ProCreate, Don't Reproduce! Propulsive Energy Diffusion for Creative Generation
Abstract
In this paper, we propose , a simple and easy-to-implement method to improve sample diversity and creativity of diffusion-based image generative models and to prevent training data reproduction. operates on a set of reference images and actively propels the generated image embedding away from the reference embeddings during the generation process. We propose FSCG-8 (Few-Shot Creative Generation 8), a few-shot creative generation dataset on eight different categories—encompassing different concepts, styles, and settings—in which achieves the highest sample diversity and fidelity. Furthermore, we show that is effective at preventing replicating training data in a large-scale evaluation using training text prompts. Code and are available at https: //github.com/Agentic-Learning-AI-Lab/procreate-diffusion-public.
Cite
Text
Lu et al. "ProCreate, Don't Reproduce! Propulsive Energy Diffusion for Creative Generation." Proceedings of the European Conference on Computer Vision (ECCV), 2024. doi:10.1007/978-3-031-73027-6_23Markdown
[Lu et al. "ProCreate, Don't Reproduce! Propulsive Energy Diffusion for Creative Generation." Proceedings of the European Conference on Computer Vision (ECCV), 2024.](https://mlanthology.org/eccv/2024/lu2024eccv-procreate/) doi:10.1007/978-3-031-73027-6_23BibTeX
@inproceedings{lu2024eccv-procreate,
title = {{ProCreate, Don't Reproduce! Propulsive Energy Diffusion for Creative Generation}},
author = {Lu, Jack and Teehan, Ryan and Ren, Mengye},
booktitle = {Proceedings of the European Conference on Computer Vision (ECCV)},
year = {2024},
doi = {10.1007/978-3-031-73027-6_23},
url = {https://mlanthology.org/eccv/2024/lu2024eccv-procreate/}
}