update mnn-sr and readme

Author: tumuyan

README.md

@@ -56,83 +56,11 @@ Another way to convert models:
 
 ![](chainner-pth2ncnn.png)
 
-## How to build RealSR-NCNN-Android-CLI
-### step1
-https://github.com/Tencent/ncnn/releases  
-download ncnn-yyyymmdd-android-vulkan-shared.zip.  
-https://github.com/webmproject/libwebp
-download the source of libwebp.  
-https://opencv.org/releases/
-download opencv-android-sdk. 
-
-### step2
-extract `ncnn-yyyymmdd-android-vulkan-shared.zip` into `../3rdparty/ncnn-android-vulkan-shared`  
-extract the source of libwebp into `../3rdparty/libwebp`  
-extract `opencv-version-android-sdk` into `../3rdparty/opencv-android-sdk`
-```
-RealSR-NCNN-Android
-├─3rdparty
-│   ├─opencv-android-sdk
-│   │   └─sdk
-│   ├─libwebp
-│   └─ncnn-android-vulkan-shared
-│       └─arm64-v8a
-├─RealSR-NCNN-Android-CLI
-│   ├─Anime4k
-│   ├─RealCUGAN
-│   ├─Waifu2x
-│   ├─RealSR
-│   ├─SRMD
-│   └─ReSize
-└─RealSR-NCNN-Android-GUI
-```
-
-### step3
-Open this project with Android Studio, rebuild it and the build result in `RealSR-NCNN-Android-CLI\*\build\intermediates\cmake\release\obj\arm64-v8a` or `RealSR-NCNN-Android-CLI\*\build\intermediates\cmake\debug\obj\arm64-v8a` could copy to the GUI project automatilly.    
-Click `3rdparty/copy_cli_build_result.bat` and it could copy the other files to GUI project.
-
-
-## How to use RealSR-NCNN-Android-CLI
-### Download models
-You could download `assets.zip` from github release page and unzip it to get models, or download models from https://github.com/tumuyan/realsr-models .
-
-### Example Command
-Make sure the elf file has execute permission. Then input command
-
-```shell
-./realsr-ncnn -i input.jpg -o output.jpg
-```
-
-### Full Usages
-The usage of others program is same as realsr-ncnn.
-```console
-Usage: realsr-ncnn -i infile -o outfile [options]...
-
-  -h                   show this help
-  -v                   verbose output
-  -i input-path        input image path (jpg/png/webp) or directory
-  -o output-path       output image path (jpg/png/webp) or directory
-  -s scale             upscale ratio (4, default=4)
-  -t tile-size         tile size (>=32/0=auto, default=0) can be 0,0,0 for multi-gpu
-  -m model-path        realsr model path (default=models-DF2K_JPEG)
-  -g gpu-id            gpu device to use (default=0) can be 0,1,2 for multi-gpu, -1 use cpu
-  -j load:proc:save    thread count for load/proc/save (default=1:2:2) can be 1:2,2,2:2 for multi-gpu
-  -x                   enable tta mode
-  -f format            output image format (jpg/png/webp, default=ext/png)
-```
-
-- `input-path` and `output-path` accept either file path or directory path
-- `scale` = scale level, 4 = upscale 4x
-- `tile-size` = tile size, use smaller value to reduce GPU memory usage, default selects automatically
-- `load:proc:save` = thread count for the three stages (image decoding + realsr upscaling + image encoding), using larger values may increase GPU usage and consume more GPU memory. You can tune this configuration with "4:4:4" for many small-size images, and "2:2:2" for large-size images. The default setting usually works fine for most situations. If you find that your GPU is hungry, try increasing thread count to achieve faster processing.
-- `format` = the format of the image to be output, png is better supported, however webp generally yields smaller file sizes, both are losslessly encoded
-
-If you encounter crash or error, try to upgrade your derive
-
 ## How to build RealSR-NCNN-Android-GUI
-Download `assets.zip` from github release page, the zip file contains models & elf files. Unzip and put them to this folder, then build it with Android Studio. 
+Build `RealSR-NCNN-Android-CLI` by yourself or download `assets.zip` from github release page, the zip file contains models & elf files. Unzip and put them to this folder, then build it with Android Studio. 
 The direct download link for current version: https://github.com/tumuyan/RealSR-NCNN-Android/releases/download/1.10.0/assets.zip
 
+
 ```
 RealSR-NCNN-Android-GUI\app\src\main\assets\
 └─realsr

README_CHS.md

@@ -75,72 +75,6 @@ models-Real-ESRGAN-AnimeSharp  // 目录需要用 models- 开头
 
 ![](chainner-pth2ncnn.png)
 
-## 如何编译 RealSR-NCNN-Android-CLI
-### step1
-https://github.com/Tencent/ncnn/releases  
-下载 `ncnn-yyyymmdd-android-vulkan-shared.zip` 或者你自己编译ncnn为so文件  
