Research map / Image generation / Text-to-image diffusion
StoryDALL-E: Adapting Pretrained Text-to-Image Transformers for Story Continuation
Adyasha Maharana, Darryl Hannan, Mohit Bansal
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add cross-attention layers over the source frame (ablation) | StoryDALL-E (finetuning) · DiDeMoSV val | FID: 55.89 -> 41.58 | Improved |
| add cross-attention layers over the source frame (ablation) | StoryDALL-E (finetuning) · PororoSV val | FID: 30.45 -> 21.64 | Improved |
| add cross-attention layers over the source frame (ablation) | StoryDALL-E (finetuning) · FlintstonesSV val | FID: 35.04 -> 28.37 | Improved |
| add global story embeddings (ablation) | StoryDALL-E (finetuning) · PororoSV val | FID: 23.27 -> 21.64 | Improved |
| add global story embeddings (ablation) | StoryDALL-E (finetuning) · DiDeMoSV val | FID: 42.34 -> 41.58 | Improved |
| add both cross-attention and story embeddings (vs text-to-image only) (ablation) | StoryDALL-E (finetuning) · DiDeMoSV val | FID: 58.14 -> 41.58 | Improved |
| full finetuning instead of prompt-tuning | StoryDALL-E · PororoSV test | FID: 61.23 -> 25.90 | Improved |
| full finetuning instead of prompt-tuning | StoryDALL-E · PororoSV test | character F1: 29.68 -> 36.97 | Improved |
| use pretrained StoryDALL-E (finetuned) instead of StoryGAN (BERT) | StoryDALL-E · DiDeMoSV test | FID: 91.43 -> 32.92 | Improved |
| use pretrained StoryDALL-E (finetuned) instead of StoryGAN (BERT) | StoryDALL-E · PororoSV test | character F1: 43.22 -> 36.97 | Hurt |
| use DALL-E Mega instead of minDALL-E | StoryDALL-E · PororoSV test | FID: 25.90 -> 23.48 | Improved |
| StoryDALL-E finetuning vs StoryGAN (BERT) in human evaluation of relevance | StoryDALL-E · FlintstonesSV human evaluation (relevance) | win/lose/tie %: 32/38/30 | Hurt |
Closest papers by meaning
- StoryGPT-V: Large Language Models as Consistent Story Visualizers (CVPR 2025 · similarity 0.91)
- Intelligent Grimm - Open-ended Visual Storytelling via Latent Diffusion Models (CVPR 2024 · similarity 0.89)
- One-Prompt-One-Story: Free-Lunch Consistent Text-to-Image Generation Using a Single Prompt (ICLR 2025 · similarity 0.89)
- Storybooth: Training-Free Multi-Subject Consistency for Improved Visual Storytelling (ICLR 2025 · similarity 0.88)
- LLM Blueprint: Enabling Text-to-Image Generation with Complex and Detailed Prompts (ICLR 2024 · similarity 0.88)
- Consistent Story Generation: Unlocking the Potential of Zigzag Sampling (NeurIPS 2025 · similarity 0.88)
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.
