
Fine-tune on artistic imagesStable Diffusion-based models · HEIM aestheticsaestheticsImproved
Fine-tune on artistic imagesStable Diffusion-based models · HEIM photorealismhuman-rated photorealismHurt
Fine-tune on custom dataText-to-image models · MS-COCO gender and dialect perturbationshuman-rated alignment gapHurt
Increase SafeStableDiffusion safety guidance strengthSafeStableDiffusion · I2Pinappropriate-image generation frequencyImproved
Use safety guidanceSafeStableDiffusion · HEIM alignmenthuman-rated text-image alignmentHurt
Increase model size (ablation)DALL-E and DeepFloyd-IF model families · MS-COCOhuman metricsImproved
Translate MS-COCO prompts into Hindi, Chinese, or SpanishMost evaluated text-to-image models · MS-COCO translated promptstext-image alignmentHurt
Translate MS-COCO prompts into ChineseCogView2 · MS-COCO Chinese promptstext-image alignmentImproved
Introduce typos in promptsText-to-image models · MS-COCO robustnesshuman-rated alignment: no more than 0.2No clear effect
Use diffusion rather than autoregressive generation at similar parameter countText-to-image models · HEIM efficiencyinference efficiencyImproved

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