Update README

README.md

@@ -11,8 +11,6 @@ tags:
 pipeline_tag: text-to-image
 extra_gated_eu_disallowed: true
 ---
-
-
 [中文阅读](./README_CN.md)
 
 <p align="center">
@@ -36,6 +34,9 @@ extra_gated_eu_disallowed: true
   <a href=https://x.com/TencentHunyuan target="_blank"><img src=https://img.shields.io/badge/Hunyuan-black.svg?logo=x height=22px></a>
 </div>
 
+<p align="center">
+    👋 Join our <a href="assets/WECHAT.md" target="_blank">WeChat</a> 
+</p>
 
 -----
 
@@ -199,10 +200,13 @@ From the results, HunyuanImage 2.1 achieved a relative win rate of -1.36% agains
 
 **Hardware and OS Requirements:**
 - NVIDIA GPU with CUDA support.
-  - **Minimum:** 59 GB GPU memory for 2048x2048 image generation (batch size = 1).
+  - **Minimum requrement for now:** 36 GB GPU memory for 2048x2048 image generation.
+  
+  > ✨ An FP8-quantized model is coming soon, enabling even lower GPU memory requirements for inference, stay tuned 👀!
+  
+  > **Note:** The memory requirements above are measured with model CPU offloading enabled. If your GPU has sufficient memory, you may disable offloading for improved inference speed.
 - Supported operating system: Linux.
 
-> **Note:** The memory requirements above are measured with model CPU offloading enabled. If your GPU has sufficient memory, you may disable offloading for improved inference speed.
 
 ## 🛠️ Dependencies and Installation
 
@@ -223,7 +227,6 @@ pip install flash-attn==2.7.3 --no-build-isolation
 The details of download pretrained models are shown [here](checkpoints-download.md).
 
 ## 🔑 Usage
-
 HunyuanImage-2.1 only supports 2K image generation (e.g. 2048x2048 for 1:1 images, 2560x1536 for 16:9 images, etc.).
 Generating images with 1K resolution will result in artifacts.
 Additionally, we recommend using the full generation pipeline for better quality (i.e. enabling prompt enhancement and refinment).
@@ -250,9 +253,9 @@ image = pipe(
     width=2048,
     height=2048,
     use_reprompt=True,  # Enable prompt enhancement
-    use_refiner=True,  # Enable refiner model
+    use_refiner=True,   # Enable refiner model
     # For the distilled model, use 8 steps for faster inference.
-    # For the non-distilled model, use 50 steps for better quality
+    # For the non-distilled model, use 50 steps for better quality.
     num_inference_steps=8 if "distilled" in model_name else 50, 
     guidance_scale=3.5,
     shift=5,
@@ -283,9 +286,9 @@ We would like to thank the following open-source projects and communities for th
 ## Github Star History
 <a href="https://star-history.com/#Tencent-Hunyuan/HunyuanImage-2.1&Date">
  <picture>
-   <source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/svg?repos=Tencent-Hunyuan/HunyuanImage-2.1&type=Date&theme=dark" />
-   <source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/svg?repos=Tencent-Hunyuan/HunyuanImage-2.1&type=Date" />
-   <img alt="Star History Chart" src="https://api.star-history.com/svg?repos=Tencent-Hunyuan/HunyuanImage-2.1&type=Date" />
+   <source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/svg?repos=Tencent-Hunyuan/HunyuanImage-2.1&type=Date1&theme=dark" />
+   <source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/svg?repos=Tencent-Hunyuan/HunyuanImage-2.1&type=Date1" />
+   <img alt="Star History Chart" src="https://api.star-history.com/svg?repos=Tencent-Hunyuan/HunyuanImage-2.1&type=Date1" />
  </picture>
 </a>
 

README_CN.md

@@ -15,6 +15,11 @@
   <a href=https://x.com/TencentHunyuan target="_blank"><img src=https://img.shields.io/badge/Hunyuan-black.svg?logo=x height=22px></a>
 </div>
 
+
+<p align="center">
+    👋 加入我们的 <a href="assets/WECHAT.md" target="_blank">WeChat</a>
+</p>
+
 ----
 
 
@@ -171,11 +176,14 @@ SSAE（结构化语义对齐评估）是一种基于先进多模态大语言模
 
 **硬件和操作系统要求：**
 - 支持 CUDA 的 NVIDIA GPU。
-  - **最低要求：** 59 GB 显存用于 2048x2048 图像生成（batch size = 1）。
+  - **最低要求：** 36 GB 显存，可用于 2048x2048 图像生成。
+
+  > ✨ 即将推出 FP8 量化模型，推理所需显存将进一步降低，敬请期待 👀！
+
+  > **注意：** 上述内存要求是在启用模型 CPU offloading 的情况下测量的。如果您的 GPU 有足够的显存，可以禁用 CPU offloading 以提高推理速度。
 - 支持的操作系统：Linux。
 
 
-> **注意：** 上述内存要求是在启用模型 CPU offloading 的情况下测量的。如果您的 GPU 有足够的显存，可以禁用 CPU offloading 以提高推理速度。
 
 ## 🛠️ 依赖与安装
 
@@ -193,14 +201,14 @@ pip install flash-attn==2.7.3 --no-build-isolation
 
 ## 🧱 模型下载
 
-模型的下载与说明请参考[这里](ckpts/checkpoints-download.md)。
+模型的下载与说明请参考[这里](checkpoints-download.md)。
 
 ## 🔑 使用
-
 HunyuanImage-2.1 仅支持 2K 分辨率图像生成（如 1:1 时为 2048x2048，16:9 时为 2560x1536 等）。
 使用其1K分辨率生成图像可能会带来画质下降与瑕疵。
 此外，我们建议使用完整的生成流程以获得更高画质（即启用提示词增强和精修功能）。
 
+
 ```python
 import torch
 from hyimage.diffusion.pipelines.hunyuanimage_pipeline import HunyuanImagePipeline
@@ -223,8 +231,8 @@ image = pipe(
     width=2048,
     height=2048,
     use_reprompt=True,  # 启用提示词增强
-    use_refiner=True,  # 启用精修模型, 以获得更高画质
-    # 对于蒸馏版模型，建议使用 8 步以加快推理速度；
+    use_refiner=True,   # 启用精修模型, 以获得更高画质
+    # 对于蒸馏版模型，建议使用 8 步以加快推理速度
     # 对于非蒸馏版模型，建议使用 50 步以获得更高画质
     num_inference_steps=8 if "distilled" in model_name else 50, 
     guidance_scale=3.5,