Hi my friend, thank you :D
Yes, it should be better, unfortunately, most artificial intelligence researchers only make the _x4 models available, BSRGAN researchers are an exception and have also put the _x2 model.
Moreover, the _x4 model is often the best model in terms of final image quality.
BSRGAN uses an incredibly high quality AI at the expense of a very heavy architecture to run, so I recommend it only for short videos and images; for videos I highly recommend RealESR_Gx4 or RealSRx4_Anime which have decent quality but are much faster (4 to 6 times faster than BSRGANx4)
However I am actively working on improving the overall speed of the app :)