language:
- en
library_name: diffusers
license: apache-2.0
tags:
- text-to-image
- stable-diffusion
pipeline_tag: image-to-image
IP-Adapter Model Card
Introduction
we present IP-Adapter, an effective and lightweight adapter to achieve image prompt capability for the pre-trained text-to-image diffusion models. An IP-Adapter with only 22M parameters can achieve comparable or even better performance to a fine-tuned image prompt model. IP-Adapter can be generalized not only to other custom models fine-tuned from the same base model, but also to controllable generation using existing controllable tools. Moreover, the image prompt can also work well with the text prompt to accomplish multimodal image generation.
Models
Image Encoder
- models/image_encoder: OpenCLIP-ViT-H-14 with 632.08M parameter
- sdxl_models/image_encoder: OpenCLIP-ViT-bigG-14 with 1844.9M parameter
More information can be found here
IP-Adapter for SD 1.5
- ip-adapter_sd15.bin: use global image embedding from OpenCLIP-ViT-H-14 as condition
- ip-adapter_sd15_light.bin: same as ip-adapter_sd15, but more compatible with text prompt
- ip-adapter-plus_sd15.bin: use patch image embeddings from OpenCLIP-ViT-H-14 as condition, closer to the reference image than ip-adapter_sd15
- ip-adapter-plus-face_sd15.bin: same as ip-adapter-plus_sd15, but use cropped face image as condition
IP-Adapter for SDXL 1.0
- ip-adapter_sdxl.bin: use global image embedding from OpenCLIP-ViT-bigG-14 as condition
- ip-adapter_sdxl_vit-h.bin: same as ip-adapter_sdxl, but use OpenCLIP-ViT-H-14
- ip-adapter-plus_sdxl_vit-h.bin: use patch image embeddings from OpenCLIP-ViT-H-14 as condition, closer to the reference image than ip-adapter_xl and ip-adapter_sdxl_vit-h
- ip-adapter-plus-face_sdxl_vit-h.bin: same as ip-adapter-plus_sdxl_vit-h, but use cropped face image as condition
Paper
The model was presented in the paper Conceptrol: Concept Control of Zero-shot Personalized Image Generation.
Abstract
Personalized image generation with text-to-image diffusion models generates unseen images based on reference image content. Zero-shot adapter methods such as IP-Adapter and OminiControl are especially interesting because they do not require test-time fine-tuning. However, they struggle to balance preserving personalized content and adherence to the text prompt. We identify a critical design flaw resulting in this performance gap: current adapters inadequately integrate personalization images with the textual descriptions. The generated images, therefore, replicate the personalized content rather than adhere to the text prompt instructions. Yet the base text-to-image has strong conceptual understanding capabilities that can be leveraged. We propose Conceptrol, a simple yet effective framework that enhances zero-shot adapters without adding computational overhead. Conceptrol constrains the attention of visual specification with a textual concept mask that improves subject-driven generation capabilities. It achieves as much as 89% improvement on personalization benchmarks over the vanilla IP-Adapter and can even outperform fine-tuning approaches such as Dreambooth LoRA. The source code is available at https://github.com/QY-H00/Conceptrol.