diff --git a/README.md b/README.md
index 9b5d6ac..8535297 100644
--- a/README.md
+++ b/README.md
@@ -19,10 +19,12 @@ Origin methods from
* [InstantCharacter](https://github.com/Tencent/InstantCharacter),
* [DreamO](https://github.com/bytedance/DreamO)
* [Bagel](https://github.com/ByteDance-Seed/Bagel)
+* [OmniConsistency](https://github.com/showlab/OmniConsistency)
## Updates:
-* 2025/06/02
+* 2025/06/04
+* OmniConsistency 并不是ID的迁移,移植过来是方便使用常规flux diffuser加载,内置多种量化方式(还未完善,目前只支持repo加载),12G以下用nf4就好(1024*1024 在12G 50秒一张图),
* 新增Bagel模型的支持,支持int8和nf4量化(官方用的十字鱼佬的PR)输入图片则是edit模式,不输入就是文生图,在量化nf4的情况下,显存峰值大约7G,实际跑4G多,edit的编辑能力,在nf4条件下一般;
@@ -247,6 +249,16 @@ download [BAGEL-7B-MoT](https://huggingface.co/ByteDance-Seed/BAGEL-7B-MoT/tree/
├──all files # 所有文件
```
+**3.12 OmniConsistency mode**
+flux repo: [flux](https://huggingface.co/black-forest-labs/FLUX.1-dev/tree/main)
+[OmniConsistency](https://hf-mirror.com/showlab/OmniConsistency/tree/main)
+```
+├── ComfyUI/models/photomaker/
+ ├──OmniConsistency.safetensors #
+├── ComfyUI/models/loras/
+ ├── any flux loras
+```
+
4 Example
----
**4.1 story-diffusion**
@@ -311,12 +323,18 @@ download [BAGEL-7B-MoT](https://huggingface.co/ByteDance-Seed/BAGEL-7B-MoT/tree/
-**4.11 Bagel**
+**4.12 Bagel**
* nf4 image2image
* nf4 txt2image
+**4.13 OmniConsistency**
+ * nf4 image2image
+
+
+
+
**4.13 comfyUI classic(comfyUI经典模式,可以接任意适配CF的流程,主要是方便使用多角色的clip)**
* any mode SD1.5 SDXL SD3.5 FLUX...
@@ -464,4 +482,4 @@ svdquant
journal = {arXiv preprint arXiv:2505.14683},
year = {2025}
}
-```
\ No newline at end of file
+```