Pixelated video...

[Hawick]

So I've seen a 720p gameplay video of GTA V and it looks really goddamn nice for a 720p video. You can check it out:

https://www.youtube.com/watch?v=vbQG70x5iyw

 

I recorded 1080p videos (very similar gta v gameplay video) and my 1080p looks like it's 480p or something. It gets extremely pixelated while I'm moving fast in cars or just moving the camera around. Even when there isn't much movement it still looks more pixelated than this 720p video.

 

Can anyone help me? How can I make 720p videos like this?

 

EDIT: I use shadowplay and use the highest bitrate and resolution settings available.

[mariushm]

 

Record with lossless capture card (for example epiphan av.io 4k or other cheaper cards in the 1-200$ range) or capture with OBS using software encoding, using either lossless codecs or x264 with near lossless presets

Render at native recording resolution or a resolution that's 1/2 or 2x the original resolution ... ex capture at 1080p and render at 1080p, or capture at 2560x1440 and resize to 1/2 to get 1280x720 and render at 720p ..

Guy may be using high resolution (4K and then resizes to 720p), or he uses some good antialiasing settings and anisotropic filtering and so on..

 

Render video using a high quality preset and high bitrate (give youtube as much quality as possible)

 

Use encoder options to optimize encoding for the actual content (animation, sharp edges etc)

shadowplay doesn't make an effort to preserve as much quality in the amount of bitrate you give it .... it's optimized for speed. You can give it 60 mbps but if you leave the preset on "fast" as soon as nvenc gets the frame below some threshold, it's satisfied and moves on to the next frame. It doesn't stay to analyze 10-16 consecutive frames and see how things move around, in order to preserve as much quality in the bitrate you have.

Youtube screws you by recompressing everything, so it's important to give as much quality to youtube in the first place.

[Hawick]

8 minutes ago, mariushm said:

 

Record with lossless capture card (for example epiphan av.io 4k or other cheaper cards in the 1-200$ range) or capture with OBS using software encoding, using either lossless codecs or x264 with near lossless presets

Render at native recording resolution or a resolution that's 1/2 or 2x the original resolution ... ex capture at 1080p and render at 1080p, or capture at 2560x1440 and resize to 1/2 to get 1280x720 and render at 720p ..

Guy may be using high resolution (4K and then resizes to 720p), or he uses some good antialiasing settings and anisotropic filtering and so on..

 

Render video using a high quality preset and high bitrate (give youtube as much quality as possible)

 

Use encoder options to optimize encoding for the actual content (animation, sharp edges etc)

shadowplay doesn't make an effort to preserve as much quality in the amount of bitrate you give it .... it's optimized for speed. You can give it 60 mbps but if you leave the preset on "fast" as soon as nvenc gets the frame below some threshold, it's satisfied and moves on to the next frame. It doesn't stay to analyze 10-16 consecutive frames and see how things move around, in order to preserve as much quality in the bitrate you have.

Youtube screws you by recompressing everything, so it's important to give as much quality to youtube in the first place.

Thanks for your answer. That guy plays PS4. So it must be the capture card.