Video quality degrades when uploading to YouTube with variable frame rate

Question

So I use OBS (latest ver) to record my gameplay on Windows 10.

OBS settings are:

• recording format: MP4
• Recording quality: indistinguishable quality, large file size
• Encoder: Software (x264 low CPU usage preset, increases file size)
• Downscale filter: Lanczos (sharpened scaling, 32 samples)
• FPS value for recording: 60
• Base and output resolution are both the same, 1080P and my monitor is 1080p as well as my in-game resolution.

So when i'm recording from a less demanding games like the new Resident Evil 2 (RE2) the video file looks perfect, both on my computer (using VLC) and on YouTube they look great. I run the game on fixed 60 FPS using V-sync and ultra graphic settings.

the videos i'm talking about:

https://www.youtube.com/watch?v=vyrzq-6eeoM

https://www.youtube.com/watch?v=sCo2GspC-jE

But, when I record from a more demanding game like Assassin's Creed Odyssey, I can't get a fixed 60 FPS, average is 45 FPS and it varies between 32 to 60 FPS. after i finish recording i play the video file on VLC and it looks great, just like my gameplay, but when I upload it to YouTube, it looks awful, blurry and so much noise in it.

https://www.youtube.com/watch?v=84ZPY1bnj4I

My YouTube channel doesn't have so much subscribers (almost 200) so YouTube doesn't use VP9 to convert my videos, it only uses x264, I don't know it's relatable though.

so my question is, where is the problem exactly? is it YouTube's fault converting my videos so poorly? because they look fine on my computer when i play them.

Or

is it my fault and i'm doing something wrong in the settings? if so what should i change?

PS:

I've tried setting the recording FPS in OBS to 30 FPS but the gameplay looked so slow on YouTube and on my PC.

Answer 1

It's not your fault neither YouTube, it's about video compression.

To simplify, when you want to encode a video, you will have a base image every few seconds (5 for example) and to have the intermediate images you are saving movement to reduce data flow (and file size by the way). To reduce more the data, you can approach movement by linear function and represent movement by key frames each 2 or 3 frames. When you have full FPS, it's perfect, but when you have less, you will regularly have two identical images and for the movement it's like stopping movement then resume it. It cost mush more data to code less frame rate because you will have to use much more key frames to code frame stop and resume.

That's why your quality decrease if you're not running your game at 60fps.

But there is solutions, the first is to record at a frame rate inferior to your game. The second is to tell your editing software to interpolate missing frames rather than display a the last frame.