· 2023
Adding Conditional Control to Text-to-Image Diffusion Models
- 3,679citations
- 2023year
Short summary
ControlNet is a new neural network architecture that adds spatial conditioning controls to existing text-to-image diffusion models without altering their original weights.
AI-generated from the title and abstract; the full text is not read.
Key points
- ControlNet adds spatial conditioning to text-to-image models without modifying original weights.
- It uses 'zero convolutions' to learn new controls while keeping pre-trained models intact.
- Tested with conditions like edges, depth, and human pose on Stable Diffusion.
- Training is robust on datasets ranging from small (<50k) to large (>1m) images.
AI-generated from the title and abstract; the full text is not read.
Abstract
We present ControlNet, a neural network architecture to add spatial conditioning controls to large, pretrained text-to-image diffusion models. ControlNet locks the production-ready large diffusion models, and reuses their deep and robust encoding layers pretrained with billions of images as a strong backbone to learn a diverse set of conditional controls. The neural architecture is connected with "zero convolutions" (zero-initialized convolution layers) that progressively grow the parameters from zero and ensure that no harmful noise could affect the finetuning. We test various conditioning controls, e.g., edges, depth, segmentation, human pose, etc., with Stable Diffusion, using single or multiple conditions, with or without prompts. We show that the training of ControlNets is robust with small (<50k) and large (>1m) datasets. Extensive results show that ControlNet may facilitate wider applications to control image diffusion models.
The authors' abstract, as published at the source., 2023 · DOI ↗
Continue with a free account
Ask the paper: 3 free questions a day about this paper; save it, get its citation, new summaries every day for your field. Takeaways are Premium.
Continue free on the webSign in with Google or Apple; no card needed. You come back to this paper.
On your phone:
Field: Radiology, Nuclear Medicine and Imaging
Radiology, Nuclear Medicine and ImagingMedicine