JiayunLi

Results 5 comments of JiayunLi

@ckkelvinchan hey, ckkelvinchan. I've refactored the structure of inference_realbasicvsr.py for memory issues and committed to [my fork](https://github.com/JiayunLi-3E8/RealBasicVSR.git), hope you read and test them and consider merging into your repo. -...

> @ckkelvinchan hey, ckkelvinchan. > > I've refactored the structure of inference_realbasicvsr.py for memory issues and committed to my fork... ![image](https://user-images.githubusercontent.com/61492105/189465717-6a53bb17-1f73-4ce6-af1d-d10a51d88441.png) ![image](https://user-images.githubusercontent.com/61492105/189465857-81eff103-8e09-4860-891f-721b3b8aa584.png)

> @JiayunLi-3E8 does it output exactly the same frames as the original code with your modifications ? I mean, if there is motion and something moves from one part of...

> @JiayunLi-3E8 I see - you have split it in 3 bands. Trying to optimize is good thing. 3070 is not low performance, it is low memory. I see another...

The reason of this problem was found. In the Sounddevice lib of Python, ![Screenshot 2023-06-21 085329](https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/assets/61492105/643bc042-b68d-4026-a47d-ac091832cb7c) it only process the exception of `decode('utf-8')` of DirectSound and MME. So, if there...