Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,85 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
library_name: diffusers
|
| 6 |
+
pipeline_tag: text-to-image
|
| 7 |
+
---
|
| 8 |
+
|
| 9 |
+
# PhotoMaker V2 Model Card
|
| 10 |
+
|
| 11 |
+
<div align="center">
|
| 12 |
+
|
| 13 |
+
[**Project Page**](https://photo-maker.github.io/) **|** [**Paper (ArXiv)**](https://arxiv.org/abs/2312.04461) **|** [**Code**](https://github.com/TencentARC/PhotoMaker)
|
| 14 |
+
|
| 15 |
+
[🤗 **Gradio demo**](https://huggingface.co/spaces/TencentARC/PhotoMaker-V2)
|
| 16 |
+
|
| 17 |
+
</div>
|
| 18 |
+
|
| 19 |
+
## Introduction
|
| 20 |
+
|
| 21 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 22 |
+
Users can input one or a few face photos, along with a text prompt, to receive a customized photo or painting within seconds (no training required!). Additionally, this model can be adapted to any base model based on SDXL or used in conjunction with other LoRA modules.
|
| 23 |
+
|
| 24 |
+
### Realistic results
|
| 25 |
+
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
|
| 29 |
+
|
| 30 |
+
### Stylization results
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
|
| 34 |
+
|
| 35 |
+
|
| 36 |
+
|
| 37 |
+
More results can be found in our [project page](https://photo-maker.github.io/)
|
| 38 |
+
|
| 39 |
+
## Model Details
|
| 40 |
+
|
| 41 |
+
It mainly contains two parts corresponding to two keys in loaded state dict:
|
| 42 |
+
|
| 43 |
+
1. `id_encoder` includes finetuned OpenCLIP-ViT-H-14 and a few fuse layers.
|
| 44 |
+
|
| 45 |
+
2. `lora_weights` applies to all attention layers in the UNet, and the rank is set to 64.
|
| 46 |
+
|
| 47 |
+
|
| 48 |
+
## Usage
|
| 49 |
+
|
| 50 |
+
You can directly download the model in this repository.
|
| 51 |
+
You also can download the model in python script:
|
| 52 |
+
|
| 53 |
+
```python
|
| 54 |
+
from huggingface_hub import hf_hub_download
|
| 55 |
+
photomaker_ckpt = hf_hub_download(repo_id="TencentARC/PhotoMaker-V2", filename="photomaker-v2.bin", repo_type="model")
|
| 56 |
+
```
|
| 57 |
+
|
| 58 |
+
Then, please follow the instructions in our [GitHub repository](https://github.com/TencentARC/PhotoMaker/README_pmv2.md).
|
| 59 |
+
|
| 60 |
+
|
| 61 |
+
## Limitations
|
| 62 |
+
|
| 63 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 64 |
+
|
| 65 |
+
- The model's customization performance degrades on Asian male faces.
|
| 66 |
+
- The model still struggles with accurately rendering human hands.
|
| 67 |
+
|
| 68 |
+
## Bias
|
| 69 |
+
|
| 70 |
+
While the capabilities of image generation models are impressive, they can also reinforce or exacerbate social biases.
|
| 71 |
+
|
| 72 |
+
## Citation
|
| 73 |
+
|
| 74 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 75 |
+
|
| 76 |
+
**BibTeX:**
|
| 77 |
+
|
| 78 |
+
```bibtex
|
| 79 |
+
@inproceedings{li2023photomaker,
|
| 80 |
+
title={PhotoMaker: Customizing Realistic Human Photos via Stacked ID Embedding},
|
| 81 |
+
author={Li, Zhen and Cao, Mingdeng and Wang, Xintao and Qi, Zhongang and Cheng, Ming-Ming and Shan, Ying},
|
| 82 |
+
booktitle={IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
|
| 83 |
+
year={2024}
|
| 84 |
+
}
|
| 85 |
+
```
|