Update README.md
Browse files
README.md
CHANGED
|
@@ -1,91 +1,98 @@
|
|
| 1 |
-
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
[
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
[
|
| 17 |
-
[
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
##
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
| 71 |
-
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
base_model:
|
| 4 |
+
- stable-diffusion-v1-5/stable-diffusion-v1-5
|
| 5 |
+
---
|
| 6 |
+
<meta name="google-site-verification" content="-XQC-POJtlDPD3i2KSOxbFkSBde_Uq9obAIh_4mxTkM" />
|
| 7 |
+
|
| 8 |
+
# DynamicID: Zero-Shot Multi-ID Image Personalization with Flexible Facial Editability
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
<div align="center">
|
| 12 |
+
|
| 13 |
+
### [ICCV 2025]
|
| 14 |
+
|
| 15 |
+
[Xirui Hu](https://openreview.net/profile?id=~Xirui_Hu1),
|
| 16 |
+
[Jiahao Wang](https://openreview.net/profile?id=~Jiahao_Wang14),
|
| 17 |
+
[Hao Chen](https://openreview.net/profile?id=~Hao_chen100),
|
| 18 |
+
[Weizhan Zhang](https://openreview.net/profile?id=~Weizhan_Zhang1),
|
| 19 |
+
[Benqi Wang](https://openreview.net/profile?id=~Benqi_Wang2),
|
| 20 |
+
[Yikun Li](https://openreview.net/profile?id=~Yikun_Li1),
|
| 21 |
+
[Haishun Nan](https://openreview.net/profile?id=~Haishun_Nan1),
|
| 22 |
+
|
| 23 |
+
[](https://arxiv.org/abs/2503.06505)
|
| 24 |
+
[](https://github.com/ByteCat-bot/DynamicID)
|
| 25 |
+
</div>
|
| 26 |
+
|
| 27 |
+
---
|
| 28 |
+
This is the official implementation of DynamicID, a framework that generates visually harmonious image featuring **multiple individuals**. Each person in the image can be specified through user-provided reference images, and most notably, our method enables **independent control of each individual's facial expression** via text prompts. Hope you have fun with this demo!
|
| 29 |
+
|
| 30 |
+
---
|
| 31 |
+
|
| 32 |
+
## 🔍 Abstract
|
| 33 |
+
|
| 34 |
+
Recent advancements in text-to-image generation have spurred interest in personalized human image generation. Although existing methods achieve high-fidelity identity preservation, they often struggle with **limited multi-ID usability** and **inadequate facial editability**.
|
| 35 |
+
|
| 36 |
+
We present DynamicID, a tuning-free framework that inherently facilitates both single-ID and multi-ID personalized generation with high fidelity and flexible facial editability. Our key innovations include:
|
| 37 |
+
|
| 38 |
+
- Semantic-Activated Attention (SAA), which employs query-level activation gating to minimize disruption to the original model when injecting ID features and achieve multi-ID personalization without requiring multi-ID samples during training.
|
| 39 |
+
|
| 40 |
+
- Identity-Motion Reconfigurator (IMR), which applies feature-space manipulation to effectively disentangle and reconfigure facial motion and identity features, supporting flexible facial editing.
|
| 41 |
+
|
| 42 |
+
- A task-decoupled training paradigm that reduces data dependency
|
| 43 |
+
|
| 44 |
+
- A curated VariFace-10k facial dataset, comprising 10k unique individuals, each represented by 35 distinct facial images.
|
| 45 |
+
|
| 46 |
+
Experimental results demonstrate that DynamicID outperforms state-of-the-art methods in identity fidelity, facial editability, and multi-ID personalization capability.
|
| 47 |
+
|
| 48 |
+
## 💡 Method
|
| 49 |
+
|
| 50 |
+
<div align="center">
|
| 51 |
+
<img src="assets/pipeline.jpg", width="1000">
|
| 52 |
+
</div>
|
| 53 |
+
|
| 54 |
+
The proposed framework is architected around two core components: SAA and IMR. (a) In the anchoring stage, we jointly optimize the SAA and a face encoder to establish robust single-ID and multi-ID personalized generation capabilities. (b) Subsequently in the reconfiguration stage, we freeze these optimized components and leverage them to train the IMR for flexible and fine-grained facial editing.
|
| 55 |
+
|
| 56 |
+
## 🚀 Checkpoint
|
| 57 |
+
|
| 58 |
+
1. Download the pretrained Stable Diffusion v1.5 checkpoint from [Stable Diffusion v1.5 on Hugging Face](https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5).
|
| 59 |
+
|
| 60 |
+
2. Download our SAA-related and IMR-related checkpoints from [DynamicID Checkpoints on Hugging Face](https://huggingface.co/meteorite2023/DynamicID).
|
| 61 |
+
|
| 62 |
+
|
| 63 |
+
## 🌈 Gallery
|
| 64 |
+
|
| 65 |
+
<div align="center">
|
| 66 |
+
<img src="assets/teaser.jpg", width="900">
|
| 67 |
+
<br><br><br>
|
| 68 |
+
<img src="assets/single.jpg", width="900">
|
| 69 |
+
<br><br><br>
|
| 70 |
+
<img src="assets/multi.jpg", width="900">
|
| 71 |
+
</div>
|
| 72 |
+
|
| 73 |
+
## 📌 ToDo List
|
| 74 |
+
|
| 75 |
+
- [x] Release technical report
|
| 76 |
+
- [x] Release **training and inference code**
|
| 77 |
+
- [x] Release **Dynamic-sd** (based on *stable diffusion v1.5*)
|
| 78 |
+
- [ ] Release **Dynamic-flux** (based on *Flux-dev*)
|
| 79 |
+
- [ ] Release a Hugging Face Demo Space
|
| 80 |
+
|
| 81 |
+
## 📖 Citation
|
| 82 |
+
If you are inspired by our work, please cite our paper.
|
| 83 |
+
```bibtex
|
| 84 |
+
@inproceedings{dynamicid,
|
| 85 |
+
title={DynamicID: Zero-Shot Multi-ID Image Personalization with Flexible Facial Editability},
|
| 86 |
+
author={Xirui Hu,
|
| 87 |
+
Jiahao Wang,
|
| 88 |
+
Hao Chen,
|
| 89 |
+
Weizhan Zhang,
|
| 90 |
+
Benqi Wang,
|
| 91 |
+
Yikun Li,
|
| 92 |
+
Haishun Nan
|
| 93 |
+
},
|
| 94 |
+
booktitle={International Conference on Computer Vision},
|
| 95 |
+
year={2025}
|
| 96 |
+
}
|
| 97 |
+
|
| 98 |
+
```
|