-https://github.com/webmproject/libwebp  
-下载libwebp的源码
-https://opencv.org/releases/  
-下载opencv-android-sdk
-
-### step2
-解压 `ncnn-yyyymmdd-android-vulkan-shared.zip` 到 `../3rdparty/ncnn-android-vulkan-shared`  
-解压libwebp源码到`../3rdparty/libwebp`  
-解压 `opencv-version-android-sdk` 到 `../3rdparty/opencv-android-sdk`  
-
-```
-RealSR-NCNN-Android
-├─3rdparty
-
-│   ├─opencv-android-sdk
-│   │   └─sdk
-│   ├─libwebp
-│   └─ncnn-android-vulkan-shared
-│       └─arm64-v8a
-├─RealSR-NCNN-Android-CLI
-│   ├─Anime4k
-│   ├─RealCUGAN
-│   ├─Waifu2x
-│   ├─RealSR
-│   ├─SRMD
-│   └─ReSize
-└─RealSR-NCNN-Android-GUI
-```
-
-### step3
-用 Android Studio 打开工程, rebuild 然后你就可以在 `RealSR-NCNN-Android-CLI\*\build\intermediates\cmake\release\obj\arm64-v8a` 或 `RealSR-NCNN-Android-CLI\*\build\intermediates\cmake\debug\obj\arm64-v8a` 找到编译好的二进制文件，这些文件会被编译脚本自动复制到 GUI 工程目录中。  
-点击 `3rdparty/copy_cli_build_result.bat` 可以更新其他库文件的二进制文件到 GUI 工程目录中
-
-## 如何使用 RealSR-NCNN-Android-CLI
-### 下载模型
-你可以从 github release 页面下载 `assets.zip`, 或者从 https://github.com/tumuyan/realsr-models 下载所需模型，需要注意不同程序需要用对应的模型
-
-### 命令范例
-确认程序有执行权限，然后输入命令：
-```shell
-./realsr-ncnn -i input.jpg -o output.jpg
-```
-
-### 完整用法
-仅以 realsr-ncnn 为例说明，其他程序使用方法完全相同，故不重复说明
-```console
-用法: realsr-ncnn -i 输入的图片路径 -o 输出的图片路径 [其他可选参数]...
-
-  -h                   显示帮助
-  -v                   显示更多输出内容
-  -i input-path        输入的图片路径（jpg/png/webp路径或者目录路径）
-  -o output-path       输出的图片路径（jpg/png/webp路径或者目录路径）
-  -s scale             缩放系数(默认4，即放大4倍)
-  -t tile-size         tile size (>=32/0=auto, default=0) can be 0,0,0 for multi-gpu
-  -m model-path        模型路径 (默认模型 models-Real-ESRGAN-anime)
-  -g gpu-id            gpu，-1使用CPU，默认0 多GPU可选 0,1,2
-  -j load:proc:save    解码/处理/保存的线程数 (默认1:2:2) 多GPU可以设 1:2,2,2:2
-  -x                   开启tta模式
-  -f format            输出格式(jpg/png/webp, 默认ext/png)
-  
-```
-
 ## 如何编译 RealSR-NCNN-Android-GUI
 从 github release 页面下载 `assets.zip` , 其中包含了模型和CLI程序, 解压并放置到如下路径, 然后使用 Android Studio 进行编译。
 当前版本的下载连接为 https://github.com/tumuyan/RealSR-NCNN-Android/releases/download/1.10.0/assets.zip

RealSR-NCNN-Android-CLI/MNN-SR/src/main/jni/main.cpp

@@ -266,9 +266,9 @@ void *load(void *args) {
             // 判断 alpha 通道是否为单一颜色
             if (countNonZero(alphaChannel != alphaChannel.at<uchar>(0, 0)) == 0) {
                 fprintf(stderr, "ignore alpha channel, %s\n", imagepath.c_str());
+                c = 3;
             } else {
                 v.inalpha = alphaChannel;
-                c = 3;
             }
             merge(channels.data(), 3, inimage);
             v.inimage = inimage;
@@ -407,24 +407,6 @@ void *save(void *args) {
 
         high_resolution_clock::time_point begin = high_resolution_clock::now();
 
-        // free input pixel data
-        {
-            unsigned char *pixeldata = (unsigned char *) v.inimage.data;
-
-
-            fprintf(stderr, "save result...\n");
-
-            if (v.webp == 1) {
-                free(pixeldata);
-            } else {
-#if _WIN32
-                free(pixeldata);
-#else
-                stbi_image_free(pixeldata);
-#endif
-            }
-        }
-
         int success = 0;
 
         path_t ext = get_file_extension(v.outpath);
@@ -446,15 +428,27 @@ void *save(void *args) {
         } else {
             if (!v.inalpha.empty()) {
 
+                fprintf(stderr, "get alpha\n");
                 // 放大 alpha 通道
                 cv::Mat scaledAlphaChannel;
                 cv::resize(v.inalpha, scaledAlphaChannel, cv::Size(), v.scale, v.scale,
                            cv::INTER_LINEAR);
                 // 将放大的 alpha 通道赋值给输出图像的 alpha 通道
                 std::vector<cv::Mat> outChannels;
                 cv::split(v.outimage, outChannels);
+
+                fprintf(stderr, "copy alpha\n");
                 scaledAlphaChannel.copyTo(outChannels[3]); // 更新 alpha 通道
-                cv::merge(outChannels, v.outimage); // 合并回输出图像
+
+                cv::Mat outputImageWithAlpha(v.outimage.rows, v.outimage.cols, CV_8UC4);
+
+                fprintf(stderr, "merge alpha\n");
+                cv::merge(outChannels, outputImageWithAlpha); // 合并回输出图像
+
+                fprintf(stderr, "merge finishe\n");
+                v.outimage = outputImageWithAlpha;
+
+                fprintf(stderr, "merge alpha done\n");
             }
 
 #if _WIN32