Update README.md

README.md

@@ -12,30 +12,38 @@ tags:
 
 使用RKNPU2运行Stable Diffusion 1.5 LCM 图像生成模型！！
 
-- 推理速度(RK3588): 单NPU核, 384x384分辨率, 4次迭代, 生成1张图片平均耗时约13.8秒
-- 内存占用: 约5.2GB
+- 推理速度(RK3588, 单NPU核):
+  - 384x384: 文本编码器 0.05s + U-Net 2.36s/it + VAE Decoder 5.48s
+  - 512x512: 文本编码器 0.05s + U-Net 5.65s/it + VAE Decoder 11.13s
 
+- 内存占用: 
+  - 384x384: 约5.2GB
+  - 512x512: 约5.6GB
 
 ## 使用方法
 
-### 1. 克隆或者下载此仓库到本地
+### 1. 克隆或者下载此仓库到本地.
 
 ### 2. 安装依赖
 
 ```bash
-pip install diffusers pillow numpy<2
+pip install diffusers pillow numpy<2 rknn-toolkit-lite2
 ```
 
-当然你还要安装rknn-toolkit2-lite2。
-
 ### 3. 运行
 
 ```bash
-python ./run_rknn-lcm.py -i ./model -o ./images --num-inference-steps 4 -s 384x384 --prompt "Majestic mountain landscape with snow-capped peaks, autumn foliage in vibrant reds and oranges, a turquoise river winding through a valley, crisp and serene atmosphere, ultra-realistic style."
+python ./run_rknn-lcm.py -i ./model -o ./images --num-inference-steps 4 -s 512x512 --prompt "Majestic mountain landscape with snow-capped peaks, autumn foliage in vibrant reds and oranges, a turquoise river winding through a valley, crisp and serene atmosphere, ultra-realistic style."
 ```
 
 ## 模型转换
 
+### 安装依赖
+
+```bash
+pip install diffusers pillow numpy<2 rknn-toolkit2
+```
+
 ### 1. 下载模型
 
 下载一个onnx格式的Stable Diffusion 1.5 LCM模型，并放到`./model`目录下。
@@ -58,15 +66,7 @@ python ./convert-onnx-to-rknn.py -m ./model -r 384x384
 
 ## 已知问题
 
-1. 截至目前，使用最新版本的rknn-toolkit2 2.2.0版本转换的模型仍然存在极其严重的精度损失！即使使用的是fp16数据类型。如图，上方是使用onnx模型推理的结果，下方是使用rknn模型推理的结果。所有参数均一致。并且分辨率越高，精度损失越严重。这是rknn-toolkit2的bug。
-
-- 384x384:  
-![image/png](https://cdn-uploads.huggingface.co/production/uploads/6319d0860d7478ae0069cd92/yDmipD6zHHVyMVWqero-l.png)
-![image/png](https://cdn-uploads.huggingface.co/production/uploads/6319d0860d7478ae0069cd92/Ieq2m-4XnAThDnTgHWjvI.png)
-
-- 256x256:  
-![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/6319d0860d7478ae0069cd92/qoagtwDKij1WGkJwqa8bz.jpeg)
-
+1. ~~截至目前，使用最新版本的rknn-toolkit2 2.2.0版本转换的模型仍然存在极其严重的精度损失！即使使用的是fp16数据类型。如图，上方是使用onnx模型推理的结果，下方是使用rknn模型推理的结果。所有参数均一致。并且分辨率越高，精度损失越严重。这是rknn-toolkit2的bug。~~ (v2.3.0已修复)
 
 2. 其实模型转换脚本可以选择多个分辨率(例如"384x384,256x256"), 但这会导致模型转换失败。这是rknn-toolkit2的bug。
 
@@ -82,8 +82,12 @@ python ./convert-onnx-to-rknn.py -m ./model -r 384x384
 
 Run the Stable Diffusion 1.5 LCM image generation model using RKNPU2!
 
-- Inference speed (RK3588): Single NPU core, 384x384 resolution, 4 iterations, average time to generate 1 image is about 13.8 seconds
-- Memory usage: About 5.2GB
+- Inference speed (RK3588, single NPU core):
+  - 384x384: Text encoder 0.05s + U-Net 2.36s/it + VAE Decoder 5.48s
+  - 512x512: Text encoder 0.05s + U-Net 5.65s/it + VAE Decoder 11.13s
+- Memory usage:
+  - 384x384: About 5.2GB
+  - 512x512: About 5.6GB
 
 ## Usage
 
@@ -92,19 +96,23 @@ Run the Stable Diffusion 1.5 LCM image generation model using RKNPU2!
 ### 2. Install dependencies
 
 ```bash
-pip install diffusers pillow numpy<2
+pip install diffusers pillow numpy<2 rknn-toolkit-lite2
 ```
 
-Of course, you also need to install rknn-toolkit2-lite2.
-
 ### 3. Run
 
 ```bash
-python ./run_rknn-lcm.py -i ./model -o ./images --num-inference-steps 4 -s 384x384 --prompt "Majestic mountain landscape with snow-capped peaks, autumn foliage in vibrant reds and oranges, a turquoise river winding through a valley, crisp and serene atmosphere, ultra-realistic style."
+python ./run_rknn-lcm.py -i ./model -o ./images --num-inference-steps 4 -s 512x512 --prompt "Majestic mountain landscape with snow-capped peaks, autumn foliage in vibrant reds and oranges, a turquoise river winding through a valley, crisp and serene atmosphere, ultra-realistic style."
 ```
 
 ## Model Conversion
 
+### Install dependencies
+
+```bash
+pip install diffusers pillow numpy<2 rknn-toolkit2
+```
+
 ### 1. Download the model
 
 Download a Stable Diffusion 1.5 LCM model in ONNX format and place it in the `./model` directory.
@@ -127,14 +135,7 @@ Note that the higher the resolution, the larger the model and the longer the con
 
 ## Known Issues
 
-1. As of now, models converted using the latest version of rknn-toolkit2 (version 2.2.0) still suffer from severe precision loss, even when using fp16 data type. As shown in the image, the top is the result of inference using the ONNX model, and the bottom is the result using the RKNN model. All parameters are the same. Moreover, the higher the resolution, the more severe the precision loss. This is a bug in rknn-toolkit2.
-- 384x384:  
-![image/png](https://cdn-uploads.huggingface.co/production/uploads/6319d0860d7478ae0069cd92/yDmipD6zHHVyMVWqero-l.png)
-![image/png](https://cdn-uploads.huggingface.co/production/uploads/6319d0860d7478ae0069cd92/Ieq2m-4XnAThDnTgHWjvI.png)
-
-- 256x256:  
-![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/6319d0860d7478ae0069cd92/qoagtwDKij1WGkJwqa8bz.jpeg)
-
+1. ~~As of now, models converted using the latest version of rknn-toolkit2 (version 2.2.0) still suffer from severe precision loss, even when using fp16 data type. As shown in the image, the top is the result of inference using the ONNX model, and the bottom is the result using the RKNN model. All parameters are the same. Moreover, the higher the resolution, the more severe the precision loss. This is a bug in rknn-toolkit2.~~ (Fixed in v2.3.0)
 
 2. Actually, the model conversion script can select multiple resolutions (e.g., "384x384,256x256"), but this causes the model conversion to fail. This is a bug in rknn-toolkit2.