I need someone to explain to me in simple words, why opening any ports is bad idea whereas I always have ports like 80, 32400 or 443 open anyway? Why an attacker cannot use those always open ports instead of others? I apologise if my question seems stupid, but all the answers I received are just - keep ports closed, you dumb arse :-)
I have been looking for an affordable yet capable AI upscaler for my home media library. After spending several days testing a number of available solutions, I eventually narrowed my shortlist down to three products: Unifab Upscaler AI, Winxvideo AI, and Video2X.
I tested all three extensively. While each produced good results, none was perfect.
Unifab Upscaler AI – A capable and reasonably priced solution. However, upscaling a 480p video to a modest 720p resolution using the Fast profile increased the file size from the original 550 MB to 9.64 GB. I found this excessive and unacceptable for my needs.
Winxvideo AI – A solid application with a very user-friendly interface. Upscaling the same 550 MB video to 720p using high-quality settings took only 3.5 hours on my GTX 1660 Super, and I was generally pleased with the output quality. Unfortunately, the software is limited to 8-bit colour depth, even when using the HEVC codec. There appears to be no option to encode in 10-bit HEVC, which is necessary for HDR content. For this reason, it was not a suitable solution for me.
Video2X – A free and surprisingly powerful tool. It is not particularly beginner-friendly and requires some effort to configure properly. For example, the GUI does not support saving custom profiles. This may not be an issue for a one-off conversion, but when processing large batches of videos, having to re-enter settings every time becomes highly impractical. Fortunately, the command-line interface allows full automation. I created my own batch script with predefined settings and batch-processing support, which eliminated these limitations. After some initial setup work, Video2X proved to be the most suitable option for my requirements.
Conclusion: Video2X ultimately offered the best balance of quality, flexibility, and cost. Although it requires more effort to set up, its automation capabilities and zero cost made it the clear winner for my workflow.
edited:
I forgot to mention that Video2x latest version 6.4.0 removed ability to use Real-ESRGAN x2PLUS profile. It is hardcoded now for x4 upscaling, which is not good for me. So, it is basically limits anyone who does not need x4 upscale to only one AI model - for Anime & Cartoons.
Having said all this, I wonder if anyone knows and can help me how can I: