Thanks it worked with some changes: - Python version 3.10.10. - My laptop is CUDA 12, I use CUDA 11.7. - Instead of conda, I use pip to install pytorch. - Run the remaining commands under requirements.txt skipping the conda ones.
Hi, nice to know that you worked it out. I am trying it too but have not succeeded yet. Could you please list your installation steps (cmds)? Do you need to install Cuda 11.7 toolkit ahead of time?
@linkozsam Жыл бұрын
@@davidsommers9271 In my case, I selected CUDA 11.7 in that step in the video. No need to install CUDA 11.7 toolkit. I also got stuck with installing "pip" due to Python 3.11.1 and need to install Python 3.10.9 instead. BTW, cmd does not recognize the command, so I need to use Anaconda Powershell Prompt to run those commands. After installation is done, I tested it with a whole image that was captured from a video clip of our dancing performance. The result is amazing.
@davidsommers9271 Жыл бұрын
@@linkozsam Yes. It WORKED for me. I had Anaconda installed w Python 3.9. So I uninstalled Python 3.11 and installed 3.10.9 and added it to path. I ran the ***Anaconda PS Prompt*** from the menu. I then install the rest commands without any issues. PyTorch was installed very fast as well as other requirements - all within a reasonable time frame. THANKS SO MUCH!
@@crisz305 Traceback (most recent call last): File "C:\Users\wq713\CodeFormer\inference_codeformer.py", line 116, in raise FileNotFoundError('No input image/video is found... ' FileNotFoundError: No input image/video is found... Note that --input_path for video should end with .mp4|.mov|.avi 我也是视频不行 报这个错
这类深度学习大模型需要喂海量的数据给上亿个参数,一般在云端用分布式系统训练,用个人机器训练到宇宙的尽头都未必能算完😅作者在项目Todo list里写了"Add training code and config files",未来可能会给爱好者提供一个预训练模型,给需要customize的玩家一些diy空间(同时不用太大的运算压力),有兴趣可以持续关注
fatal: destination path 'CodeFormer' already exists and is not an empty directory. 这个怎么处理呢
@ss-kq2en Жыл бұрын
python basicsr/setup.py develop Traceback (most recent call last): File "C:\Users\a\CodeFormer\basicsr\setup.py", line 9, in import torch File "C:\Users\a\anaconda3\lib\site-packages\torch\__init__.py", line 139, in raise err OSError: [WinError 126] 找不到指定的模块。 Error loading "C:\Users\a\anaconda3\lib\site-packages\torch\lib\caffe2_nvrtc.dll" or one of its dependencies.
我是有安装ffmeg ,但视频不行,不知少了什么? Face detection model: retinaface_resnet50 Background upsampling: True, Face upsampling: True All results are saved in results/_1.0 我是跑CPU的,results 的forder都沒看到
(base) C:\Users\twtwtw\CodeFormer>pip install -q gradiopip install -q gradio ERROR: Could not find a version that satisfies the requirement gradiopip (from versions: none) ERROR: No matching distribution found for gradiopip
可能因为照片档比较大(4~5Mb),在做全图清晰化时出错(RuntimeError: CUDA out of memory. Tried to allocate 2.16 GiB (GPU 0; 6.00 GiB total capacity; 3.64 GiB already allocated; 0 bytes free; 5.27 GiB reserved in total by PyTorch))。如果改用人脸图清晰化(DFL专用),可以操作。但是产生的照片严重扭曲。请问有方法解决吗?